def detect(self, image, frame=12345):
limitspeed = 0
if self.debug > 0:
print(type(image))
array = np.frombuffer(image.raw_data, dtype=np.dtype("uint8"))
array = np.reshape(array, (image.height, image.width, 4))
img_np = array[:, :, :3]
opt_augment = False
opt_agnostic_nms = False
opt_classes = None
opt_iou_thres = 0.45
opt_conf_thres = 0.40
opt_save_conf = False
newimg_np = letterbox(img_np, self.img_sz, stride=self.stride)[0]
im0s = newimg_np #cv2.cvtColor(newimg_np, cv2.COLOR_RGB2BGR)
if self.debug > 0:
print(f'im0s shape:', im0s.shape)
img = im0s[:, :, ::-1].transpose(2, 0, 1) # BGR to RGB, to 3x416x416
if self.debug > 0:
print(f'img shape:', img.shape)
img = np.ascontiguousarray(img)
if self.debug > 0:
print(f'mewimg shape:', img.shape)
img = torch.from_numpy(img).to(self.device)
img = img.half() if self.half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = self.model(img, augment=opt_augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
opt_conf_thres,
opt_iou_thres,
classes=opt_classes,
agnostic=opt_agnostic_nms)
t2 = time_synchronized()
# Process detections
for i, det in enumerate(pred): # detections per image
s = ''
gn = torch.tensor(im0s.shape)[[1, 0, 1,
0]] # normalization gain whwh
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0s.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f"{n} {self.names[int(c)]}{'s' * (n > 1)}, " # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh,
conf) if opt_save_conf else (cls,
*xywh) # label format
if self.debug > 0:
print('line:', line)
clsname = self.names[int(cls)]
label = f'{clsname} {conf:.2f}'
if self.debug > 0:
print('label:', label)
if (clsname.find('traffic.speed_limit.30') >= 0):
limitspeed = 30
elif (clsname.find('traffic.speed_limit.60') >= 0):
limitspeed = 60
elif (clsname.find('traffic.speed_limit.90') >= 0):
limitspeed = 90
#if clsname in ['traffic light','person','car','bicycle','bus','truck']:
plot_one_box(xyxy,
newimg_np,
color=tuple(self.colors[int(cls)]),
label=label,
line_thickness=1)
# Print time (inference + NMS)
if self.debug > 0:
print(f'{s}Done. ({t2 - t1:.3f}s)')
cv2.imshow('detect', newimg_np)
cv2.waitKey(1) # 1 millisecond
bgr_image = cv2.cvtColor(newimg_np, cv2.COLOR_RGB2BGR)
return bgr_image, limitspeed
def inferance(self, cv2Img, imgName='base64.jpg'):
t0 = time.time()
im0s, img, save_dir = self.preprocessing(cv2Img)
# Inference
t1 = time_synchronized()
pred = self.model(img, augment=False)[0]
# Apply NMS
pred = non_max_suppression(pred,
0.25,
0.45,
classes=None,
agnostic=False)
t2 = time_synchronized()
# Process detections
result_text = ''
for i, det in enumerate(pred): # detections per image
s, im0 = '', im0s
save_path = str(save_dir / imgName)
s += '%gx%g ' % img.shape[2:] # print string
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f"{n} {self.names[int(c)]}{'s' * (n > 1)}, " # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
label = f'{self.names[int(cls)]} {conf:.2f}'
plot_one_box(xyxy,
im0,
label=label,
color=self.colors[int(cls)],
line_thickness=3)
# Print time (inference + NMS)
result_text = f'{s}Done. ({t2 - t1:.3f}s)'
print(result_text)
if self.save_img:
if not cv2.imwrite(save_path, im0):
raise Exception("Could not write image to: " +
str(save_path))
extension = os.path.splitext(imgName)[1]
is_sucess, img_encoded = cv2.imencode(extension, im0)
if is_sucess:
byte_im = img_encoded.tobytes()
else:
print(f'>>Debug: filename={imgName}, exension={extension}')
raise Exception("Could not cvt to bytes: " + str(save_path))
print(f"Results saved to {save_dir}")
print(f'Done. ({time.time() - t0:.3f}s)')
del pred, im0s, img
return byte_im, str(result_text)
def test(
data,
weights=None,
batch_size=16,
imgsz=640,
conf_thres=0.001,
iou_thres=0.6, # for NMS
save_json=False,
single_cls=False,
augment=False,
verbose=False,
model=None,
dataloader=None,
save_dir=Path(''), # for saving images
save_txt=False, # for auto-labelling
save_conf=False,
plots=True,
log_imgs=0): # number of logged images
# Initialize/load model and set device
training = model is not None
if training: # called by train.py
device = next(model.parameters()).device # get model device
else: # called directly
set_logging()
device = select_device(opt.device, batch_size=batch_size)
save_txt = opt.save_txt # save *.txt labels
# Directories
if save_dir == Path('runs/test'): # if default
save_dir.mkdir(parents=True, exist_ok=True) # make base
save_dir = Path(increment_dir(save_dir / 'exp',
opt.name)) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(
parents=True, exist_ok=True) # make new dir
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
# Multi-GPU disabled, incompatible with .half() https://github.com/ultralytics/yolov5/issues/99
# if device.type != 'cpu' and torch.cuda.device_count() > 1:
# model = nn.DataParallel(model)
# Half
half = device.type != 'cpu' # half precision only supported on CUDA
if half:
model.half()
# Configure
model.eval()
with open(data) as f:
data = yaml.load(f, Loader=yaml.FullLoader) # model dict
check_dataset(data) # check
nc = 1 if single_cls else int(data['nc']) # number of classes
iouv = torch.linspace(0.5, 0.95,
10).to(device) # iou vector for [email protected]:0.95
niou = iouv.numel()
# Logging
log_imgs = min(log_imgs, 100) # ceil
try:
import wandb # Weights & Biases
except ImportError:
log_imgs = 0
# Dataloader
if not training:
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img
) if device.type != 'cpu' else None # run once
path = data['test'] if opt.task == 'test' else data[
'val'] # path to val/test images
dataloader = create_dataloader(path,
imgsz,
batch_size,
model.stride.max(),
opt,
hyp=None,
augment=False,
cache=False,
pad=0.5,
rect=True)[0]
seen = 0
names = {
k: v
for k, v in enumerate(
model.names if hasattr(model, 'names') else model.module.names)
}
coco91class = coco80_to_coco91_class()
s = ('%20s' + '%12s' * 6) % ('Class', 'Images', 'Targets', 'P', 'R',
'[email protected]', '[email protected]:.95')
p, r, f1, mp, mr, map50, map, t0, t1 = 0., 0., 0., 0., 0., 0., 0., 0., 0.
loss = torch.zeros(3, device=device)
jdict, stats, ap, ap_class, wandb_images = [], [], [], [], []
for batch_i, (img, targets, paths,
shapes) in enumerate(tqdm(dataloader, desc=s)):
img = img.to(device, non_blocking=True)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
targets = targets.to(device)
nb, _, height, width = img.shape # batch size, channels, height, width
whwh = torch.Tensor([width, height, width, height]).to(device)
# Disable gradients
with torch.no_grad():
# Run model
t = time_synchronized()
inf_out, train_out = model(
img, augment=augment) # inference and training outputs
t0 += time_synchronized() - t
# Compute loss
if training: # if model has loss hyperparameters
loss += compute_loss([x.float() for x in train_out], targets,
model)[1][:3] # box, obj, cls
# Run NMS
t = time_synchronized()
output = non_max_suppression(inf_out,
conf_thres=conf_thres,
iou_thres=iou_thres)
t1 += time_synchronized() - t
# Statistics per image
for si, pred in enumerate(output):
labels = targets[targets[:, 0] == si, 1:]
nl = len(labels)
tcls = labels[:, 0].tolist() if nl else [] # target class
seen += 1
if pred is None:
if nl:
stats.append((torch.zeros(0, niou, dtype=torch.bool),
torch.Tensor(), torch.Tensor(), tcls))
continue
# Append to text file
if save_txt:
gn = torch.tensor(shapes[si][0])[[1, 0, 1, 0
]] # normalization gain whwh
x = pred.clone()
x[:, :4] = scale_coords(img[si].shape[1:], x[:, :4],
shapes[si][0],
shapes[si][1]) # to original
for *xyxy, conf, cls in x:
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh,
conf) if save_conf else (cls,
*xywh) # label format
with open(
str(save_dir / 'labels' / Path(paths[si]).stem) +
'.txt', 'a') as f:
f.write(('%g ' * len(line) + '\n') % line)
# W&B logging
if len(wandb_images) < log_imgs:
box_data = [{
"position": {
"minX": xyxy[0],
"minY": xyxy[1],
"maxX": xyxy[2],
"maxY": xyxy[3]
},
"class_id": int(cls),
"box_caption": "%s %.3f" % (names[cls], conf),
"scores": {
"class_score": conf
},
"domain": "pixel"
} for *xyxy, conf, cls in pred.clone().tolist()]
boxes = {
"predictions": {
"box_data": box_data,
"class_labels": names
}
}
wandb_images.append(wandb.Image(img[si], boxes=boxes))
# Clip boxes to image bounds
clip_coords(pred, (height, width))
# Append to pycocotools JSON dictionary
if save_json:
# [{"image_id": 42, "category_id": 18, "bbox": [258.15, 41.29, 348.26, 243.78], "score": 0.236}, ...
image_id = Path(paths[si]).stem
box = pred[:, :4].clone() # xyxy
scale_coords(img[si].shape[1:], box, shapes[si][0],
shapes[si][1]) # to original shape
box = xyxy2xywh(box) # xywh
box[:, :2] -= box[:, 2:] / 2 # xy center to top-left corner
for p, b in zip(pred.tolist(), box.tolist()):
jdict.append({
'image_id':
int(image_id) if image_id.isnumeric() else image_id,
'category_id':
coco91class[int(p[5])],
'bbox': [round(x, 3) for x in b],
'score':
round(p[4], 5)
})
# Assign all predictions as incorrect
correct = torch.zeros(pred.shape[0],
niou,
dtype=torch.bool,
device=device)
if nl:
detected = [] # target indices
tcls_tensor = labels[:, 0]
# target boxes
tbox = xywh2xyxy(labels[:, 1:5]) * whwh
# Per target class
for cls in torch.unique(tcls_tensor):
ti = (cls == tcls_tensor).nonzero(as_tuple=False).view(
-1) # prediction indices
pi = (cls == pred[:, 5]).nonzero(as_tuple=False).view(
-1) # target indices
# Search for detections
if pi.shape[0]:
# Prediction to target ious
ious, i = box_iou(pred[pi, :4], tbox[ti]).max(
1) # best ious, indices
# Append detections
detected_set = set()
for j in (ious > iouv[0]).nonzero(as_tuple=False):
d = ti[i[j]] # detected target
if d.item() not in detected_set:
detected_set.add(d.item())
detected.append(d)
correct[
pi[j]] = ious[j] > iouv # iou_thres is 1xn
if len(
detected
) == nl: # all targets already located in image
break
# Append statistics (correct, conf, pcls, tcls)
stats.append(
(correct.cpu(), pred[:, 4].cpu(), pred[:, 5].cpu(), tcls))
# Plot images
if plots and batch_i < 1:
f = save_dir / f'test_batch{batch_i}_gt.jpg' # filename
plot_images(img, targets, paths, str(f), names) # ground truth
f = save_dir / f'test_batch{batch_i}_pred.jpg'
plot_images(img, output_to_target(output, width, height), paths,
str(f), names) # predictions
# W&B logging
if wandb_images:
wandb.log({"outputs": wandb_images})
# Compute statistics
stats = [np.concatenate(x, 0) for x in zip(*stats)] # to numpy
if len(stats) and stats[0].any():
p, r, ap, f1, ap_class = ap_per_class(*stats,
plot=plots,
fname=save_dir /
'precision-recall_curve.png')
p, r, ap50, ap = p[:, 0], r[:, 0], ap[:, 0], ap.mean(
1) # [P, R, [email protected], [email protected]:0.95]
mp, mr, map50, map = p.mean(), r.mean(), ap50.mean(), ap.mean()
nt = np.bincount(stats[3].astype(np.int64),
minlength=nc) # number of targets per class
else:
nt = torch.zeros(1)
# Print results
pf = '%20s' + '%12.3g' * 6 # print format
print(pf % ('all', seen, nt.sum(), mp, mr, map50, map))
# Print results per class
if verbose and nc > 1 and len(stats):
for i, c in enumerate(ap_class):
print(pf % (names[c], seen, nt[c], p[i], r[i], ap50[i], ap[i]))
# Print speeds
t = tuple(x / seen * 1E3
for x in (t0, t1, t0 + t1)) + (imgsz, imgsz, batch_size) # tuple
if not training:
print(
'Speed: %.1f/%.1f/%.1f ms inference/NMS/total per %gx%g image at batch-size %g'
% t)
# Save JSON
if save_json and len(jdict):
w = Path(weights[0] if isinstance(weights, list) else weights
).stem if weights is not None else '' # weights
file = save_dir / f"detections_val2017_{w}_results.json" # predicted annotations file
print('\nCOCO mAP with pycocotools... saving %s...' % file)
with open(file, 'w') as f:
json.dump(jdict, f)
try: # https://github.com/cocodataset/cocoapi/blob/master/PythonAPI/pycocoEvalDemo.ipynb
from pycocotools.coco import COCO
from pycocotools.cocoeval import COCOeval
imgIds = [int(Path(x).stem) for x in dataloader.dataset.img_files]
cocoGt = COCO(
glob.glob('../coco/annotations/instances_val*.json')
[0]) # initialize COCO ground truth api
cocoDt = cocoGt.loadRes(str(file)) # initialize COCO pred api
cocoEval = COCOeval(cocoGt, cocoDt, 'bbox')
cocoEval.params.imgIds = imgIds # image IDs to evaluate
cocoEval.evaluate()
cocoEval.accumulate()
cocoEval.summarize()
map, map50 = cocoEval.stats[:
2] # update results ([email protected]:0.95, [email protected])
except Exception as e:
print('ERROR: pycocotools unable to run: %s' % e)
# Return results
if not training:
print('Results saved to %s' % save_dir)
model.float() # for training
maps = np.zeros(nc) + map
for i, c in enumerate(ap_class):
maps[c] = ap[i]
return (mp, mr, map50, map,
*(loss.cpu() / len(dataloader)).tolist()), maps, t
def detect(save_img=False):
out, source, weights, imgsz = \
opt.output, opt.source, opt.weights, opt.img_size
# Initialize
set_logging()
device = select_device(opt.device)
if os.path.exists(out):
shutil.rmtree(out) # delete output folder
os.makedirs(out) # make new output folder
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(
torch.load('weights/resnet101.pt',
map_location=device)['model']) # load weights
modelc.to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
save_img = True
dataset = LoadImages(source, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)]
for _ in range(len(names))]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half()
if half else img) #if device.type != 'cpu' else None # run once
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
opt.conf_thres,
opt.iou_thres,
classes=opt.classes,
agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
p, s, im0 = path, '', im0s
save_path = str(Path(out) / Path(p).name)
txt_path = str(Path(out) / Path(p).stem) + (
'_%g' % dataset.frame if dataset.mode == 'video' else '')
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Write results
img2 = im0.copy()
nperson = []
nname = []
for *xyxy, conf, cls in reversed(det):
if save_img: # Add bbox to image
label = '%s %.2f' % (names[int(cls)], conf)
########################################################################################################
##classes 변수 생성 (이름)
classes = names[int(cls)]
##classes 변수 함수에 추가
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3,
classes=classes)
##사람이라고 판단한 물체의 각 좌표 리스트에 저장
if classes == 'person':
nperson.append([
int(xyxy[0]),
int(xyxy[1]),
int(xyxy[2]),
int(xyxy[3])
])
if classes == 'name_tag':
nname.append([
int(xyxy[0]),
int(xyxy[1]),
int(xyxy[2]),
int(xyxy[3])
])
##사람이 아닌 리스트의 크기가 0보다 클 때 미리 복사해둔 프레임의 구역으로 이미지 덮기
if len(nname) > 0:
for ii in range(len(nname)):
for pi in range(len(nperson)):
if nname[ii][1] >= nperson[pi][1] and nname[ii][
3] <= nperson[pi][3] and nname[ii][
0] >= nperson[pi][0] and nname[ii][
2] <= nperson[pi][2]:
proi = img2[nname[ii][1]:nname[ii][3],
nname[ii][0]:nname[ii][2]]
cv2.imwrite(
"./temp/{0}_{1}_{2}_{3}.jpg".format(
nname[ii][1], nname[ii][3],
nname[ii][0], nname[ii][2]), proi)
roi = img2[nperson[pi][1]:nperson[pi][3],
nperson[pi][0]:nperson[pi][2]]
im0[nperson[pi][1]:nperson[pi][3],
nperson[pi][0]:nperson[pi][2]] = roi
########################################################################################################
# Print time (inference + NMS)
print('%sDone. (%.3fs)' % (s, t2 - t1))
# Save results (image with detections)
if save_img:
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release() # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
vid_writer = cv2.VideoWriter(
'./inference/output/output.mp4',
cv2.VideoWriter_fourcc(*fourcc), fps, (w, h))
vid_writer.write(im0)
def run(
data,
weights=None, # model.pt path(s)
batch_size=32, # batch size
imgsz=640, # inference size (pixels)
conf_thres=0.001, # confidence threshold
iou_thres=0.6, # NMS IoU threshold
task='val', # train, val, test, speed or study
device='', # cuda device, i.e. 0 or 0,1,2,3 or cpu
workers=8, # max dataloader workers (per RANK in DDP mode)
single_cls=False, # treat as single-class dataset
augment=False, # augmented inference
verbose=False, # verbose output
save_txt=False, # save results to *.txt
save_hybrid=False, # save label+prediction hybrid results to *.txt
save_conf=False, # save confidences in --save-txt labels
save_json=False, # save a COCO-JSON results file
project=ROOT / 'runs/val', # save to project/name
name='exp', # save to project/name
exist_ok=False, # existing project/name ok, do not increment
half=True, # use FP16 half-precision inference
dnn=False, # use OpenCV DNN for ONNX inference
model=None,
dataloader=None,
save_dir=Path(''),
plots=True,
callbacks=Callbacks(),
compute_loss=None,
):
# Initialize/load model and set device
training = model is not None
if training: # called by train.py
device, pt, jit, engine = next(model.parameters()).device, True, False, False # get model device, PyTorch model
half &= device.type != 'cpu' # half precision only supported on CUDA
model.half() if half else model.float()
else: # called directly
device = select_device(device, batch_size=batch_size)
# Directories
save_dir = increment_path(Path(project) / name, exist_ok=exist_ok) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
# Load model
model = DetectMultiBackend(weights, device=device, dnn=dnn, data=data, fp16=half)
stride, pt, jit, engine = model.stride, model.pt, model.jit, model.engine
imgsz = check_img_size(imgsz, s=stride) # check image size
half = model.fp16 # FP16 supported on limited backends with CUDA
if engine:
batch_size = model.batch_size
else:
device = model.device
if not (pt or jit):
batch_size = 1 # export.py models default to batch-size 1
LOGGER.info(f'Forcing --batch-size 1 square inference (1,3,{imgsz},{imgsz}) for non-PyTorch models')
# Data
data = check_dataset(data) # check
# Configure
model.eval()
cuda = device.type != 'cpu'
is_coco = isinstance(data.get('val'), str) and data['val'].endswith(f'coco{os.sep}val2017.txt') # COCO dataset
nc = 1 if single_cls else int(data['nc']) # number of classes
iouv = torch.linspace(0.5, 0.95, 10, device=device) # iou vector for [email protected]:0.95
niou = iouv.numel()
# Dataloader
if not training:
if pt and not single_cls: # check --weights are trained on --data
ncm = model.model.nc
assert ncm == nc, f'{weights[0]} ({ncm} classes) trained on different --data than what you passed ({nc} ' \
f'classes). Pass correct combination of --weights and --data that are trained together.'
