def mainFunc(args):
# Set the main function flag
print("Main Function Start...")
# Check the GPU device
print("Number of available GPUs: {}".format(torch.cuda.device_count()))
# Check whether using the distributed runing for the network
is_distributed = initDistributed(args)
master = True
if is_distributed and os.environ["RANK"]:
master = int(
os.environ["RANK"]) == 0 # check whether this node is master node
# Configuration for device setting
set_logging()
if is_distributed:
device = torch.device('cuda:{}'.format(args.local_rank))
else:
device = select_device(args.device)
half = device.type != 'cpu' # half precision only supported on CUDA
# Load the configuration
config = loadConfig(args.config)
# CuDNN related setting
if torch.cuda.is_available():
cudnn.benchmark = config.DEVICE.CUDNN.BENCHMARK
cudnn.deterministic = config.DEVICE.CUDNN.DETERMINISTIC
cudnn.enabled = config.DEVICE.CUDNN.ENABLED
# Configurations for dirctories
save_img, save_dir, source, yolov5_weights, view_img, save_txt, imgsz = \
False, Path(args.save_dir), args.source, args.weights, args.view_img, args.save_txt, args.img_size
webcam = source.isnumeric() or source.startswith(
('rtsp://', 'rtmp://', 'http://')) or source.endswith('.txt')
if save_dir == Path('runs/detect'): # if default
os.makedirs('runs/detect', exist_ok=True) # make base
save_dir = Path(increment_dir(save_dir / 'exp',
args.name)) # increment run
os.makedirs(save_dir / 'labels' if save_txt else save_dir,
exist_ok=True) # make new dir
# Load yolov5 model for human detection
model_yolov5 = attempt_load(config.MODEL.PRETRAINED.YOLOV5,
map_location=device)
imgsz = check_img_size(imgsz,
s=model_yolov5.stride.max()) # check img_size
if half:
model_yolov5.half() # to FP16
# Second-stage classifier
classify = False
if classify:
model_classifier = load_classifier(name='resnet101', n=2) # initialize
model_classifier.load_state_dict(
torch.load('weights/resnet101.pt',
map_location=device)['model']) # load weights
model_classifier.to(device).eval()
# Load resnet model for human keypoints estimation
model_resnet = eval('pose_models.' + config.MODEL.NAME.RESNET +
'.get_pose_net')(config, is_train=False)
if config.EVAL.RESNET.MODEL_FILE:
print('=> loading model from {}'.format(config.EVAL.RESNET.MODEL_FILE))
model_resnet.load_state_dict(torch.load(config.EVAL.RESNET.MODEL_FILE),
strict=False)
else:
print('expected model defined in config at EVAL.RESNET.MODEL_FILE')
model_resnet.to(device)
model_resnet.eval()
# Set Dataloader
vid_path, vid_writer = None, None
if webcam:
view_img = True
dataset = LoadStreams(source, img_size=imgsz)
else:
save_img = True
dataset = LoadImages(source, img_size=imgsz)
pose_transform = transforms.Compose(
[ # input transformation for 2d human pose estimation
transforms.ToTensor(),
transforms.Normalize(mean=[0.485, 0.456, 0.406],
std=[0.229, 0.224, 0.225])
])
# Get names and colors
names = model_yolov5.module.names if hasattr(
model_yolov5, 'module') else model_yolov5.names
colors = [[random.randint(0, 255) for _ in range(3)]
for _ in range(len(names))]
# Construt filters for filtering 2D/3D human keypoints
# filters_2d = constructFilters((1,16,2), freq=25, mincutoff=1, beta=0.01) # for test
# filters_3d = constructFilters((1,16,3), freq=25, mincutoff=1, beta=0.01)
# Run the yolov5 and resnet for 2d human pose estimation
# with torch.no_grad():
# Run inference
t0 = time.time()
img = torch.zeros((1, 3, imgsz, imgsz), device=device) # init img
_ = model_yolov5(img.half() if half else img
) if device.type != 'cpu' else None # run once
# Process every video frame
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_boxes = model_yolov5(img, augment=args.augment)[0]
# Apply NMS
pred_boxes = non_max_suppression(pred_boxes,
args.conf_thres,
args.iou_thres,
classes=args.classes,
agnostic=args.agnostic_nms)
t2 = time_synchronized()
# Can not find people and move to next frame
if pred_boxes[0] is None:
# show the frame with no human detected
cv2.namedWindow("2D Human Pose Estimation", cv2.WINDOW_NORMAL)
cv2.imshow("2D Human Pose Estimation", im0s[0].copy())
# wait manual operations
# with kb.Listener(on_press=on_press) as listener:
# listener.join()
# return
# if kb.is_pressed('t'):
# return
print("No Human Detected and Move on.")
print("-" * 30)
continue
# Print time (inference + NMS)
detect_time = t2 - t1
detect_fps = 1.0 / detect_time
print("Human Detection Time: {}, Human Detection FPS: {}".format(
detect_time, detect_fps))
# Apply Classifier
if classify: # false
pred_boxes = apply_classifier(pred_boxes, model_classifier, img,
im0s)
# Estimate 2d human pose(multiple person)
centers = []
scales = []
for id, boxes in enumerate(pred_boxes):
if boxes is not None and len(boxes):
boxes[:, :4] = scale_coords(img.shape[2:], boxes[:, :4],
im0s[id].copy().shape).round()
# convert tensor to list format
boxes = np.delete(boxes.cpu().numpy(), [-2, -1], axis=1).tolist()
for l in range(len(boxes)):
boxes[l] = [tuple(boxes[l][0:2]), tuple(boxes[l][2:4])]
# convert box to center and scale
for box in boxes:
center, scale = box_to_center_scale(box, imgsz, imgsz)
centers.append(center)
scales.append(scale)
t3 = time_synchronized()
pred_pose_2d = get_pose_estimation_prediction(config,
model_resnet,
im0s[0],
centers,
scales,
transform=pose_transform,
device=device)
t4 = time_synchronized()
# Print time (2d human pose estimation)
estimate_time = t4 - t3
estimate_fps = 1.0 / estimate_time
print("Pose Estimation Time: {}, Pose Estimation FPS: {}".format(
estimate_time, estimate_fps))
# Filter the predicted 2d human pose(multiple person)
t5 = time_synchronized()
# if False: # for test
if config.EVAL.RESNET.USE_FILTERS_2D:
# construct filters for every keypoints of every person in 2D
filters_2d = constructFilters(pred_pose_2d.shape,
freq=1,
mincutoff=1,
beta=0