model.warmup(imgsz=(1 if pt else batch_size, 3, imgsz, imgsz)) # warmup
pad = 0.0 if task in ('speed', 'benchmark') else 0.5
rect = False if task == 'benchmark' else pt # square inference for benchmarks
task = task if task in ('train', 'val', 'test') else 'val' # path to train/val/test images
dataloader = create_dataloader(data[task],
imgsz,
batch_size,
stride,
single_cls,
pad=pad,
rect=rect,
workers=workers,
prefix=colorstr(f'{task}: '))[0]
seen = 0
confusion_matrix = ConfusionMatrix(nc=nc)
names = {k: v for k, v in enumerate(model.names if hasattr(model, 'names') else model.module.names)}
class_map = coco80_to_coco91_class() if is_coco else list(range(1000))
s = ('%20s' + '%11s' * 6) % ('Class', 'Images', 'Labels', 'P', 'R', '[email protected]', '[email protected]:.95')
dt, p, r, f1, mp, mr, map50, map = [0.0, 0.0, 0.0], 0.0, 0.0, 0.0, 0.0, 0.0, 0.0, 0.0
loss = torch.zeros(3, device=device)
jdict, stats, ap, ap_class = [], [], [], []
callbacks.run('on_val_start')
pbar = tqdm(dataloader, desc=s, bar_format='{l_bar}{bar:10}{r_bar}{bar:-10b}') # progress bar
for batch_i, (im, targets, paths, shapes) in enumerate(pbar):
callbacks.run('on_val_batch_start')
t1 = time_sync()
if cuda:
im = im.to(device, non_blocking=True)
targets = targets.to(device)
im = im.half() if half else im.float() # uint8 to fp16/32
im /= 255 # 0 - 255 to 0.0 - 1.0
nb, _, height, width = im.shape # batch size, channels, height, width
t2 = time_sync()
dt[0] += t2 - t1
# Inference
out, train_out = model(im) if training else model(im, augment=augment, val=True) # inference, loss outputs
dt[1] += time_sync() - t2
# Loss
if compute_loss:
loss += compute_loss([x.float() for x in train_out], targets)[1] # box, obj, cls
# NMS
targets[:, 2:] *= torch.tensor((width, height, width, height), device=device) # to pixels
lb = [targets[targets[:, 0] == i, 1:] for i in range(nb)] if save_hybrid else [] # for autolabelling
t3 = time_sync()
out = non_max_suppression(out, conf_thres, iou_thres, labels=lb, multi_label=True, agnostic=single_cls)
dt[2] += time_sync() - t3
# Metrics
for si, pred in enumerate(out):
labels = targets[targets[:, 0] == si, 1:]
nl, npr = labels.shape[0], pred.shape[0] # number of labels, predictions
path, shape = Path(paths[si]), shapes[si][0]
correct = torch.zeros(npr, niou, dtype=torch.bool, device=device) # init
seen += 1
if npr == 0:
if nl:
stats.append((correct, *torch.zeros((3, 0), device=device)))
continue
# Predictions
if single_cls:
pred[:, 5] = 0
predn = pred.clone()
scale_coords(im[si].shape[1:], predn[:, :4], shape, shapes[si][1]) # native-space pred
# Evaluate
if nl:
tbox = xywh2xyxy(labels[:, 1:5]) # target boxes
scale_coords(im[si].shape[1:], tbox, shape, shapes[si][1]) # native-space labels
labelsn = torch.cat((labels[:, 0:1], tbox), 1) # native-space labels
correct = process_batch(predn, labelsn, iouv)
if plots:
confusion_matrix.process_batch(predn, labelsn)
stats.append((correct, pred[:, 4], pred[:, 5], labels[:, 0])) # (correct, conf, pcls, tcls)
# Save/log
if save_txt:
save_one_txt(predn, save_conf, shape, file=save_dir / 'labels' / (path.stem + '.txt'))
if save_json:
save_one_json(predn, jdict, path, class_map) # append to COCO-JSON dictionary
callbacks.run('on_val_image_end', pred, predn, path, names, im[si])
# Plot images
if plots and batch_i < 3:
plot_images(im, targets, paths, save_dir / f'val_batch{batch_i}_labels.jpg', names) # labels
plot_images(im, output_to_target(out), paths, save_dir / f'val_batch{batch_i}_pred.jpg', names) # pred
callbacks.run('on_val_batch_end')
# Compute metrics
stats = [torch.cat(x, 0).cpu().numpy() for x in zip(*stats)] # to numpy
if len(stats) and stats[0].any():
tp, fp, p, r, f1, ap, ap_class = ap_per_class(*stats, plot=plots, save_dir=save_dir, names=names)
ap50, ap = ap[:, 0], ap.mean(1) # [email protected], [email protected]:0.95
mp, mr, map50, map = p.mean(), r.mean(), ap50.mean(), ap.mean()
nt = np.bincount(stats[3].astype(int), minlength=nc) # number of targets per class
else:
nt = torch.zeros(1)
# Print results
pf = '%20s' + '%11i' * 2 + '%11.3g' * 4 # print format
LOGGER.info(pf % ('all', seen, nt.sum(), mp, mr, map50, map))
# Print results per class
if (verbose or (nc < 50 and not training)) and nc > 1 and len(stats):
for i, c in enumerate(ap_class):
LOGGER.info(pf % (names[c], seen, nt[c], p[i], r[i], ap50[i], ap[i]))
# Print speeds
t = tuple(x / seen * 1E3 for x in dt) # speeds per image
if not training:
shape = (batch_size, 3, imgsz, imgsz)
LOGGER.info(f'Speed: %.1fms pre-process, %.1fms inference, %.1fms NMS per image at shape {shape}' % t)
# Plots
if plots:
confusion_matrix.plot(save_dir=save_dir, names=list(names.values()))
callbacks.run('on_val_end')
# Save JSON
if save_json and len(jdict):
w = Path(weights[0] if isinstance(weights, list) else weights).stem if weights is not None else '' # weights
anno_json = str(Path(data.get('path', '../coco')) / 'annotations/instances_val2017.json') # annotations json
pred_json = str(save_dir / f"{w}_predictions.json") # predictions json
LOGGER.info(f'\nEvaluating pycocotools mAP... saving {pred_json}...')
with open(pred_json, 'w') as f:
json.dump(jdict, f)
try: # https://github.com/cocodataset/cocoapi/blob/master/PythonAPI/pycocoEvalDemo.ipynb
check_requirements(['pycocotools'])
from pycocotools.coco import COCO
from pycocotools.cocoeval import COCOeval
anno = COCO(anno_json) # init annotations api
pred = anno.loadRes(pred_json) # init predictions api
eval = COCOeval(anno, pred, 'bbox')
if is_coco:
eval.params.imgIds = [int(Path(x).stem) for x in dataloader.dataset.im_files] # image IDs to evaluate
eval.evaluate()
eval.accumulate()
eval.summarize()
map, map50 = eval.stats[:2] # update results ([email protected]:0.95, [email protected])
except Exception as e:
LOGGER.info(f'pycocotools unable to run: {e}')
# Return results
model.float() # for training
if not training:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
LOGGER.info(f"Results saved to {colorstr('bold', save_dir)}{s}")
maps = np.zeros(nc) + map
for i, c in enumerate(ap_class):
maps[c] = ap[i]
return (mp, mr, map50, map, *(loss.cpu() / len(dataloader)).tolist()), maps, t
def detect(self,save_img=False):
# Get names and colors
names = self.model.module.names if hasattr(self.model, 'module') else self.model.names
#colors = [[random.randint(0, 255) for _ in range(3)] for _ in range(len(names))]
# Run inference
#t0 = time.time()
img = torch.zeros((1, 3, self.imgsz, self.imgsz), device=self.device) # init img
_ = self.model(img.half() if self.half else img) if self.device.type != 'cpu' else None # run once
for path, img, im0s, vid_cap in self.dataset:
img = torch.from_numpy(img).to(self.device)
img = img.half() if self.half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
path_split = path.split("/")
file_info= path_split[len(path_split)-1]
file_info_split = file_info.split("_")
date_time = file_info_split[3] +":" + file_info_split[4] + ":" + file_info_split[5]
# Inference
t1 = time_synchronized()
pred = self.model(img, augment=self.opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred, self.opt.conf_thres, self.opt.iou_thres, classes=self.opt.classes,
agnostic=self.opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if self.classify:
pred = apply_classifier(pred, self.modelc, img, im0s)
#print("pred",pred)
# Process detections
for i, det in enumerate(pred): # detections per image
'''
if self.webcam: # batch_size >= 1
p, s, im0 = path[i], '%g: ' % i, im0s[i].copy()
else:
'''
p, s, im0 = path, '', im0s
save_path = str(Path(self.out) / Path(p).name)
txt_path = str(Path(self.out) / Path(p).stem) + ('_%g' % self.dataset.frame if self.dataset.mode == 'video' else '')
#s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
# detect 했을 경우
if det is not None and len(det):
total = 0.0
# Rescale boxes from img_size to im0 size
#print("type : " , type(det))
#print("det : " , det)
goods_type = None
percent = None
more_than_90 = 0
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += '%g %s, ' % (n, names[int(c)]) # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
if self.save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist() # normalized xywh
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * 5 + '\n') % (cls, *xywh)) # label format
#print(cls)
if self.save_img or self.view_img: # Add bbox to image
goods_type = names[int(cls)]
percent = '%.2f' % (conf)
#print(type(percent))
percent = float(percent)
#print("percent",type(percent))
label = '%s %.2f' % (names[int(cls)], conf)
print(percent)
if percent > 0.85:
#plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=3)
plot_one_box(xyxy, im0, label=label, color=(0,0,255), line_thickness=3)
total = total + percent
more_than_90 += 1
#avg = total/len(det)
if more_than_90 != 0 :
avg = total/more_than_90
avg = round(avg,2)
#print(total)
#print(more_than_90)
#print(avg)
print("names : ", names[int(cls)])
#print("확률 : %.2f", percent )
cv_img, name, color = self.d.load_image(goods_type)
if cv_img is not None :
qt_img = self.convert_cv_qt(cv_img)
self.updateFeatureLable(qt_img)
#self.infomsg_append("[DETECT] 품종 : %s, 코드 : %s, 개수 : %d" % (name, goods_type, len(det)))
#img_time = datetime.datetime.now().strftime("%Y-%m-%d,%H:%M:%S")
img_time = datetime.datetime.now().strftime("%H:%M:%S")
img_date = datetime.datetime.now().strftime("%Y_%m_%d")
self.infomsg_append(img_time+",%s,%s,%d" % (name, goods_type, more_than_90))
#log_string = img_time + "," + name + ","+goods_type+"," + str(len(det)) +","+ avg
log_string = date_time + "," + name + ","+goods_type+"," + str(more_than_90) +","+ str(avg)
try:
if not os.path.exists("log"):
os.makedirs("log")
except OSError:
print('Error: Creating directory. log')
f = open("./log/2020_10_13_log.csv", mode='at', encoding='utf-8')
f.writelines(log_string+'\n')
f.close()
print(log_string)
#print("db 이미지 업로드 성공")
#detect 없을 시
else :
print("해당 품목 db에서 조회불가 ")
self.iv.clear()
self.f_label.clear()
else :
print("detect 없음")
#self.infomsg_append("[DETECT] 위 품종은 신규 학습이 필요합니다.")
#self.infomsg_append("detect 학습 필요")
print(s)
# Print time (inference + NMS)
print('%sDone. (%.3fs)' % (s, t2 - t1))
self.iv.setImage(self.convert_cv_qt(im0))
# Stream results
if self.view_img:
cv2.imshow(p, im0)
if cv2.waitKey(1) == ord('q'): # q to quit
raise StopIteration
# Save results (image with detections)
if self.save_img:
if self.dataset.mode == 'images':
cv2.imwrite(save_path, im0)
else:
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release() # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
vid_writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*fourcc), fps, (w, h))
vid_writer.write(im0)
if self.save_txt or self.save_img:
print('Results saved to %s' % Path(self.out))
if platform.system() == 'Darwin' and not self.opt.update: # MacOS
os.system('open ' + save_path)
def detect(self,
app,
weights,
source='data/images',
image_size=736,
debug=False):
""" Detects the image or video of the given source by using the specified weights.
Parameters
----------
weights: str
Weights path.
source: str
Source of the detection. 0 for webcam.
image_size: int
Inference size (pixels).
debug: bool
Whether the debug mode is on.
"""
print(f'Detecting using weights: {weights}')
source = str(source)
imgsz = image_size
opt_project = 'runs/detect'
opt_name = 'exp'
opt_exist_ok = False
opt_device = ''
opt_augment = False
opt_conf_thres = 0.25
opt_iou_thres = 0.01
opt_classes = 0
opt_agnostic_nms = True
opt_save_conf = False
webcam = False
if source.isnumeric() or source.endswith(
'.txt') or source.lower().startswith(
('rtsp://', 'rtmp://', 'http://')):
print(source)
webcam = source.isnumeric() or source.endswith(
'.txt') or source.lower().startswith(
('rtsp://', 'rtmp://', 'http://'))
# Initialize
set_logging()
device = select_device(opt_device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(
torch.load('weights/resnet101.pt',
map_location=device)['model']).to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
if webcam:
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadStreams(source, img_size=imgsz)
print("webcam") # debug
logging.debug("webcam")
else:
dataset = LoadImages(source, img_size=imgsz)
print("image") # debug
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img
) if device.type != 'cpu' else None # run once
logging.debug("start inference")
for path, img, im0s, vid_cap in dataset:
if self.detection is False and webcam is True:
logging.debug("kill thread")
dataset.kill_thread()
break
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt_augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
opt_conf_thres,
opt_iou_thres,
classes=opt_classes,
agnostic=opt_agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0, frame = path[i], '%g: ' % i, im0s[i].copy(
), dataset.count
else:
p, s, im0, frame = path, '', im0s, getattr(
dataset, 'frame', 0)
#p = Path(p) # to Path
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f'{n} {names[int(c)]}s, ' # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
logging.debug('Prediction: {}; Confidence {}'.format(
f'{names[int(cls)]}', f'{conf:.2f}'))
label = ''.join(
map(lambda x: x
if x.islower() else ' ' + x, names[int(cls)]))
label = f'{label} {conf:.2f}' if debug else label
if debug:
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3)
plot_one_point(xyxy,
im0,
label=label,
color=colors[int(cls)],
point_thickness=None,
r=10)
app.image = QImage(bytearray(im0), im0.shape[1], im0.shape[0],
QImage.Format_RGB888).rgbSwapped()
app.Label_Bild.setPixmap(QPixmap(app.image))
time.sleep(1 / 60)
# Print time (inference + NMS)
print(f'{s}Done. ({t2 - t1:.3f}s)')
logging.debug(f'Done. ({time.time() - t0:.3f}s)')
print(f'Done. ({time.time() - t0:.3f}s)')
def multithreading(video, device, half, names, colors, model, q):
frame = video
t0 = time.time()
fps_t1 = time.time()
img, img0 = img_preprocess(frame)
# img: Resize , img0:Orginal
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float()
# uint8 to fp16/32
img /= 255.0
# 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS : 取得每項預測的數值
pred = non_max_suppression(pred,
opt.conf_thres,
opt.iou_thres,
classes=opt.classes,
agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier : 取得該數值的LAbel
if False:
pred = apply_classifier(pred, modelc, img, img0)
# Draw Box
for i, det in enumerate(pred):
s = '%gx%g ' % img.shape[2:]
# print string
gn = torch.tensor(img0.shape)[[1, 0, 1, 0]]
# normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
img0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum()
# detections per class
s += '%g %ss, ' % (n, names[int(c)])
# add to string
# Write results
for *xyxy, conf, cls in reversed(det):
label = '%s %.2f' % (names[int(cls)], conf)
plot_one_box(xyxy,
img0,
label=label,
color=colors[int(cls)],
line_thickness=3)
# Print Results(inference + NMS)
print('%sDone. (%.3fs)' % (s, t2 - t1))
# Draw Image
x, y, w, h = (img0.shape[1] // 4), 25, (img0.shape[1] // 2), 30
cv2.rectangle(img0, (x, 10), (x + w, y + h), (0, 0, 0), -1)
cv2.putText(
img0,
'{} | inference: {:.4f}s | fps: {:.4f}'.format(opt.weights[0], t2 - t1,
1 / (time.time() - t0)),
(x + 20, y + 20), cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 0, 255), 2)
q.put(img0)
def detect(save_img=False):
warning = 0
_sum = 0
source, weights, view_img, save_txt, imgsz = opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size
save_img = not opt.nosave and not source.endswith(
'.txt') # save inference images
webcam = source.isnumeric() or source.endswith(
'.txt') or source.lower().startswith(
('rtsp://', 'rtmp://', 'http://')) or source == 'realsense'
# Directories
save_dir = Path(
increment_path(Path(opt.project) / opt.name,
exist_ok=opt.exist_ok)) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(
parents=True, exist_ok=True) # make dir
# Initialize
set_logging()
device = select_device(opt.device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
stride = int(model.stride.max()) # model stride
imgsz = check_img_size(imgsz, s=stride) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(
torch.load('weights/resnet101.pt',
map_location=device)['model']).to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
if webcam:
view_img = check_imshow()
cudnn.benchmark = True # set True to speed up constant image size inference
# dataset = LoadStreams(source, img_size=imgsz, stride=stride)
if source == '1':
#print('우헤헤헤헤')
dataset = LoadRealSense2()
else:
#print('하ㅏ하하하')
dataset = LoadStreams(source, img_size=imgsz)
else:
dataset = LoadImages(source, img_size=imgsz, stride=stride)
#print('오이잉이이이')
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
# Run inference
if device.type != 'cpu':
model(
torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(
next(model.parameters()))) # run once
t0 = time.time()
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
opt.conf_thres,
opt.iou_thres,
classes=opt.classes,
agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0, frame = path[i], '%g: ' % i, im0s[i].copy(
), dataset.count
else:
p, s, im0, frame = path, '', im0s, getattr(dataset, 'frame', 0)
p = Path(p) # to Path
save_path = str(save_dir / p.name) # img.jpg
txt_path = str(save_dir / 'labels' / p.stem) + (
'' if dataset.mode == 'image' else f'_{frame}') # img.txt
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
if names[int(c)] == ('no-helmet'):
warning = 1
_sum += 1
if names[int(c)] == ('no-helmets'):
warning = 1
_sum += 1
if names[int(c)] == ('no-mask'):
warning = 1
_sum += 1
if names[int(c)] == ('no-masks'):
warning = 1
_sum += 1
if names[int(c)] == ('person'):
if n == 1:
warning = 1
_sum += 1
# Write results
for *xyxy, conf, cls in reversed(det):
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh, conf) if opt.save_conf else (
cls, *xywh) # label format
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * len(line)).rstrip() % line + '\n')
if save_img or view_img: # Add bbox to image
label = f'{names[int(cls)]} {conf:.2f}'
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3)
#if 'helmet' in names[int(cls)]:
#print('캬캬캬캬캬캬',names[int(cls)])
#print('ㅋㅋㅋㅋㅋ',xyxy)
# Print time (inference + NMS)
print(f'{s}Done. ({t2 - t1:.3f}s)')
file = open(
'/home/piai/MinkyuKim/Week6~10_AI/Project/SocketConnection/detect.txt',
'a')
idx = 0
if warning == 1:
#file = open('detect.txt', 'w')
idx += 1
if idx <= 200:
file.write('1' + '\n')
#idx += 1
# print(_sum)
# if _sum>=200:
# print('_sum은', _sum)
# file.write('20' + '\n')
# file.close()
print('stop')
warning = 0
else:
file.write('0' + '\n')
print('pass')
# file.close()
# Stream results
if view_img:
cv2.imshow(str(p), im0)
cv2.waitKey(1) # 1 millisecond
# Save results (image with detections)
if save_img:
if dataset.mode == 'image':
cv2.imwrite(save_path, im0)
else: # 'video' or 'stream'
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release(
) # release previous video writer
if vid_cap: # video
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
else: # stream
fps, w, h = 30, im0.shape[1], im0.shape[0]
save_path += '.mp4'
vid_writer = cv2.VideoWriter(
save_path, cv2.VideoWriter_fourcc(*'mp4v'), fps,
(w, h))
vid_writer.write(im0)
file.close()
if save_txt or save_img:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}")
print(f'Done. ({time.time() - t0:.3f}s)')
def run(
data,
weights=None, # model.pt path(s)
batch_size=32, # batch size
imgsz=640, # inference size (pixels)
conf_thres=0.001, # confidence threshold
iou_thres=0.6, # NMS IoU threshold
task='val', # train, val, test, speed or study
device='', # cuda device, i.e. 0 or 0,1,2,3 or cpu
single_cls=False, # treat as single-class dataset
augment=False, # augmented inference
verbose=False, # verbose output
save_txt=False, # save results to *.txt
save_hybrid=False, # save label+prediction hybrid results to *.txt
save_conf=False, # save confidences in --save-txt labels
save_json=False, # save a COCO-JSON results file
project='runs/val', # save to project/name
name='exp', # save to project/name
exist_ok=False, # existing project/name ok, do not increment
half=True, # use FP16 half-precision inference
model=None,
dataloader=None,
save_dir=Path(''),
plots=True,
wandb_logger=None,
compute_loss=None,
):
# Initialize/load model and set device
training = model is not None
if training: # called by train.py
device = next(model.parameters()).device # get model device
else: # called directly
device = select_device(device, batch_size=batch_size)
# Directories
save_dir = increment_path(Path(project) / name,
exist_ok=exist_ok) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(
parents=True, exist_ok=True) # make dir
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
gs = max(int(model.stride.max()), 32) # grid size (max stride)
imgsz = check_img_size(imgsz, s=gs) # check image size
# Multi-GPU disabled, incompatible with .half() https://github.com/ultralytics/yolov5/issues/99
# if device.type != 'cpu' and torch.cuda.device_count() > 1:
# model = nn.DataParallel(model)
# Data
with open(data) as f:
data = yaml.safe_load(f)
check_dataset(data) # check
# Half
half &= device.type != 'cpu' # half precision only supported on CUDA
if half:
model.half()
# Configure
model.eval()
is_coco = type(data['val']) is str and data['val'].endswith(
'coco/val2017.txt') # COCO dataset
nc = 1 if single_cls else int(data['nc']) # number of classes
iouv = torch.linspace(0.5, 0.95,
10).to(device) # iou vector for [email protected]:0.95
niou = iouv.numel()
# Dataloader
if not training:
if device.type != 'cpu':
model(
torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(
next(model.parameters()))) # run once
task = task if task in (
'train', 'val', 'test') else 'val' # path to train/val/test images
dataloader = create_dataloader(data[task],
imgsz,
batch_size,
gs,
single_cls,
pad=0.5,
rect=True,
prefix=colorstr(f'{task}: '))[0]
seen = 0
confusion_matrix = ConfusionMatrix(nc=nc)
names = {
k: v
for k, v in enumerate(
model.names if hasattr(model, 'names') else model.module.names)
}
class_map = coco80_to_coco91_class() if is_coco else list(range(1000))
s = ('%20s' + '%11s' * 6) % ('Class', 'Images', 'Labels', 'P', 'R',
'[email protected]', '[email protected]:.95')
p, r, f1, mp, mr, map50, map, t0, t1, t2 = 0., 0., 0., 0., 0., 0., 0., 0., 0., 0.
loss = torch.zeros(3, device=device)
jdict, stats, ap, ap_class = [], [], [], []
for batch_i, (img, targets, paths,
shapes) in enumerate(tqdm(dataloader, desc=s)):
t_ = time_sync()
img = img.to(device, non_blocking=True)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
targets = targets.to(device)
nb, _, height, width = img.shape # batch size, channels, height, width
t = time_sync()
t0 += t - t_
# Run model
out, train_out = model(
img, augment=augment) # inference and training outputs
t1 += time_sync() - t
# Compute loss
if compute_loss:
loss += compute_loss([x.float() for x in train_out],
targets)[1][:3] # box, obj, cls
# Run NMS
targets[:, 2:] *= torch.Tensor([width, height, width,
height]).to(device) # to pixels
lb = [targets[targets[:, 0] == i, 1:]
for i in range(nb)] if save_hybrid else [] # for autolabelling
t = time_sync()
out = non_max_suppression(out,
conf_thres,
iou_thres,
labels=lb,
multi_label=True,
agnostic=single_cls)
t2 += time_sync() - t
# Statistics per image
for si, pred in enumerate(out):
labels = targets[targets[:, 0] == si, 1:]
nl = len(labels)
tcls = labels[:, 0].tolist() if nl else [] # target class
path, shape = Path(paths[si]), shapes[si][0]
seen += 1
if len(pred) == 0:
if nl:
stats.append((torch.zeros(0, niou, dtype=torch.bool),
torch.Tensor(), torch.Tensor(), tcls))
continue
# Predictions
if single_cls:
pred[:, 5] = 0
predn = pred.clone()
scale_coords(img[si].shape[1:], predn[:, :4], shape,
shapes[si][1]) # native-space pred
# Evaluate
if nl:
tbox = xywh2xyxy(labels[:, 1:5]) # target boxes
scale_coords(img[si].shape[1:], tbox, shape,
shapes[si][1]) # native-space labels
labelsn = torch.cat((labels[:, 0:1], tbox),
1) # native-space labels
correct = process_batch(predn, labelsn, iouv)
if plots:
confusion_matrix.process_batch(predn, labelsn)
else:
correct = torch.zeros(pred.shape[0], niou, dtype=torch.bool)
stats.append((correct.cpu(), pred[:, 4].cpu(), pred[:, 5].cpu(),
tcls)) # (correct, conf, pcls, tcls)
# Save/log
if save_txt:
save_one_txt(predn,
save_conf,
shape,
file=save_dir / 'labels' / (path.stem + '.txt'))
if save_json:
save_one_json(predn, jdict, path,
class_map) # append to COCO-JSON dictionary
if wandb_logger and wandb_logger.wandb_run:
wandb_logger.val_one_image(pred, predn, path, names, img[si])
# Plot images
if plots and batch_i < 3:
f = save_dir / f'val_batch{batch_i}_labels.jpg' # labels
Thread(target=plot_images,
args=(img, targets, paths, f, names),
daemon=True).start()
f = save_dir / f'val_batch{batch_i}_pred.jpg' # predictions
Thread(target=plot_images,
args=(img, output_to_target(out), paths, f, names),
daemon=True).start()
# Compute statistics
stats = [np.concatenate(x, 0) for x in zip(*stats)] # to numpy
if len(stats) and stats[0].any():
p, r, ap, f1, ap_class = ap_per_class(*stats,
plot=plots,
save_dir=save_dir,
names=names)
ap50, ap = ap[:, 0], ap.mean(1) # [email protected], [email protected]:0.95
mp, mr, map50, map = p.mean(), r.mean(), ap50.mean(), ap.mean()
nt = np.bincount(stats[3].astype(np.int64),
minlength=nc) # number of targets per class
else:
nt = torch.zeros(1)
# Print results
pf = '%20s' + '%11i' * 2 + '%11.3g' * 4 # print format
print(pf % ('all', seen, nt.sum(), mp, mr, map50, map))
# Print results per class
if (verbose or (nc < 50 and not training)) and nc > 1 and len(stats):
for i, c in enumerate(ap_class):
print(pf % (names[c], seen, nt[c], p[i], r[i], ap50[i], ap[i]))
# Print speeds
t = tuple(x / seen * 1E3 for x in (t0, t1, t2)) # speeds per image
if not training:
shape = (batch_size, 3, imgsz, imgsz)
print(
f'Speed: %.1fms pre-process, %.1fms inference, %.1fms NMS per image at shape {shape}'
% t)
# Plots
if plots:
confusion_matrix.plot(save_dir=save_dir, names=list(names.values()))
if wandb_logger and wandb_logger.wandb:
val_batches = [
wandb_logger.wandb.Image(str(f), caption=f.name)
for f in sorted(save_dir.glob('val*.jpg'))
]
wandb_logger.log({"Validation": val_batches})
# Save JSON
if save_json and len(jdict):
w = Path(weights[0] if isinstance(weights, list) else weights
).stem if weights is not None else '' # weights
anno_json = str(
Path(data.get('path', '../coco')) /
'annotations/instances_val2017.json') # annotations json
pred_json = str(save_dir / f"{w}_predictions.json") # predictions json
print(f'\nEvaluating pycocotools mAP... saving {pred_json}...')
with open(pred_json, 'w') as f:
json.dump(jdict, f)
try: # https://github.com/cocodataset/cocoapi/blob/master/PythonAPI/pycocoEvalDemo.ipynb
check_requirements(['pycocotools'])
from pycocotools.coco import COCO
from pycocotools.cocoeval import COCOeval
anno = COCO(anno_json) # init annotations api
pred = anno.loadRes(pred_json) # init predictions api
eval = COCOeval(anno, pred, 'bbox')
if is_coco:
eval.params.imgIds = [
int(Path(x).stem) for x in dataloader.dataset.img_files
] # image IDs to evaluate
eval.evaluate()
eval.accumulate()
eval.summarize()
map, map50 = eval.stats[:
2] # update results ([email protected]:0.95, [email protected])
except Exception as e:
print(f'pycocotools unable to run: {e}')
# Return results
model.float() # for training
if not training:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}")
maps = np.zeros(nc) + map
for i, c in enumerate(ap_class):
maps[c] = ap[i]
return (mp, mr, map50, map,
*(loss.cpu() / len(dataloader)).tolist()), maps, t
def test(
data,
weights=None,
batch_size=32,
imgsz=640,
conf_thres=0.001,
iou_thres=0.6, # for NMS
save_json=False,
single_cls=False,
verbose=False,
model=None,
dataloader=None,
save_dir=Path(''), # for saving images
save_txt=False, # for auto-labelling
save_hybrid=False, # for hybrid auto-labelling
save_conf=False, # save auto-label confidences
plots=True,
log_imgs=0, # number of logged images
compute_loss=None):
# Initialize/load model and set device
training = model is not None
if training: # called by train.py
device = next(model.parameters()).device # get model device
else: # called directly
set_logging()
device = select_device(opt.device, batch_size=batch_size)
# Directories
save_dir = Path(
increment_path(Path(opt.project) / opt.name,
exist_ok=opt.exist_ok)) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(
parents=True, exist_ok=True) # make dir
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
# Multi-GPU disabled, incompatible with .half() https://github.com/ultralytics/yolov5/issues/99
# if device.type != 'cpu' and torch.cuda.device_count() > 1:
# model = nn.DataParallel(model)
# Half
half = device.type != 'cpu' # half precision only supported on CUDA
if half:
model.half()
# Configure
model.eval()
is_coco = data.endswith('coco.yaml') # is COCO dataset
with open(data) as f:
data = yaml.load(f, Loader=yaml.SafeLoader) # model dict
check_dataset(data) # check
nc = 1 if single_cls else int(data['nc']) # number of classes
iouv = torch.linspace(0.5, 0.95,
10).to(device) # iou vector for [email protected]:0.95
niou = iouv.numel()
# Logging
log_imgs, wandb = min(log_imgs, 100), None # ceil
try:
import wandb # Weights & Biases
except ImportError:
log_imgs = 0
# Dataloader
if not training:
if device.type != 'cpu':
model(
torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(
next(model.parameters()))) # run once
path = data['test'] if opt.task == 'test' else data[
'val'] # path to val/test images
dataloader = create_dataloader(
path,
imgsz,
batch_size,
model.stride.max(),
opt,
pad=0.5,
rect=True,
prefix=colorstr('test: ' if opt.task == 'test' else 'val: '))[0]
seen = 0
confusion_matrix = ConfusionMatrix(nc=nc)
names = {
k: v
for k, v in enumerate(
model.names if hasattr(model, 'names') else model.module.names)
}
coco91class = coco80_to_coco91_class()
s = ('%20s' + '%12s' * 6) % ('Class', 'Images', 'Targets', 'P', 'R',
'[email protected]', '[email protected]:.95')
p, r, f1, mp, mr, map50, map, t0, t1 = 0., 0., 0., 0., 0., 0., 0., 0., 0.
loss = torch.zeros(3, device=device)
jdict, stats, ap, ap_class, wandb_images = [], [], [], [], []
for batch_i, (img, targets, paths,
shapes) in enumerate(tqdm(dataloader, desc=s)):
img = img.to(device, non_blocking=True)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
targets = targets.to(device)
nb, _, height, width = img.shape # batch size, channels, height, width
with torch.no_grad():
# Run model
t = time_synchronized()
inf_out, train_out = model(img) # inference and training outputs
t0 += time_synchronized() - t
# Compute loss
if compute_loss:
loss += compute_loss([x.float() for x in train_out],
targets)[1][:3] # box, obj, cls
# Run NMS
targets[:, 2:] *= torch.Tensor([width, height, width,
height]).to(device) # to pixels
lb = [targets[targets[:, 0] == i, 1:] for i in range(nb)
] if save_hybrid else [] # for autolabelling
t = time_synchronized()
output = non_max_suppression(inf_out,
conf_thres=conf_thres,
iou_thres=iou_thres,
labels=lb)
t1 += time_synchronized() - t
# Statistics per image
for si, pred in enumerate(output):
labels = targets[targets[:, 0] == si, 1:]
nl = len(labels)
tcls = labels[:, 0].tolist() if nl else [] # target class
path = Path(paths[si])
seen += 1
if len(pred) == 0:
if nl:
stats.append((torch.zeros(0, niou, dtype=torch.bool),
torch.Tensor(), torch.Tensor(), tcls))
continue
# Predictions
predn = pred.clone()
scale_coords(img[si].shape[1:], predn[:, :4], shapes[si][0],
shapes[si][1]) # native-space pred
# Append to text file
if save_txt:
gn = torch.tensor(shapes[si][0])[[1, 0, 1, 0
]] # normalization gain whwh
for *xyxy, conf, cls in predn.tolist():
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh,
conf) if save_conf else (cls,
*xywh) # label format
with open(save_dir / 'labels' / (path.stem + '.txt'),
'a') as f:
f.write(('%g ' * len(line)).rstrip() % line + '\n')
# W&B logging
if plots and len(wandb_images) < log_imgs:
box_data = [{
"position": {
"minX": xyxy[0],
"minY": xyxy[1],
"maxX": xyxy[2],
"maxY": xyxy[3]
},
"class_id": int(cls),
"box_caption": "%s %.3f" % (names[cls], conf),
"scores": {
"class_score": conf
},
"domain": "pixel"
} for *xyxy, conf, cls in pred.tolist()]
boxes = {
"predictions": {
"box_data": box_data,
"class_labels": names
}
} # inference-space
wandb_images.append(
wandb.Image(img[si], boxes=boxes, caption=path.name))
# Append to pycocotools JSON dictionary
if save_json:
# [{"image_id": 42, "category_id": 18, "bbox": [258.15, 41.29, 348.26, 243.78], "score": 0.236}, ...
image_id = int(
path.stem) if path.stem.isnumeric() else path.stem
box = xyxy2xywh(predn[:, :4]) # xywh
box[:, :2] -= box[:, 2:] / 2 # xy center to top-left corner
for p, b in zip(pred.tolist(), box.tolist()):
jdict.append({
'image_id':
image_id,
'category_id':
coco91class[int(p[5])] if is_coco else int(p[5]),
'bbox': [round(x, 3) for x in b],
'score':
round(p[4], 5)
})
# Assign all predictions as incorrect
correct = torch.zeros(pred.shape[0],
niou,
dtype=torch.bool,
device=device)
if nl:
detected = [] # target indices
tcls_tensor = labels[:, 0]
# target boxes
tbox = xywh2xyxy(labels[:, 1:5])
scale_coords(img[si].shape[1:], tbox, shapes[si][0],
shapes[si][1]) # native-space labels
if plots:
confusion_matrix.process_batch(
predn, torch.cat((labels[:, 0:1], tbox), 1))
# Per target class
for cls in torch.unique(tcls_tensor):
ti = (cls == tcls_tensor).nonzero(as_tuple=False).view(
-1) # prediction indices
pi = (cls == pred[:, 5]).nonzero(as_tuple=False).view(
-1) # target indices
# Search for detections
if pi.shape[0]:
# Prediction to target ious
ious, i = box_iou(predn[pi, :4], tbox[ti]).max(
1) # best ious, indices
# Append detections
detected_set = set()
for j in (ious > iouv[0]).nonzero(as_tuple=False):
d = ti[i[j]] # detected target
if d.item() not in detected_set:
detected_set.add(d.item())
detected.append(d)
correct[
pi[j]] = ious[j] > iouv # iou_thres is 1xn
if len(
detected
) == nl: # all targets already located in image
break
# Append statistics (correct, conf, pcls, tcls)
stats.append(
(correct.cpu(), pred[:, 4].cpu(), pred[:, 5].cpu(), tcls))
# Plot images
if plots and batch_i < 3:
f = save_dir / f'test_batch{batch_i}_labels.jpg' # labels
Thread(target=plot_images,
args=(img, targets, paths, f, names),
daemon=True).start()
f = save_dir / f'test_batch{batch_i}_pred.jpg' # predictions
Thread(target=plot_images,
args=(img, output_to_target(output), paths, f, names),
daemon=True).start()
# Compute statistics
stats = [np.concatenate(x, 0) for x in zip(*stats)] # to numpy
if len(stats) and stats[0].any():
p, r, ap, f1, ap_class = ap_per_class(*stats,
plot=plots,
save_dir=save_dir,
names=names)
ap50, ap = ap[:, 0], ap.mean(1) # [email protected], [email protected]:0.95
mp, mr, map50, map = p.mean(), r.mean(), ap50.mean(), ap.mean()
nt = np.bincount(stats[3].astype(np.int64),
minlength=nc) # number of targets per class
else:
nt = torch.zeros(1)
# Print results
pf = '%20s' + '%12.3g' * 6 # print format
print(pf % ('all', seen, nt.sum(), mp, mr, map50, map))
# Print results per class
if (verbose or (nc < 50 and not training)) and nc > 1 and len(stats):
for i, c in enumerate(ap_class):
print(pf % (names[c], seen, nt[c], p[i], r[i], ap50[i], ap[i]))
# Print speeds
t = tuple(x / seen * 1E3
for x in (t0, t1, t0 + t1)) + (imgsz, imgsz, batch_size) # tuple
if not training:
print(
'Speed: %.1f/%.1f/%.1f ms inference/NMS/total per %gx%g image at batch-size %g'
% t)
# Plots
if plots:
confusion_matrix.plot(save_dir=save_dir, names=list(names.values()))
if wandb and wandb.run:
val_batches = [
wandb.Image(str(f), caption=f.name)
for f in sorted(save_dir.glob('test*.jpg'))
]
wandb.log({
"Images": wandb_images,
"Validation": val_batches
},
commit=False)
# Save JSON
if save_json and len(jdict):
w = Path(weights[0] if isinstance(weights, list) else weights
).stem if weights is not None else '' # weights
anno_json = '../coco/annotations/instances_val2017.json' # annotations json
pred_json = str(save_dir / f"{w}_predictions.json") # predictions json
print('\nEvaluating pycocotools mAP... saving %s...' % pred_json)
with open(pred_json, 'w') as f:
json.dump(jdict, f)
try: # https://github.com/cocodataset/cocoapi/blob/master/PythonAPI/pycocoEvalDemo.ipynb
from pycocotools.coco import COCO
from pycocotools.cocoeval import COCOeval
anno = COCO(anno_json) # init annotations api
pred = anno.loadRes(pred_json) # init predictions api
eval = COCOeval(anno, pred, 'bbox')
if is_coco:
eval.params.imgIds = [
int(Path(x).stem) for x in dataloader.dataset.img_files
] # image IDs to evaluate
eval.evaluate()
eval.accumulate()
eval.summarize()
map, map50 = eval.stats[:
2] # update results ([email protected]:0.95, [email protected])
except Exception as e:
print(f'pycocotools unable to run: {e}')
# Return results
if not training:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}")
model.float() # for training
maps = np.zeros(nc) + map
for i, c in enumerate(ap_class):
maps[c] = ap[i]
return (mp, mr, map50, map,
*(loss.cpu() / len(dataloader)).tolist()), maps, t
def detect_images(self):
with torch.no_grad():
# Get names and colors
names = self.model.module.names if hasattr(
self.model, 'module') else self.model.names
colors = [[random.randint(0, 255) for _ in range(3)]
for _ in range(len(names))]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, self.imgsz, self.imgsz),
device=self.device) # init img
o = self.model(img.half() if self.half else img
) if self.device.type != 'cpu' else None # run once
for path, img, im0s, vid_cap in self.dataset:
img = torch.from_numpy(img).to(self.device)
img = img.half() if self.half else img.float(
) # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = self.model(img, augment=self.augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
self.conf_thres,
self.iou_thres,
classes=self.classes)
t2 = time_synchronized()
# Process detections
for i, det in enumerate(pred): # detections per image
p, s, im0 = path, '', im0s
save_path = str(Path('inference/output/') / Path(p).name)
txt_path = str(
Path('inference/output/') /
Path(p).stem) + ('_%g' % self.dataset.frame if
self.dataset.mode == 'video' else '')
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1, 0
]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]
) # add to string
# for *xyxy, conf, cls in reversed(det):
# if self.save_img: # Add bbox to image
# label = '%s %.2f' % (names[int(cls)], conf)
# plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=3)
# Print time (inference + NMS)
print('%sDone. (%.3fFPS)' % (s, 1 / (t2 - t1)))
def detect_image(self, im0s):
names = self.model.module.names if hasattr(
self.model, 'module') else self.model.names
colors = [[random.randint(0, 255) for _ in range(3)]
for _ in range(len(names))]
img = self.preprocess_image(im0s)
print(img.shape)
img = torch.from_numpy(img).to(self.device)
img = img.half() if self.half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = self.model(img, augment=self.augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
self.conf_thres,
self.iou_thres,
classes=self.classes)
# Process detections
boxes = []
confs = []
labels = []
for i, det in enumerate(pred): # detections per image
s, im0 = '', im0s
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]) # add to string
for *xyxy, conf, cls in reversed(det):
bbox = [
int(xyxy[0]),
int(xyxy[1]),
int(xyxy[2]),
int(xyxy[3])
]
boxes.append(bbox)
confs.append(float(conf))
labels.append(int(cls))
t2 = time_synchronized()
# Print time (inference + NMS)
print('%sDone. (%.3f FPS)' % (s, 1 / (t2 - t1)))
return boxes, confs, labels
def custom_detect(
weights='yolov5l.pt', # model.pt path(s)
source='data/images', # source folder
imgsz=640, # inference size (pixels)
conf_thres=0.5, # object confidence threshold
iou_thres=0.45, # IOU threshold for NMS
device='', # cuda device, i.e. 0 or 0,1,2,3 or cpu
view_img=False, # display results
box_only=False, # output only contains bounding boxes, not original image
save_txt=False, # save results to *.txt
save_conf=False, # save confidences in --save-txt labels
classes: int=None, # only look for certain classes. Probably a list of ints
agnostic_nms=False, # class-agnostic NMS
augment=False, # augmented inference
update=False, # update all models
project='runs', # save results to project/name
name='', # save results to project/name
exist_ok=False # existing project/name ok, do not increment
):
# Directories
save_dir = Path(increment_path(Path(project) / name, exist_ok=exist_ok)) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
# Initialize
set_logging()
device = select_device(device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
stride = int(model.stride.max()) # model stride
imgsz = check_img_size(imgsz, s=stride) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(torch.load('weights/resnet101.pt', map_location=device)['model']).to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
save_img = True
dataset = LoadImages(source, img_size=imgsz, stride=stride)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
# Run inference
if device.type != 'cpu':
model(torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(next(model.parameters()))) # run once
t0 = time.time()
for path, img, im0, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=augment)[0]
# Apply NMS
pred = non_max_suppression(pred, conf_thres, iou_thres, classes=classes, agnostic=agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0)
# Process detections
for i, det in enumerate(pred): # detections per image
p, s, frame = path, '', getattr(dataset, 'frame', 0)
p = Path(p) # to Path
save_path = str(save_dir / p.name) # img.jpg
txt_path = str(save_dir / 'labels' / p.stem) + ('' if dataset.mode == 'image' else f'_{frame}') # img.txt
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
if box_only:
im0 = np.zeros_like(im0) # Makes to img black to output only bounding boxes
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh, conf) if save_conf else (cls, *xywh) # label format
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * len(line)).rstrip() % line + '\n')
if save_img or view_img: # Add bbox to image
label = f'{names[int(cls)]} {conf:.2f}'
plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=3)
# Print time (inference + NMS)
print(f'{s}Done. ({t2 - t1:.3f}s)')
# Stream results
if view_img:
cv2.imshow(str(p), im0)
cv2.waitKey(1) # 1 millisecond
# Save results (image with detections)
if save_img:
if dataset.mode == 'image':
cv2.imwrite(save_path, im0)
else: # 'video'
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release() # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
vid_writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*fourcc), fps, (w, h))
vid_writer.write(im0)
if save_txt or save_img:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}")
print(f'Done. ({time.time() - t0:.3f}s)')
def detect(source,
weights,
conf_thres=0.25,
imgsz=640,
iou_thres=0.45,
classes=None,
device=''):
# Initialize
set_logging()
device = select_device(device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(
torch.load('weights/resnet101.pt',
map_location=device)['model']) # load weights
modelc.to(device).eval()
# Set Dataloader
dataset = LoadImages(source, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)]
for _ in range(len(names))]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img
) if device.type != 'cpu' else None # run once
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=True)[0]
# Apply NMS
pred = non_max_suppression(pred,
conf_thres,
iou_thres,
classes=classes,
agnostic=True)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
p, s, im0 = path, '', im0s
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]) # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
label = '%s %.2f' % (names[int(cls)], conf)
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3)
# Print time (inference + NMS)
print('%sDone. (%.3fs)' % (s, t2 - t1))
print('Done. (%.3fs)' % (time.time() - t0))
return im0
def test(
data,
weights=None,
batch_size=16,
imgsz=640,
conf_thres=0.001,
iou_thres=0.6, # for NMS
save_json=False,
single_cls=False,
augment=False,
verbose=False,
model=None,
dataloader=None,
save_dir='',
merge=False,
save_txt=False):
# Initialize/load model and set device
training = model is not None
if training: # called by train.py
device = next(model.parameters()).device # get model device
else: # called directly
device = select_device(opt.device, batch_size=batch_size)
merge, save_txt = opt.merge, opt.save_txt # use Merge NMS, save *.txt labels
if save_txt:
out = Path('inference/output')
if os.path.exists(out):
shutil.rmtree(out) # delete output folder
os.makedirs(out) # make new output folder
# Remove previous
for f in glob.glob(str(Path(save_dir) / 'test_batch*.jpg')):
os.remove(f)
# Load model
model = Darknet(opt.cfg).to(device)
# load model
try:
ckpt = torch.load(weights[0],
map_location=device) # load checkpoint
ckpt['model'] = {
k: v
for k, v in ckpt['model'].items()
if model.state_dict()[k].numel() == v.numel()
}
model.load_state_dict(ckpt['model'], strict=False)
except:
load_darknet_weights(model, weights[0])
imgsz = check_img_size(imgsz, s=32) # check img_size
# Half
half = device.type != 'cpu' # half precision only supported on CUDA
if half:
model.half()
# Configure
model.eval()
with open(data) as f:
data = yaml.load(f, Loader=yaml.FullLoader) # model dict
nc = 1 if single_cls else int(data['nc']) # number of classes
iouv = torch.linspace(0.5, 0.95,
10).to(device) # iou vector for [email protected]:0.95
niou = iouv.numel()
# Dataloader
if not training:
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img
) if device.type != 'cpu' else None # run once
path = data['test'] if opt.task == 'test' else data[
'val'] # path to val/test images
dataloader = create_dataloader(path,
imgsz,
batch_size,
32,
opt,
hyp=None,
augment=False,
cache=False,
pad=0.5,
rect=True)[0]
seen = 0
try:
names = model.names if hasattr(model, 'names') else model.module.names
except:
names = load_classes(opt.names)
coco91class = coco80_to_coco91_class()
s = ('%20s' + '%12s' * 6) % ('Class', 'Images', 'Targets', 'P', 'R',
'[email protected]', '[email protected]:.95')
p, r, f1, mp, mr, map50, map, t0, t1 = 0., 0., 0., 0., 0., 0., 0., 0., 0.
loss = torch.zeros(3, device=device)
jdict, stats, ap, ap_class = [], [], [], []
for batch_i, (img, targets, paths,
shapes) in enumerate(tqdm(dataloader, desc=s)):
img = img.to(device, non_blocking=True)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
targets = targets.to(device)
nb, _, height, width = img.shape # batch size, channels, height, width
whwh = torch.Tensor([width, height, width, height]).to(device)
# Disable gradients
with torch.no_grad():
# Run model
t = time_synchronized()
inf_out, train_out = model(
img, augment=augment) # inference and training outputs
t0 += time_synchronized() - t
# Compute loss
if training: # if model has loss hyperparameters
loss += compute_loss([x.float() for x in train_out], targets,
model)[1][:3] # GIoU, obj, cls
# Run NMS
t = time_synchronized()
output = non_max_suppression(inf_out,
conf_thres=conf_thres,
iou_thres=iou_thres,
merge=merge)
t1 += time_synchronized() - t
# Statistics per image
for si, pred in enumerate(output):
labels = targets[targets[:, 0] == si, 1:]
nl = len(labels)
tcls = labels[:, 0].tolist() if nl else [] # target class
seen += 1
if pred is None:
if nl:
stats.append((torch.zeros(0, niou, dtype=torch.bool),
torch.Tensor(), torch.Tensor(), tcls))
continue
# Append to text file
if save_txt:
gn = torch.tensor(shapes[si][0])[[1, 0, 1, 0
]] # normalization gain whwh
txt_path = str(out / Path(paths[si]).stem)
pred[:, :4] = scale_coords(img[si].shape[1:], pred[:, :4],
shapes[si][0],
shapes[si][1]) # to original
for *xyxy, conf, cls in pred:
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
with open(txt_path + '.txt', 'a') as f:
f.write(
('%g ' * 5 + '\n') % (cls, *xywh)) # label format
# Clip boxes to image bounds
clip_coords(pred, (height, width))
# Append to pycocotools JSON dictionary
if save_json:
# [{"image_id": 42, "category_id": 18, "bbox": [258.15, 41.29, 348.26, 243.78], "score": 0.236}, ...
image_id = Path(paths[si]).stem
box = pred[:, :4].clone() # xyxy
scale_coords(img[si].shape[1:], box, shapes[si][0],
shapes[si][1]) # to original shape
box = xyxy2xywh(box) # xywh
box[:, :2] -= box[:, 2:] / 2 # xy center to top-left corner
for p, b in zip(pred.tolist(), box.tolist()):
jdict.append({
'image_id':
int(image_id) if image_id.isnumeric() else image_id,
'category_id':
coco91class[int(p[5])],
'bbox': [round(x, 3) for x in b],
'score':
round(p[4], 5)
})
# Assign all predictions as incorrect
correct = torch.zeros(pred.shape[0],
niou,
dtype=torch.bool,
device=device)
if nl:
detected = [] # target indices
tcls_tensor = labels[:, 0]
# target boxes
tbox = xywh2xyxy(labels[:, 1:5]) * whwh
# Per target class
for cls in torch.unique(tcls_tensor):
ti = (cls == tcls_tensor).nonzero(as_tuple=False).view(
-1) # prediction indices
pi = (cls == pred[:, 5]).nonzero(as_tuple=False).view(
-1) # target indices
# Search for detections
if pi.shape[0]:
# Prediction to target ious
ious, i = box_iou(pred[pi, :4], tbox[ti]).max(
1) # best ious, indices
# Append detections
for j in (ious > iouv[0]).nonzero(as_tuple=False):
d = ti[i[j]] # detected target
if d not in detected:
detected.append(d)
correct[
pi[j]] = ious[j] > iouv # iou_thres is 1xn
if len(
detected
) == nl: # all targets already located in image
break
# Append statistics (correct, conf, pcls, tcls)
stats.append(
(correct.cpu(), pred[:, 4].cpu(), pred[:, 5].cpu(), tcls))
# Plot images
if batch_i < 1:
f = Path(save_dir) / ('test_batch%g_gt.jpg' % batch_i) # filename
plot_images(img, targets, paths, str(f), names) # ground truth
f = Path(save_dir) / ('test_batch%g_pred.jpg' % batch_i)
plot_images(img, output_to_target(output, width, height), paths,
str(f), names) # predictions
# Compute statistics
stats = [np.concatenate(x, 0) for x in zip(*stats)] # to numpy
if len(stats) and stats[0].any():
p, r, ap, f1, ap_class = ap_per_class(*stats)
p, r, ap50, ap = p[:, 0], r[:, 0], ap[:, 0], ap.mean(
1) # [P, R, [email protected], [email protected]:0.95]
mp, mr, map50, map = p.mean(), r.mean(), ap50.mean(), ap.mean()
nt = np.bincount(stats[3].astype(np.int64),
minlength=nc) # number of targets per class
else:
nt = torch.zeros(1)
# Print results
pf = '%20s' + '%12.3g' * 6 # print format
print(pf % ('all', seen, nt.sum(), mp, mr, map50, map))
# Print results per class
if verbose and nc > 1 and len(stats):
for i, c in enumerate(ap_class):
print(pf % (names[c], seen, nt[c], p[i], r[i], ap50[i], ap[i]))
# Print speeds
t = tuple(x / seen * 1E3
for x in (t0, t1, t0 + t1)) + (imgsz, imgsz, batch_size) # tuple
if not training:
print(
'Speed: %.1f/%.1f/%.1f ms inference/NMS/total per %gx%g image at batch-size %g'
% t)
# Save JSON
if save_json and len(jdict):
f = 'detections_val2017_%s_results.json' % \
(weights.split(os.sep)[-1].replace('.pt', '') if isinstance(weights, str) else '') # filename
print('\nCOCO mAP with pycocotools... saving %s...' % f)
with open(f, 'w') as file:
json.dump(jdict, file)
try: # https://github.com/cocodataset/cocoapi/blob/master/PythonAPI/pycocoEvalDemo.ipynb
from pycocotools.coco import COCO
from pycocotools.cocoeval import COCOeval
imgIds = [int(Path(x).stem) for x in dataloader.dataset.img_files]
cocoGt = COCO(
glob.glob('../coco/annotations/instances_val*.json')
[0]) # initialize COCO ground truth api
cocoDt = cocoGt.loadRes(f) # initialize COCO pred api
cocoEval = COCOeval(cocoGt, cocoDt, 'bbox')
cocoEval.params.imgIds = imgIds # image IDs to evaluate
cocoEval.evaluate()
cocoEval.accumulate()
cocoEval.summarize()
map, map50 = cocoEval.stats[:
2] # update results ([email protected]:0.95, [email protected])
except Exception as e:
print('ERROR: pycocotools unable to run: %s' % e)
# Return results
model.float() # for training
maps = np.zeros(nc) + map
for i, c in enumerate(ap_class):
maps[c] = ap[i]
return (mp, mr, map50, map,
*(loss.cpu() / len(dataloader)).tolist()), maps, t
def forward(self, imgs, size=640, augment=False, profile=False):
# Inference from various sources. For height=640, width=1280, RGB images example inputs are:
# file: imgs = 'data/images/zidane.jpg' # str or PosixPath
# URI: = 'https://ultralytics.com/images/zidane.jpg'
# OpenCV: = cv2.imread('image.jpg')[:,:,::-1] # HWC BGR to RGB x(640,1280,3)
# PIL: = Image.open('image.jpg') or ImageGrab.grab() # HWC x(640,1280,3)
# numpy: = np.zeros((640,1280,3)) # HWC
# torch: = torch.zeros(16,3,320,640) # BCHW (scaled to size=640, 0-1 values)
# multiple: = [Image.open('image1.jpg'), Image.open('image2.jpg'), ...] # list of images
t = [time_sync()]
p = next(self.model.parameters()) # for device and type
if isinstance(imgs, torch.Tensor): # torch
with amp.autocast(enabled=p.device.type != 'cpu'):
return self.model(
imgs.to(p.device).type_as(p), augment,
profile) # inference
# Pre-process
n, imgs = (len(imgs), imgs) if isinstance(imgs, list) else (
1, [imgs]) # number of images, list of images
shape0, shape1, files = [], [], [
] # image and inference shapes, filenames
for i, im in enumerate(imgs):
f = f'image{i}' # filename
if isinstance(im, (str, Path)): # filename or uri
im, f = Image.open(
requests.get(im, stream=True).raw if str(im).
startswith('http') else im), im
im = np.asarray(exif_transpose(im))
elif isinstance(im, Image.Image): # PIL Image
im, f = np.asarray(
exif_transpose(im)), getattr(im, 'filename', f) or f
files.append(Path(f).with_suffix('.jpg').name)
if im.shape[0] < 5: # image in CHW
im = im.transpose(
(1, 2, 0)) # reverse dataloader .transpose(2, 0, 1)
im = im[..., :3] if im.ndim == 3 else np.tile(
im[..., None], 3) # enforce 3ch input
s = im.shape[:2] # HWC
shape0.append(s) # image shape
g = (size / max(s)) # gain
shape1.append([y * g for y in s])
imgs[i] = im if im.data.contiguous else np.ascontiguousarray(
im) # update
shape1 = [
make_divisible(x, int(self.stride.max()))
for x in np.stack(shape1, 0).max(0)
] # inference shape
x = [letterbox(im, new_shape=shape1, auto=False)[0]
for im in imgs] # pad
x = np.stack(x, 0) if n > 1 else x[0][None] # stack
x = np.ascontiguousarray(x.transpose((0, 3, 1, 2))) # BHWC to BCHW
x = torch.from_numpy(x).to(
p.device).type_as(p) / 255 # uint8 to fp16/32
t.append(time_sync())
with amp.autocast(enabled=p.device.type != 'cpu'):
# Inference
y = self.model(x, augment, profile)[0] # forward
t.append(time_sync())
# Post-process
y = non_max_suppression(y,
self.conf,
iou_thres=self.iou,
classes=self.classes,
multi_label=self.multi_label,
max_det=self.max_det) # NMS
for i in range(n):
scale_coords(shape1, y[i][:, :4], shape0[i])
t.append(time_sync())
return Detections(imgs, y, files, t, self.names, x.shape)
def test(
data,
weights=None,
batch_size=32,
imgsz=640,
conf_thres=0.001,
iou_thres=0.6, # for NMS
save_json=False,
single_cls=False,
augment=False,
verbose=False,
model=None,
dataloader=None,
save_dir=Path(''), # for saving images
save_txt=False, # for auto-labelling
save_hybrid=False, # for hybrid auto-labelling
save_conf=False, # save auto-label confidences
plots=False,
log_imgs=0, # number of logged images
compute_loss=None):
# Initialize/load model and set device
logger = setup_logger('Test', './')
write_info(logger, True)
training = model is not None
if training: # called by train.py
device = next(model.parameters()).device # get model device
else: # called directly
set_logging()
device = select_device(opt.device, batch_size=batch_size)
# Directories
save_dir = Path(
increment_path(Path(opt.project) / opt.name,
exist_ok=opt.exist_ok)) # increment run
# (save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
# Half
half = device.type != 'cpu' # half precision only supported on CUDA
if half:
model.half()
# Configure
model.eval()
with open(data) as f:
data = yaml.load(f, Loader=yaml.FullLoader) # model dict
check_dataset(data) # check
nc = 1 if single_cls else int(data['nc']) # number of classes
iouv = torch.linspace(0.5, 0.95,
10).to(device) # iou vector for [email protected]:0.95
niou = iouv.numel()
# Logging
log_imgs, wandb = min(log_imgs, 100), None # ceil
try:
import wandb # Weights & Biases
except ImportError:
log_imgs = 0
# Dataloader
if not training:
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img
) if device.type != 'cpu' else None # run once
path = data['test'] if opt.task == 'test' else data[
'val'] # path to val/test images
dataloader = create_dataloader(
path,
imgsz,
batch_size,
model.stride.max(),
opt,
pad=0.5,
rect=True,
prefix=colorstr('test: ' if opt.task == 'test' else 'val: '))[0]
# write_imglist(path, logger)
seen = 0
confusion_matrix = ConfusionMatrix(nc=nc)
names = {
k: v
for k, v in enumerate(
model.names if hasattr(model, 'names') else model.module.names)
}
p, r, f1, mp, mr, map50, map, t0, t1 = 0., 0., 0., 0., 0., 0., 0., 0., 0.
loss = torch.zeros(3, device=device)
jdict, stats, ap, ap_class, wandb_images = [], [], [], [], []
for batch_i, (img, targets, paths, shapes) in enumerate(dataloader):
stats_perimg = []
img = img.to(device, non_blocking=True)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
targets = targets.to(device)
nb, _, height, width = img.shape # batch size, channels, height, width
with torch.no_grad():
# Run model
t = time_synchronized()
inf_out, train_out = model(
img, augment=augment) # inference and training outputs
t0 += time_synchronized() - t
# Compute loss
if compute_loss:
loss += compute_loss([x.float() for x in train_out],
targets)[1][:3] # box, obj, cls
# Run NMS
targets[:, 2:] *= torch.Tensor([width, height, width,
height]).to(device) # to pixels
lb = [targets[targets[:, 0] == i, 1:] for i in range(nb)
] if save_hybrid else [] # for autolabelling
t = time_synchronized()
output = non_max_suppression(inf_out,
conf_thres=conf_thres,
iou_thres=iou_thres,
labels=lb)
t1 += time_synchronized() - t
# Statistics per image
for si, pred in enumerate(output):
labels = targets[targets[:, 0] == si, 1:]
nl = len(labels)
tcls = labels[:, 0].tolist() if nl else [] # target class
path = Path(paths[si])
seen += 1
if len(pred) == 0:
if nl:
stats.append((torch.zeros(0, niou, dtype=torch.bool),
torch.Tensor(), torch.Tensor(), tcls))
stats_perimg.append((torch.zeros(0, niou,
dtype=torch.bool),
torch.Tensor(), torch.Tensor(), tcls))
continue
# Predictions
predn = pred.clone()
scale_coords(img[si].shape[1:], predn[:, :4], shapes[si][0],
shapes[si][1]) # native-space pred
# Assign all predictions as incorrect
correct = torch.zeros(pred.shape[0],
niou,
dtype=torch.bool,
device=device)
if nl:
detected = [] # target indices
tcls_tensor = labels[:, 0]
# target boxes
tbox = xywh2xyxy(labels[:, 1:5])
scale_coords(img[si].shape[1:], tbox, shapes[si][0],
shapes[si][1]) # native-space labels
if plots:
confusion_matrix.process_batch(
pred, torch.cat((labels[:, 0:1], tbox), 1))
# Per target class
for cls in torch.unique(tcls_tensor):
ti = (cls == tcls_tensor).nonzero(as_tuple=False).view(
-1) # prediction indices
pi = (cls == pred[:, 5]).nonzero(as_tuple=False).view(
-1) # target indices
# Search for detections
if pi.shape[0]:
# Prediction to target ious
ious, i = box_iou(predn[pi, :4], tbox[ti]).max(
1) # best ious, indices
# Append detections
detected_set = set()
for j in (ious > iouv[0]).nonzero(as_tuple=False):
d = ti[i[j]] # detected target
if d.item() not in detected_set:
detected_set.add(d.item())
detected.append(d)
correct[
pi[j]] = ious[j] > iouv # iou_thres is 1xn
if len(
detected
) == nl: # all targets already located in image
break
# Append statistics (correct, conf, pcls, tcls)
stats.append(
(correct.cpu(), pred[:, 4].cpu(), pred[:, 5].cpu(), tcls))
stats_perimg.append(
(correct.cpu(), pred[:, 4].cpu(), pred[:, 5].cpu(), tcls))
# # f1 per image
# mf1_perimg = 0
# mp_perimg = 0
# mr_perimg = 0
# stats_perimg = [np.concatenate(x, 0) for x in zip(*stats_perimg)]
# if len(stats_perimg) and stats_perimg[0].any():
# p_perimg, r_perimg, _, f1_perimg, ap_class_perimg = ap_per_class(*stats_perimg, plot=plots, save_dir=save_dir, names=names)
# p_perimg, r_perimg, f1_perimg = p_perimg[:, 0], r_perimg[:, 0], f1_perimg[:, 0] # [P, R, [email protected], [email protected]:0.95]
# nt_perimg = np.bincount(stats_perimg[3].astype(np.int64), minlength=nc) # number of targets per class
# for ind, c in enumerate(ap_class_perimg):
# mf1_perimg += f1_perimg[ind] * nt_perimg[c]
# mp_perimg += p_perimg[ind] * nt_perimg[c]
# mr_perimg += r_perimg[ind] * nt_perimg[c]
# mf1_perimg = mf1_perimg/nt_perimg.sum()
# mp_perimg = mp_perimg / nt_perimg.sum()
# mr_perimg = mr_perimg / nt_perimg.sum()
#
# logger.info('[{}] {} [F1 score:{:4f} (Prec: {:4f}, Rec: {:4f})]'.format(str(batch_i + 1), paths[0].split('/')[-1], mf1_perimg, mp_perimg, mr_perimg))
# save GPS log
# Logger_System('./xmls', './Logger', output, paths, names)
# Compute statistics
stats = [np.concatenate(x, 0) for x in zip(*stats)] # to numpy
if len(stats) and stats[0].any():
p, r, ap, f1, ap_class = ap_per_class(*stats,
plot=plots,
save_dir=save_dir,
names=names)
p, r, ap50, ap, f1 = p[:, 0], r[:, 0], ap[:, 0], ap.mean(
1), f1[:, 0] # [P, R, [email protected], [email protected]:0.95]
mp, mr, map50, map, mf1 = p.mean(), r.mean(), ap50.mean(), ap.mean(
), f1.mean()
nt = np.bincount(stats[3].astype(np.int64),
minlength=nc) # number of targets per class
else:
nt = torch.zeros(1)
meanf1 = 0
meanp = 0
meanr = 0
meanap = 0
# Print results per class
if (verbose or (nc <= 20)) and nc > 1 and len(stats):
for i, c in enumerate(ap_class):
meanf1 += f1[i] * nt[c]
meanp += p[i] * nt[c]
meanr += r[i] * nt[c]
meanap += ap50[i] * nt[c]
meanf1 = meanf1 / nt.sum()
meanp = meanp / nt.sum()
meanr = meanr / nt.sum()
meanap = meanap / nt.sum()
logger.info('[Final] F1 score:{:4f} (Prec: {:4f}, Rec: {:4f})\n'.format(
meanf1, meanp, meanr))
# Print speeds
t = tuple(x / seen * 1E3
for x in (t0, t1, t0 + t1)) + (imgsz, imgsz, batch_size) # tuple
if not training:
print(
'Speed: %.1f/%.1f/%.1f ms inference/NMS/total per %gx%g image at batch-size %g'
% t)
# Return results
model.float() # for training
maps = np.zeros(nc) + map
for i, c in enumerate(ap_class):
maps[c] = ap[i]
write_info(logger, False)
return (meanf1, mp, mr, map50, meanap,
*(loss.cpu() / len(dataloader)).tolist()), maps, t
def detect():
source, weights, save_txt, imgsz, save_img = config.source, config.weights, config.save_txt, config.img_size, config.save_img
#对图片进行重命名
rename_file()
# Directories
save_dir = Path(increment_path(Path(config.project) / config.name, exist_ok=config.exist_ok)) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
# Initialize
set_logging()
device = select_device(config.device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Set Dataloader
vid_path, vid_writer = None, None
dataset = LoadImages(source, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img) if device.type != 'cpu' else None # run once
#载入标签文件，用于将切割好的图片进行归档
label_list = []
with open(config.label_list_path, 'r') as f:
temp_list = f.readlines()
for i in temp_list:
temp = i.strip().split(',',maxsplit=3)
label_list.append(temp)
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=True)[0]
# Apply NMS
pred = non_max_suppression(pred, config.conf_thres, config.iou_thres)
t2 = time_synchronized()
# Process detections
for i, det in enumerate(pred): # detections per image
p, s, im0, frame = Path(path), '', im0s, getattr(dataset, 'frame', 0)#im0是原图[长，宽，通道]
save_path = str(save_dir / p.name)
txt_path = str(save_dir / 'labels' / p.stem) + ('' if dataset.mode == 'image' else f'_{frame}')
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
if len(det):
#存储经转换后的标记框信息
det_info_list = []
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f'{n} {names[int(c)]}s, ' # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist() # normalized xywh
line = [int(cls), xywh, float(conf.cpu())] if config.save_conf else [int(cls), xywh] # label format
det_info_list.append(line)
if save_img: # Add bbox to image
label = f'{names[int(cls)]} {conf:.2f}'
plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=3)
#处理图片信息
cut_img(p.name, det_info_list, im0.shape[1], im0.shape[0], save_dir, label_list)
else:
#此时表明没有框选出图片，需要进行记录：
log_result(p.name + '不存在框选目标')
# Print time (inference + NMS)
print(f'{s}Done. ({t2 - t1:.3f}s)')
# 存储识别结果
if save_img:
cv2.imwrite(save_path, im0)
if save_txt or save_img:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}")
print(f'Done. ({time.time() - t0:.3f}s)')
def run(weights='yolov5s.pt', # model.pt path(s)
source='data/images', # file/dir/URL/glob, 0 for webcam
imgsz=640, # inference size (pixels)
conf_thres=0.25, # confidence threshold
iou_thres=0.45, # NMS IOU threshold
max_det=1000, # maximum detections per image
device='', # cuda device, i.e. 0 or 0,1,2,3 or cpu
view_img=False, # show results
save_txt=False, # save results to *.txt
save_conf=False, # save confidences in --save-txt labels
save_crop=False, # save cropped prediction boxes
nosave=False, # do not save images/videos
classes=None, # filter by class: --class 0, or --class 0 2 3
agnostic_nms=False, # class-agnostic NMS
augment=False, # augmented inference
update=False, # update all models
project='runs/detect', # save results to project/name
name='exp', # save results to project/name
exist_ok=False, # existing project/name ok, do not increment
line_thickness=3, # bounding box thickness (pixels)
hide_labels=False, # hide labels
hide_conf=False, # hide confidences
half=False, # use FP16 half-precision inference
):
save_img = not nosave and not source.endswith('.txt') # save inference images
webcam = source.isnumeric() or source.endswith('.txt') or source.lower().startswith(
('rtsp://', 'rtmp://', 'http://', 'https://'))
# Directories
save_dir = increment_path(Path(project) / name, exist_ok=exist_ok) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
# Initialize
set_logging()
device = select_device(device)
half &= device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
stride = int(model.stride.max()) # model stride
imgsz = check_img_size(imgsz, s=stride) # check image size
names = model.module.names if hasattr(model, 'module') else model.names # get class names
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet50', n=2) # initialize
modelc.load_state_dict(torch.load('resnet50.pt', map_location=device)['model']).to(device).eval()
# Dataloader
if webcam:
view_img = check_imshow()
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadStreams(source, img_size=imgsz, stride=stride)
bs = len(dataset) # batch_size
else:
dataset = LoadImages(source, img_size=imgsz, stride=stride)
bs = 1 # batch_size
vid_path, vid_writer = [None] * bs, [None] * bs
# Run inference
if device.type != 'cpu':
model(torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(next(model.parameters()))) # run once
t0 = time.time()
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=augment)[0]
# Apply NMS
pred = non_max_suppression(pred, conf_thres, iou_thres, classes, agnostic_nms, max_det=max_det)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0, frame = path[i], f'{i}: ', im0s[i].copy(), dataset.count
else:
p, s, im0, frame = path, '', im0s.copy(), getattr(dataset, 'frame', 0)
p = Path(p) # to Path
save_path = str(save_dir / p.name) # img.jpg
txt_path = str(save_dir / 'labels' / p.stem) + ('' if dataset.mode == 'image' else f'_{frame}') # img.txt
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
imc = im0.copy() if save_crop else im0 # for save_crop
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh, conf) if save_conf else (cls, *xywh) # label format
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * len(line)).rstrip() % line + '\n')
if save_img or save_crop or view_img: # Add bbox to image
c = int(cls) # integer class
label = None if hide_labels else (names[c] if hide_conf else f'{names[c]} {conf:.2f}')
plot_one_box(xyxy, im0, label=label, color=colors(c, True), line_thickness=line_thickness)
if save_crop:
save_one_box(xyxy, imc, file=save_dir / 'crops' / names[c] / f'{p.stem}.jpg', BGR=True)
# Print time (inference + NMS)
print(f'{s}Done. ({t2 - t1:.3f}s)')
# Stream results
if view_img:
cv2.imshow(str(p), im0)
cv2.waitKey(1) # 1 millisecond
# Save results (image with detections)
if save_img:
if dataset.mode == 'image':
cv2.imwrite(save_path, im0)
else: # 'video' or 'stream'
if vid_path[i] != save_path: # new video
vid_path[i] = save_path
if isinstance(vid_writer[i], cv2.VideoWriter):
vid_writer[i].release() # release previous video writer
if vid_cap: # video
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
else: # stream
fps, w, h = 30, im0.shape[1], im0.shape[0]
save_path += '.mp4'
vid_writer[i] = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*'mp4v'), fps, (w, h))
vid_writer[i].write(im0)
if save_txt or save_img:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}")
if update:
strip_optimizer(weights) # update model (to fix SourceChangeWarning)
print(f'Done. ({time.time() - t0:.3f}s)')
def detect(save_img=False):
# saving dhakaai submission format
if opt.dhakaai_format:
results = {
'image_id': [],
'score': [],
'class': [],
'xyxy':[],
'hw':[],
}
source, weights, view_img, save_txt, imgsz = opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size
webcam = source.isnumeric() or source.endswith('.txt') or source.lower().startswith(
('rtsp://', 'rtmp://', 'http://'))
# Directories
save_dir = Path(increment_path(Path(opt.project) / opt.name, exist_ok=opt.exist_ok)) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
# Initialize
set_logging()
device = select_device(opt.device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(torch.load('weights/resnet101.pt', map_location=device)['model']).to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
if webcam:
view_img = True
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadStreams(source, img_size=imgsz)
else:
save_img = True
dataset = LoadImages(source, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img) if device.type != 'cpu' else None # run once
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred, opt.conf_thres, opt.iou_thres, classes=opt.classes, agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0 = Path(path[i]), '%g: ' % i, im0s[i].copy()
else:
p, s, im0 = Path(path), '', im0s
save_path = str(save_dir / p.name)
txt_path = str(save_dir / 'labels' / p.stem) + ('_%g' % dataset.frame if dataset.mode == 'video' else '')
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]) # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
if opt.dhakaai_format:
conf_score = '%.2f' % (conf)
label_with_cls = '%s' % (names[int(cls)])
results['image_id'].append(Path(p).name)
results['score'].append(conf_score)
results['class'].append(label_with_cls)
results['xyxy'].append(xyxy)
results['hw'].append((im0.shape[0], im0.shape[1]))
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh, conf) if opt.save_conf else (cls, *xywh) # label format
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * len(line)).rstrip() % line + '\n')
if save_img or view_img: # Add bbox to image
label = '%s %.2f' % (names[int(cls)], conf)
plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=3)
# Print time (inference + NMS)
print('%sDone. (%.3fs)' % (s, t2 - t1))
# Stream results
if view_img:
cv2.imshow(str(p), im0)
if cv2.waitKey(1) == ord('q'): # q to quit
raise StopIteration
# Save results (image with detections)
if save_img:
if dataset.mode == 'images':
cv2.imwrite(save_path, im0)
else:
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release() # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
vid_writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*fourcc), fps, (w, h))
vid_writer.write(im0)
if save_txt or save_img:
print('Results saved to %s' % save_dir)
print('Done. (%.3fs)' % (time.time() - t0))
if opt.dhakaai_format:
submit_dhakaAI(results, 'submission')
def detect():
device = select_device(str(0))
print('device:', device)
#half = device.type != 'cpu'
# Load model
weights = 'runs/exp0/weights/best.pt'
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(512, s=model.stride.max()) # check img_size
save_img = True
##dataset = LoadImages(source, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)]
for _ in range(len(names))]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img) #if device.type != 'cpu' else None # run once
img0 = cv2.imread('test/0013_0.jpg')
#img0 = Image.open('test/0013_0.jpg')
#print(img0)
img = letterbox(img0, new_shape=512)[0]
# Convert
img = img[:, :, ::-1].transpose(2, 0, 1) # BGR to RGB, to 3x416x416
img = np.ascontiguousarray(img)
img = torch.from_numpy(img).to(device)
img = img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=False)[0]
# Apply NMS
pred = non_max_suppression(pred, 0.4, 0.5, classes=0, agnostic=False)
t2 = time_synchronized()
# Process detections
for i, det in enumerate(pred):
gn = torch.tensor(img0.shape)[[1, 0, 1, 0]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
img0.shape).round()
# Write results
n = 1
for *xyxy, conf, cls in reversed(det):
# Add bbox to image
label = '%s %.2f' % (names[int(cls)], conf)
x1, y1, x2, y2 = int(xyxy[0]), int(xyxy[1]), int(xyxy[2]), int(
xyxy[3])
print(x1, y1, x2, y2)
crop = img0[y1:y2, x1:x2]
cv2.imwrite('test/{}.jpg'.format(str(n)), crop)
n = n + 1
plot_one_box(xyxy,
img0,
label=label,
color=colors[int(cls)],
line_thickness=3)
if save_img:
cv2.imwrite('test/test.jpg', img0)
print('Done. (%.3fs)' % (time.time() - t0))
def detect(save_img=False):
out, source, weights, view_img, save_txt, imgsz = opt.output, opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size
webcam = source.isnumeric() or source.endswith(
'.txt') or source.lower().startswith(('rtsp://', 'rtmp://', 'http://'))
#
# # Directories
# save_dir = Path(increment_path(Path(opt.project) / opt.name, exist_ok=opt.exist_ok)) # increment run
# (save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
# Initialize
set_logging()
device = select_device(opt.device)
if os.path.exists(out):
shutil.rmtree(out) # delete output folder
os.makedirs(out) # make new output folder
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(
torch.load('weights/resnet101.pt',
map_location=device)['model']).to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
# if webcam:
# view_img = True
# cudnn.benchmark = True # set True to speed up constant image size inference
# dataset = LoadStreams(source, img_size=imgsz)
# else:
# save_img = True
# dataset = LoadImages(source, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img
) if device.type != 'cpu' else None # run once
videopath_list = (
# '2020double_company',
# '2020double_company_1',
# 'child_79_company',
# 'child_137_huaxia',
# 'child_137_huaxia_1',
# 'double_54_zhuhai',
# 'double_54_zhuhai_1',
# 'double_59_huaxiaxueyuan',
# 'double_59_huaxiaxueyuan_1',
# 'double_990_close_company',
# 'double_beijing',
# 'double_beijing_1',
'single_28_huaxia',
# 'single_28_huaxia_2',
# 'single_897_yinchuan',
# 'single_897_yinchuan_2',
# 'single_1000_beijng_shoudu',
# 'single_1000_guangzhjou',
# 'single_1000_wuhan',
)
video_dir_pass = [
'single_1000_beijng_shoudu_kuan',
'single_1000_wuhan_kuan',
]
# video_path='/home/lishuang/Disk/shengshi_data/video_test_split_all/single_1000_beijng_shoudu_test_frame'
video_dir_path = '/home/lishuang/Disk/shengshi_data/video_test_split_all'
video_paths = os.listdir(video_dir_path)
for video_dir in video_paths:
if video_dir not in videopath_list:
print(video_dir, " pass")
continue
video_path = os.path.join(video_dir_path, video_dir)
# if video_dir !='double_54_zhuhai':
# continue
csv_path = os.path.join(video_dir_path, 'video_test_csv',
f'{video_dir}_video_cut.csv')
# csv_path = os.path.join(os.path.join(videopath, ".."), f'{basedirname}_video_cut.csv')
video_name = []
video_name_dic = {}
with open(csv_path) as f:
lines = f.readlines()[1:]
for line in lines:
line = line.rstrip()
items = line.split(',')
video_name.append(items[1])
video_name_dic[items[1]] = [
items[2], items[3], items[4], items[5]
]
if os.path.isdir(video_path):
video_files = os.listdir(video_path)
alarmvideo_list = {}
for video_file in video_files:
if video_file != '616643FEF1380C0E_2019-10-19-11-37-49-812_passenger_00000061_2.mp4':
continue
if video_file[:-4] not in video_name_dic:
continue
videosource = os.path.join(video_path, video_file)
# if len(os.listdir(videosource))==0:
# continue
save_img = True
view_img = True
videodataset = LoadImages(videosource, img_size=imgsz)
video_file, extension = os.path.splitext(video_file)
alarmvideo_list[video_file] = 0
frame_record = [0, 0, 0, 0, 0, 0, 0, 0, 0, 0]
frame_num = 0
outvideo = str(Path(out) / video_dir / video_file)
x1t, y1t, x2t, y2t = video_name_dic[video_file]
ratio_width = 1
ratio_height = 1
x1t = int(x1t) * ratio_width
x2t = int(x2t) * ratio_width
y1t = int(y1t) * ratio_height
y2t = int(y2t) * ratio_height
if os.path.exists(outvideo):
shutil.rmtree(outvideo) # delete output folder
os.makedirs(outvideo) # make new output folder
for path, img, im0s, vid_cap in videodataset: # one video
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float(
) # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
opt.conf_thres,
opt.iou_thres,
classes=opt.classes,
agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
boxnum = 0
boxnumbody = 0
boxnumhead = 0
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0 = path[i], '%g: ' % i, im0s[i].copy()
else:
p, s, im0 = path, '', im0s
save_path = str(Path(out) / video_dir / Path(p).name)
# txt_path = str(Path(out) /video_dir/video_file/ Path(p).stem) + ('_%g' % videodataset.frame if videodataset.mode == 'video' else '')
txt_path = str(
Path(out) / video_dir / video_file /
str(videodataset.frame))
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(
im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:],
det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (
det[:,
-1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]
) # add to string
# Write results
for *xyxy, conf, cls in det:
# if cls == 0:
# # label = 'person'
# boxnum += 1
# boxnumbody += 1
# elif cls == 1:
# # label = 'head'
# boxnumhead += 1
# if point_in_box(box_center, [x1, y1, x2, y2]):
# boxnumhead += 1 * person_result['class'] == 2
# boxnumbody += 1 * person_result['class'] == 1
if save_txt: # Write to file
xywh = (xyxy2xywh(
torch.tensor(xyxy).view(1, 4)) /
gn).view(
-1).tolist() # normalized xywh
with open(txt_path + '.txt', 'a') as f:
x0, y0, w0, h0 = xywh
h, w = im0.shape[:2]
x0 *= w
y0 *= h
w0 *= w
h0 *= h
x1 = x0 - w0 / 2
y1 = y0 - h0 / 2
if point_in_box([x0, y0],
[x1t, y1t, x2t, y2t]):
boxnumhead += 1 * cls == 1
boxnumbody += 1 * cls == 0
f.write(('%s ' + '%.2g ' + '%d ' * 3 +
'%d' + '\n') %
(names[int(cls)], conf, x1, y1,
w0, h0)) # label format
# f.write(('%ss '+'%.2g ' * 5 + '\n') % (names[int(cls)], conf,*xywh)) # label format
if save_img or view_img: # Add bbox to image
label = '%s %.2f' % (names[int(cls)], conf)
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3)
# Print time (inference + NMS)
print('%sDone. (%.3fs)' % (s, t2 - t1))
# Stream results
if view_img:
cv2.imshow(p, im0)
if cv2.waitKey(1) == ord('q'): # q to quit
raise StopIteration
# Save results (image with detections)
if save_img:
if videodataset.mode == 'images':
cv2.imwrite(save_path, im0)
else:
# if vid_path != save_path: # new video
# vid_path = save_path
# if isinstance(vid_writer, cv2.VideoWriter):
# vid_writer.release() # release previous video writer
#
# fourcc = 'mp4v' # output video codec
# fps = vid_cap.get(cv2.CAP_PROP_FPS)
# w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
# h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
# vid_writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*fourcc), fps,
# (w, h))
# vid_writer.write(im0)
image_path = os.path.join(
outvideo,
str(videodataset.frame) + '.jpg')
cv2.imwrite(image_path, im0)
if boxnumbody > 1 or boxnumhead > 1:
frame_record[frame_num % 10] = 1
else:
frame_record[frame_num % 10] = 0
frame_num += 1
if alarmvideo_list[video_file] == 0 and sum(
frame_record) > 7:
alarmvideo_list[video_file] = 1
image_path = os.path.join(
outvideo,
str(videodataset.frame) + '_alarmvideo.jpg')
cv2.imwrite(image_path, im0)
file_data = ""
for single_video in alarmvideo_list:
file_data += str(single_video) + ', value: ' + str(
alarmvideo_list[single_video]) + '\n'
with open(
f'{os.path.basename(video_path)}_video_result_{opt.conf_thres}.txt',
'a') as f:
f.write(file_data)
# if save_txt or save_img:
# print('Results saved to %s' % save_dir)
print('Done. (%.3fs)' % (time.time() - t0))
def detect(save_img=False):
frame_number = 0
source, weights, view_img, save_txt, imgsz, second_classifier = opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size, opt.second_classifier
webcam = source.isnumeric() or source.endswith('.txt') or source.lower().startswith(
('rtsp://', 'rtmp://', 'http://'))
# Directories
save_dir = Path(increment_path(Path(opt.project) / opt.name, exist_ok=opt.exist_ok)) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(parents=True, exist_ok=True) # make dir
# Initialize
set_logging()
device = select_device(opt.device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
stride = int(model.stride.max()) # model stride
imgsz = check_img_size(imgsz, s=stride) # check img_size
if half:
model.half() # to FP16
class_names1 = model.module.names if hasattr(model, 'module') else model.names
print(class_names1)
# Second-stage classifier
if second_classifier:
# TODO: Provide the list of class names for 2nd model
class_names2 = []
print(class_names2)
# If YOLO is trained for more than one class, then we mention which class is sub_class
if len(class_names1)>1:
super_class = 1 # Index of class that does not have sub-classes, currently for 2 classes
# it can be made a list for multiple superclasses
sub_class = 0 # Index of class that has sub-classes that image classifier will classify
# If YOLO is trained for just one class of objects, which is to be further classified by image classifier
else:
sub_class=0
super_class = None
modelc = create_model(len(class_names2),device)
# TODO: Provide the path to load the pre-trained image classifier model .pt file below
checkpoint = torch.load('')
modelc.load_state_dict(checkpoint['model_state_dict'])
modelc.to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
if webcam:
view_img = check_imshow()
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadStreams(source, img_size=imgsz, stride=stride)
else:
save_img = True
dataset = LoadImages(source, img_size=imgsz, stride=stride)
if second_classifier:
colors = [[random.randint(0, 255) for _ in range(3)] for _ in class_names2]
else:
colors = [[random.randint(0, 255) for _ in range(3)] for _ in class_names1]
# Run inference
if device.type != 'cpu':
model(torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(next(model.parameters()))) # run once
t0 = time.time()
for path, img, im0s, vid_cap in dataset:
frame_number+=1
original_image = img
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred, opt.conf_thres, opt.iou_thres, classes=opt.classes, agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply 2nd stage classifier
if second_classifier:
if pred[0].nelement()>0:
pred = apply_second_stage_classifier(pred, modelc, original_image, device, mean, std)
all_label = ""
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0, frame = path[i], '%g: ' % i, im0s[i].copy(), dataset.count
else:
p, s, im0, frame = path, '', im0s, getattr(dataset, 'frame', 0)
p = Path(p) # to Path
save_path = str(save_dir / p.name) # img.jpg
txt_path = str(save_dir / 'labels' / p.stem) + ('' if dataset.mode == 'image' else f'_{frame}') # img.txt
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
# s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
# Write results
for elements in reversed(det):
xyxy,conf,*cls = elements[:4],elements[4],elements[5:]
if save_txt:
with open(txt_path + '.txt', 'a') as f:
f.write(f"Frame number {frame_number} \n XYXY {xyxy}, conf {conf}, cls {cls}")
if save_img or view_img: # Add bbox to image
# If there was no sub_class detection
if len(cls[0])==1:
label = f'{class_names1[int(cls[0].item())]} {conf:.2f}'
plot_one_box(xyxy, im0, label=label, color=colors[int(cls[0].item())], line_thickness=3)
all_label+=label+","
# else, we find super class and sub class
else:
all_classes = cls[0].tolist()
# In general.py, we appended a column to every detection 'det' tensor
# If super class exists in the tensor, then we just show its class name
if super_class and super_class in all_classes:
# name_ind = int(sub_classes[sub_class])
label = f'{class_names1[super_class]} {conf:.2f}'
# If subclass existed in detection 'det' tensor and superclass didnt
# then we show name of subclass
else:
name_ind = int(all_classes[-1]) # Last element is the sub_class from image classifier
label = f'{class_names2[name_ind]} {conf:.2f}'
all_label+=label+","
plot_one_box(xyxy, im0, label=label, color=colors[name_ind], line_thickness=3)
# Print time (inference + NMS)
print(f'{s}Done. ({t2 - t1:.3f}s)')
# Stream results
if view_img:
cv2.imshow(str(p), im0)
cv2.waitKey(1) # 1 millisecond
# Save results (image with detections)
if save_img:
if dataset.mode == 'image':
cv2.imwrite(save_path, im0)
else: # 'video'
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release() # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
vid_writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*fourcc), fps, (w, h))
vid_writer.write(im0)
if save_txt or save_img:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}")
print(f'Done. ({time.time() - t0:.3f}s)')
def detect(save_img=False):
out, source, weights, view_img, save_txt, imgsz = \
opt.output, opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size
webcam = source.isnumeric() or source.startswith(
'rtsp') or source.startswith('http') or source.endswith('.txt')
# take a pic !!
img_arr, depth_arr = realsense.get_image(show=False) #array RGB
# print('shape=',depth_arr.shape) #(480, 640)
print('central depth=', depth_arr[
240,
320]) #depth at center in mm #should it be depth_arr[row,col]?
x1, y1, x2, y2, depth_avg = 0, 0, 0, 0, 0
xyz_obj = np.array([0, 0, 0])
K = np.array([[609.674560546875, 0.0, 323.9862365722656],
[0.0, 608.5648193359375, 227.5126495361328],
[0.0, 0.0, 1.0]]) # intrinsic ?
# convert
img_pad = letterbox(img_arr[:, :, ::-1],
new_shape=imgsz)[0] # first to BGR and padding
img_pad = img_pad[:, :, ::-1].transpose(2, 0,
1) # BGR to RGB, to 3x416x416
img_pad = np.ascontiguousarray(img_pad) # 将一个内存不连续存储的数组转换为内存连续存储的数组
# cv2.imwrite('letterbox.jpg', 255 * img.transpose((1, 2, 0))[:, :, ::-1]) # save letterbox image
realsense_once = 1
# 1: take an image from realsense and infer without imwrite locally;
# 0: infer all images from the path, including one taken from realsense
if realsense_once == 0:
cv2.imwrite('./inference/images/snap.jpg',
cv2.cvtColor(img_arr,
cv2.COLOR_RGB2BGR)) #opencv assume BGR
# Initialize
set_logging()
device = select_device(opt.device)
if os.path.exists(out):
shutil.rmtree(out) # delete output folder
os.makedirs(out) # make new output folder
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(
torch.load('weights/resnet101.pt',
map_location=device)['model']) # load weights
modelc.to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
if webcam:
view_img = True
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadStreams(source, img_size=imgsz)
else:
save_img = True
dataset = LoadImages(source, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)]
for _ in range(len(names))]
# Run inference !!
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img
) if device.type != 'cpu' else None # run once
# infer one image from realsense
if realsense_once:
img = img_pad
im0s = img_arr[:, :, ::-1] # BGR
path = '/home/hanwen/test_ros_ws/src/yolov5_test/scripts/inference/images/snap.jpg'
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
opt.conf_thres,
opt.iou_thres,
classes=opt.classes,
agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0 = path[i], '%g: ' % i, im0s[i].copy()
else:
p, s, im0 = path, '', im0s
save_path = str(Path(out) / Path(p).name)
txt_path = str(Path(out) / Path(p).stem) + (
'_%g' % dataset.frame if dataset.mode == 'video' else '')
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]) # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
# average depth
x1, y1, x2, y2 = xyxy[0].cpu().numpy(), xyxy[1].cpu(
).numpy(), xyxy[2].cpu().numpy(), xyxy[3].cpu().numpy()
print('x1,y1,x2,y2 = ', x1, y1, x2, y2)
xc, yc = (x1 + x2) / 2, (y1 + y2) / 2
x1c, x2c, y1c, y2c = (x1 + xc) / 2, (x2 + xc) / 2, (
y1 + yc) / 2, (y2 + yc) / 2
depth_4samples = [
depth_arr[int(y1c), int(x1c)], depth_arr[int(y1c),
int(x2c)],
depth_arr[int(y2c), int(x1c)], depth_arr[int(y2c),
int(x2c)]
]
print('depth_4samples:', depth_4samples)
depth_validsamples = [b for b in depth_4samples if b > 0]
depth_avg = np.mean(depth_validsamples)
print('depth for grasping =', depth_avg)
xyz_obj = np.dot(np.linalg.inv(K),
depth_avg *
np.array([xc, yc, 1]).transpose()
) # estimated object center-XYZ (mm)
xyz_obj = xyz_obj.transpose()
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * 5 + '\n') %
(cls, *xywh)) # label format
if save_img or view_img: # Add bbox to image
label = '%s %.2f' % (names[int(cls)], conf)
im0 = np.array(
im0
) # this fix the error https://github.com/opencv/opencv/issues/18120
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3) # error
# Print time (inference + NMS)
print('%sDone. (%.3fs)' % (s, t2 - t1)) #done with one det
# Stream results
if view_img:
cv2.imshow(p, im0)
if cv2.waitKey(1) == ord('q'): # q to quit
raise StopIteration
# Save results (image with detections)
if save_img:
if dataset.mode == 'images':
cv2.imwrite(save_path, im0)
else:
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release(
) # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
vid_writer = cv2.VideoWriter(
save_path, cv2.VideoWriter_fourcc(*fourcc), fps,
(w, h))
vid_writer.write(im0)
# realsense_once=0, use the default dataloader
else:
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
opt.conf_thres,
opt.iou_thres,
classes=opt.classes,
agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0 = path[i], '%g: ' % i, im0s[i].copy()
else:
p, s, im0 = path, '', im0s
save_path = str(Path(out) / Path(p).name)
txt_path = str(Path(out) / Path(p).stem) + (
'_%g' % dataset.frame if dataset.mode == 'video' else '')
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]) # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
# average depth
x1, y1, x2, y2 = xyxy[0].cpu().numpy(), xyxy[1].cpu(
).numpy(), xyxy[2].cpu().numpy(), xyxy[3].cpu().numpy(
)
print('x1,y1,x2,y2 = ', x1, y1, x2, y2)
xc, yc = (x1 + x2) / 2, (y1 + y2) / 2
x1c, x2c, y1c, y2c = (x1 + xc) / 2, (x2 + xc) / 2, (
y1 + yc) / 2, (y2 + yc) / 2
depth_4samples = [
depth_arr[int(y1c), int(x1c)], depth_arr[int(y1c),
int(x2c)],
depth_arr[int(y2c), int(x1c)], depth_arr[int(y2c),
int(x2c)]
]
print('depth_4samples:', depth_4samples)
depth_validsamples = [
b for b in depth_4samples if b > 0
]
depth_avg = np.mean(depth_validsamples)
print('depth for grasping =', depth_avg)
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * 5 + '\n') %
(cls, *xywh)) # label format
if save_img or view_img: # Add bbox to image
label = '%s %.2f' % (names[int(cls)], conf)
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3)
# Print time (inference + NMS)
print('%sDone. (%.3fs)' % (s, t2 - t1)) #done with one det
# Stream results
if view_img:
cv2.imshow(p, im0)
if cv2.waitKey(1) == ord('q'): # q to quit
raise StopIteration
# Save results (image with detections)
if save_img:
if dataset.mode == 'images':
cv2.imwrite(save_path, im0)
else:
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release(
) # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
vid_writer = cv2.VideoWriter(
save_path, cv2.VideoWriter_fourcc(*fourcc),
fps, (w, h))
vid_writer.write(im0)
if save_txt or save_img:
print('Results saved to %s' % Path(out))
if platform.system() == 'Darwin' and not opt.update: # MacOS
os.system('open ' + save_path)
print('All Done. (%.3fs)' % (time.time() - t0)) #done with all imgs
return x1, y1, x2, y2, xyz_obj[0], xyz_obj[1], xyz_obj[2], depth_avg
def run(
weights='yolov5s.pt', # model.pt path(s)
source='./test_1', # file/dir/URL/glob, 0 for webcam
imgsz=640, # inference size (pixels)
conf_thres=0.25, # confidence threshold
iou_thres=0.45, # NMS IOU threshold
max_det=1000, # maximum detections per image
device='', # = device, i.e. 0 or 0,1,2,3 or cpu
view_img=False, # show results
save_txt=False, # save results to *.txt
save_conf=False, # save confidences in --save-txt labelss
save_crop=False, # save cropped prediction boxes
nosave=False, # do not save images/videos
classes=None, # filter by class: --class 0, or --class 0 2 3
agnostic_nms=False, # class-agnostic NMS
augment=False, # augmented inference
update=False, # update all models
project='runs/detect', # save results to project/name
name='exp', # save results to project/name
exist_ok=False, # existing project/name ok, do not increment
line_thickness=3, # bounding box thickness (pixels)
hide_labels=False, # hide labels
hide_conf=False, # hide confidences
half=False, # use FP16 half-precision inference
):
desire_param = []
coor = []
all_info = []
save_img = not nosave and not source.endswith(
'.txt') # save inference images
# Directories
save_dir = increment_path(Path(project) / name,
exist_ok=exist_ok) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(
parents=True, exist_ok=True) # make dir
# Initialize
set_logging()
device = select_device(device)
half &= device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
stride = int(model.stride.max()) # model stride
imgsz = check_img_size(imgsz, s=stride) # check image size
names = model.module.names if hasattr(
model, 'module') else model.names # get class names
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet50', n=2) # initialize
modelc.load_state_dict(
torch.load('resnet50.pt',
map_location=device)['model']).to(device).eval()
# Set Dataloader
dataset = LoadImages(source, img_size=imgsz, stride=stride)
# Run inference
if device.type != 'cpu':
model(
torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(
next(model.parameters()))) # run once
t0 = time.time()
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
conf_thres,
iou_thres,
classes,
agnostic_nms,
max_det=max_det)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
p, s, im0, frame = path, '', im0s.copy(), getattr(
dataset, 'frame', 0)
p = Path(p) # to Path
save_path = str(save_dir / p.name) # img.jpg
#print()
txt_path = str(save_dir / 'labels' / p.stem) + (
'' if dataset.mode == 'image' else f'_{frame}') # img.txt
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
imc = im0.copy() if save_crop else im0 # for save_crop
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
pr = ' '
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
pr += f"{n} {names[int(c)]}{'s' * (n > 1)},"
desire_param.append({"image_id": p.name, "prediction": pr})
# for img_name in enumerate(p.name):
# if img_name not in desire_param:
# add_image = (img_name, "predictioni 0")
# desire_param.append(add_image)
# print(desire_param)
# Write results
for *xyxy, conf, cls in reversed(det):
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh,
conf) if save_conf else (cls,
*xywh) # label format
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * len(line)).rstrip() % line + '\n')
if save_img or save_crop or view_img: # Add bbox to image
c = int(cls) # integer class
label = None if hide_labels else (
names[c]
if hide_conf else f'{names[c]} {conf:.2f}')
plot_one_box(xyxy,
im0,
label=label,
color=colors(c, True),
line_thickness=line_thickness)
if save_crop:
save_one_box(xyxy,
imc,
file=save_dir / 'crops' / names[c] /
f'{p.stem}.jpg',
BGR=True)
# Print time (inference + NMS)
# print(f'{s}Done. ({t2 - t1:.3f}s)')
result = [(f'{s}Done. ({t2 - t1:.3f}s)')]
print(result)
# Stream results
if view_img:
cv2.imshow(str(p), im0)
cv2.waitKey(1) # 1 millisecond
# Save results (image with detections)
if save_img:
if dataset.mode == 'image':
cv2.imwrite(save_path, im0)
else:
None
if save_txt or save_img:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}")
if update:
strip_optimizer(weights) # update model (to fix SourceChangeWarning)
print(f'Done. ({time.time() - t0:.3f}s)')
# lst = []
# for child in desire_param:
# info = ["img_name", "prediction"]
# lst1 = {k: v for k, v in zip(info, child)}
# lst.append(lst1)
with open('result.json', 'w') as f:
json.dump(desire_param, f)
def detect(save_img=False):
source, weights, SAVE_PATH, view_img, save_txt, imgsz = opt.source, opt.weights, opt.SAVE_PATH, opt.view_img, opt.save_txt, opt.img_size
webcam = source.isnumeric() or source.endswith(
'.txt') or source.lower().startswith(('rtsp://', 'rtmp://', 'http://'))
print(f'SAVE_PATHは{SAVE_PATH}')
# Directories
# save_dir = Path(increment_path(Path(opt.project) / opt.name, exist_ok=opt.exist_ok)) # increment run
save_dir = Path(SAVE_PATH, exist_ok=opt.exist_ok)
(save_dir / 'labels' if save_txt else save_dir).mkdir(
parents=True, exist_ok=True) # make dir
# Initialize
set_logging()
device = select_device(opt.device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(
torch.load('weights/resnet101.pt',
map_location=device)['model']).to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
if webcam:
view_img = True
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadStreams(source, img_size=imgsz)
else:
save_img = True
dataset = LoadImages(source, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img
) if device.type != 'cpu' else None # run once
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred,
opt.conf_thres,
opt.iou_thres,
classes=opt.classes,
agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0, frame = path[i], '%g: ' % i, im0s[i].copy(
), dataset.count
else:
p, s, im0, frame = path, '', im0s, getattr(dataset, 'frame', 0)
p = Path(p) # to Path
save_path = str(save_dir / p.name) # img.jpg
txt_path = str(save_dir / 'labels' / p.stem) + (
'' if dataset.mode == 'image' else f'_{frame}') # img.txt
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh, conf) if opt.save_conf else (
cls, *xywh) # label format
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * len(line)).rstrip() % line + '\n')
if save_img or view_img: # Add bbox to image
label = f'{names[int(cls)]} {conf:.2f}'
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3)
# Print time (inference + NMS)
print(f'{s}Done. ({t2 - t1:.3f}s)')
# Stream results
if view_img:
cv2.imshow(str(p), im0)
# Save results (image with detections)
if save_img:
if dataset.mode == 'image':
cv2.imwrite(save_path, im0)
else: # 'video'
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release(
) # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
vid_writer = cv2.VideoWriter(
save_path, cv2.VideoWriter_fourcc(*fourcc), fps,
(w, h))
vid_writer.write(im0)
if save_txt or save_img:
s = f"\n{len(list(save_dir.glob('labels/*.txt')))} labels saved to {save_dir / 'labels'}" if save_txt else ''
print(f"Results saved to {save_dir}{s}")
print(f'Done. ({time.time() - t0:.3f}s)')
def handle_content_message(event):
if isinstance(event.message, ImageMessage):
ext = 'jpg'
else:
return
message_content = line_bot_api.get_message_content(event.message.id)
with tempfile.NamedTemporaryFile(dir=static_tmp_path,
prefix=ext + '-',
delete=False) as tf:
for chunk in message_content.iter_content():
tf.write(chunk)
tempfile_path = tf.name
dist_path = tempfile_path + '.' + ext
dist_name = os.path.basename(dist_path)
os.rename(tempfile_path, dist_path)
# Set Dataloader
dataset = LoadImages(dist_path, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img
) if device.type != 'cpu' else None # run once
for path, img, im0s, vid_cap in dataset:
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=False)[0]
# Apply NMS
pred = non_max_suppression(pred,
conf_thres,
iou_thres,
classes=classes,
agnostic=agnostic_nms)
t2 = time_synchronized()
# Process detections
for i, det in enumerate(pred): # detections per image
p, s, im0, frame = path, '', im0s, getattr(dataset, 'frame', 0)
p = Path(p) # to Path
save_path = str(save_dir + p.name) # img.jpg
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += f'{n} {names[int(c)]}s, ' # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) /
gn).view(-1).tolist() # normalized xywh
line = (cls, *xywh,
conf) if save_conf else (cls,
*xywh) # label format
if save_img or view_img: # Add bbox to image
label = f'{names[int(cls)]} {conf:.2f}'
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3)
# Save results (image with detections)
if save_img:
if dataset.mode == 'image':
cv2.imwrite(save_path, im0)
# Print time (inference + NMS)
print(f'{s}Done. ({t2 - t1:.3f}s)')
url = request.url_root + '/static/tmp/' + dist_name
line_bot_api.reply_message(event.reply_token, [
TextSendMessage(text='Object detection result:'),
ImageSendMessage(url, url)
])
def detect(save_img=False):
out, source, weights, view_img, save_txt, imgsz = \
opt.output, opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size
webcam = source == '0' or source.startswith('rtsp') or source.startswith('http') or source.endswith('.txt')
screenshot = source == '1' or source == '2' #1 is to screenshot main screen, 2 is to screenshot second screen
if screenshot:
num_screen_tmp = int(source)
# Initialize
set_logging()
device = select_device(opt.device)
if os.path.exists(out):
shutil.rmtree(out) # delete output folder
os.makedirs(out) # make new output folder
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
imgsz = check_img_size(imgsz, s=model.stride.max()) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(torch.load('weights/resnet101.pt', map_location=device)['model']) # load weights
modelc.to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
if screenshot:
view_img = True
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadScreen(num_screen=num_screen_tmp, img_size=imgsz) #img_size=640
elif webcam:
view_img = True
cudnn.benchmark = True # set True to speed up constant image size inference
dataset = LoadStreams(source, img_size=imgsz)
else:
save_img = True
dataset = LoadImages(source, img_size=imgsz)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in range(len(names))]
# Run inference
t0 = time.time()
t4 = time_synchronized()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model(img.half() if half else img) if device.type != 'cpu' else None # run once
for path, img, im0s, vid_cap in dataset:
timer = cv2.getTickCount()
img = torch.from_numpy(img).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
# Inference
t1 = time_synchronized()
pred = model(img, augment=opt.augment)[0]
# Apply NMS
pred = non_max_suppression(pred, opt.conf_thres, opt.iou_thres, classes=opt.classes, agnostic=opt.agnostic_nms)
t2 = time_synchronized()
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
cv2.namedWindow(path, cv2.WND_PROP_FULLSCREEN)
# cv2.setWindowProperty(path, cv2.WND_PROP_FULLSCREEN, cv2.WINDOW_FULLSCREEN)
# Process detections
for i, det in enumerate(pred): # detections per image
if webcam: # batch_size >= 1
p, s, im0 = path[i], '%g: ' % i, im0s[i].copy()
else:
p, s, im0 = path, '', im0s
save_path = str(Path(out) / Path(p).name)
txt_path = str(Path(out) / Path(p).stem) + ('_%g' % dataset.frame if dataset.mode == 'video' else '')
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1, 0]] # normalization gain whwh
if det is not None and len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4], im0.shape).round()
# Print results
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]) # add to string
# Write results
for *xyxy, conf, cls in reversed(det):
if save_txt: # Write to file
xywh = (xyxy2xywh(torch.tensor(xyxy).view(1, 4)) / gn).view(-1).tolist() # normalized xywh
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * 5 + '\n') % (cls, *xywh)) # label format
if save_img or view_img: # Add bbox to image
label = '%s %.2f' % (names[int(cls)], conf)
plot_one_box(xyxy, im0, label=label, color=colors[int(cls)], line_thickness=3)
# Stream results
if view_img:
img_cropped = cv2.cvtColor(im0, cv2.COLOR_BGR2RGB)
fps = cv2.getTickFrequency() / (cv2.getTickCount() - timer)
cv2.putText(img_cropped, "FPS=" + str(int(fps)), (50, 50), cv2.FONT_HERSHEY_COMPLEX, 0.8, (0, 0, 255), 2)
cv2.imshow(p, img_cropped)
#cv2.imshow(p, im0)
if cv2.waitKey(1) == ord('q'): # q to quit
raise StopIteration
# Save results (image with detections)
if save_img:
if dataset.mode == 'images':
cv2.imwrite(save_path, im0)
else:
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release() # release previous video writer
fourcc = 'mp4v' # output video codec
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
vid_writer = cv2.VideoWriter(save_path, cv2.VideoWriter_fourcc(*fourcc), fps, (w, h))
vid_writer.write(im0)
t3 = time_synchronized()
# Print time (inference + NMS)
print('%sDone. (%.1f (%.1f) FPS total (just NN processing: %.3fs inference, %.3fs post))' % (s, 1.0/(t3-t4), 1.0/(t3-t1), t2 - t1, t3 - t2))
t4 = time_synchronized()
if save_txt or save_img:
print('Results saved to %s' % Path(out))
if platform.system() == 'Darwin' and not opt.update: # MacOS
os.system('open ' + save_path)
print('Done. (%.3fs)' % (time.time() - t0))
def detect(save_img=False):
os.environ['CUDA_VISIBLE_DEVICES'] = opt.device
source, weights, save_img, save_txt, imgsz = opt.source, opt.weights, opt.save_img, opt.save_txt, opt.img_size
#save_img = not opt.nosave and not source.endswith('.txt') # save inference images
webcam = source.isnumeric() or source.endswith(
'.txt') or source.lower().startswith(('rtsp://', 'rtmp://', 'http://'))
# Directories
save_dir = Path(
increment_path(Path(opt.project) / opt.name,
exist_ok=opt.exist_ok)) # increment run
(save_dir / 'labels' if save_txt else save_dir).mkdir(
parents=True, exist_ok=True) # make dir
print(save_dir)
# Initialize
set_logging()
device = select_device(opt.device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = attempt_load(weights, map_location=device) # load FP32 model
stride = int(model.stride.max()) # model stride
imgsz = check_img_size(imgsz, s=stride) # check img_size
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = load_classifier(name='resnet101', n=2) # initialize
modelc.load_state_dict(
torch.load('weights/resnet101.pt',
map_location=device)['model']).to(device).eval()
# Set Dataloader
vid_path, vid_writer = None, None
# dataset = LoadImages_panda(source, img_size=imgsz, stride=stride, split_size=[[1000, 1000],[1500, 1500], [3000, 3000],[6000, 6000],[10000,10000]], over_lap=0.3)
dataset = LoadImages_panda(source,
img_size=imgsz,
stride=stride,
split_size=[[1500, 1500], [3000, 3000],
[6000, 6000], [10000, 10000]],
over_lap=0.3)
#dataset = LoadImages_panda(source, img_size=imgsz, stride=stride,split_size=[], over_lap=0.3)
# Get names and colors
names = model.module.names if hasattr(model, 'module') else model.names
colors = [[random.randint(0, 255) for _ in range(3)] for _ in names]
ret_matrix = np.zeros((2, 2))
# Run inference
if device.type != 'cpu':
model(
torch.zeros(1, 3, imgsz, imgsz).to(device).type_as(
next(model.parameters()))) # run once
t0 = time.time()
jdict = []
results = []
for path, img_list, start_list, split_size_list, boundary_list, img0_list, im0s, vid_cap in dataset:
det_list = []
det_scale_dict = {}
# Inference
t1 = time_synchronized()
for img_i in range(len(img_list)):
img = torch.from_numpy(img_list[img_i]).to(device)
img = img.half() if half else img.float() # uint8 to fp16/32
img /= 255.0 # 0 - 255 to 0.0 - 1.0
if img.ndimension() == 3:
img = img.unsqueeze(0)
pred = model(img, augment=opt.augment)[0]
# Apply
pred_list = []
pred_output = non_max_suppression(pred,
opt.conf_thres,
opt.iou_thres,
classes=opt.classes,
agnostic=opt.agnostic_nms)
if pred_list == []:
pred_list = pred_output[0]
if len(pred_output[0]):
pred_list = torch.cat((pred_list, pred_output[0]), dim=0)
pred_out = [pred_list]
# Apply Classifier
if classify:
pred_out = apply_classifier(pred_out, modelc, img,
img0_list[img_i])
# Process detections
for i, det in enumerate(pred_out): # detections per image
if webcam: # batch_size >= 1
p, s, im0, frame = path[i], '%g: ' % i, img0_list[img_i][
i].copy(), dataset.count
else:
p, s, im0, frame = path, '', img0_list[img_i], getattr(
dataset, 'frame', 0)
p = Path(p) # to Path
save_path = str(save_dir / p.name) # img.jpg
txt_path = str(save_dir / 'labels' / p.stem) + (
'' if dataset.mode == 'image' else f'_{frame}') # img.txt
s += '%gx%g ' % img.shape[2:] # print string
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
det = del_more(det,
img.shape[2:],
big_thres=0.3,
boundary=boundary_list[img_i])
# del_small
if len(det) != 0:
small_thres = 15
dets_wh_thres = det[:, 2:4] - det[:, :2]
det_thres = torch.minimum(dets_wh_thres[:, 0],
dets_wh_thres[:, 1])
if split_size_list[img_i] > 2000:
det = det[det_thres > small_thres]
if len(det):
# Rescale boxes from img_size to im0 size
det[:, :4] = scale_coords(img.shape[2:], det[:, :4],
im0.shape).round()
det[..., 0] += start_list[img_i][0]
det[..., 2] += start_list[img_i][0]
det[..., 1] += start_list[img_i][1]
det[..., 3] += start_list[img_i][1]
if det_list == []:
det_list = det
else:
det_list = torch.cat((det_list, det), dim=0)
if split_size_list[img_i] not in det_scale_dict.keys():
det_scale_dict[split_size_list[img_i]] = det
else:
det_scale_dict[split_size_list[img_i]] = torch.cat(
(det_scale_dict[split_size_list[img_i]], det),
dim=0)
det_list = fuse_all_det(det_list[:, :6],
im0,
conf_thres=opt.conf_thres,
nms_thres=opt.iou_thres,
method='standard',
merge=False)
scale_key = []
for key in det_scale_dict.keys():
scale_key.append(key)
det_list = WBF_fuse(im0, [det_list, det_scale_dict[scale_key[0]]],
weights=[1, 1],
iou_thres=0.5,
conf_thres=0.5)
# Print results
for c in det_list[:, -1].unique():
n = (det_list[:, -1] == c).sum() # detections per class
s += f"{n} {names[int(c)]}{'s' * (n > 1)}, " # add to string
# Write results
frame_id = int(path.split(".")[-2].split("_")[-1])
id_list = []
online_tlwhs = []
for *xyxy, conf, cls in reversed(det_list):
online_tlwhs.append((float(xyxy[0]), float(xyxy[1]),
float(xyxy[2]) - float(xyxy[0]),
float(xyxy[3]) - float(xyxy[1])))
id_list.append(conf)
if save_img: # Add bbox to image
plot_one_box(xyxy,
im0s,
label=0,
color=colors[int(cls)],
line_thickness=3)
results.append((frame_id, online_tlwhs, id_list))
# Print time (per image)
t2 = time_synchronized()
print(f'{s}Done. ({t2 - t1:.3f}s)')
sys.stdout.flush()
# Save results (image with detections)
if save_img:
if dataset.mode == 'image':
cv2.imwrite(save_path, im0s)
else: # 'video' or 'stream'
if vid_path != save_path: # new video
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release() # release previous video writer
if vid_cap: # video
fps = vid_cap.get(cv2.CAP_PROP_FPS)
w = int(vid_cap.get(cv2.CAP_PROP_FRAME_WIDTH))
h = int(vid_cap.get(cv2.CAP_PROP_FRAME_HEIGHT))
else: # stream
fps, w, h = 30, im0.shape[1], im0.shape[0]
save_path += '.mp4'
vid_writer = cv2.VideoWriter(
save_path, cv2.VideoWriter_fourcc(*'mp4v'), fps,
(w, h))
vid_writer.write(im0)
result_detection, id_num = write_results(
opt.source.split("/")[-1] + ".txt", results, "mot")
print("detection_num", result_detection)
print("id_num", id_num)
print(f'Done. ({time.time() - t0:.3f}s)')
print(f'object num. ({len(jdict)})')
sys.stdout.flush()
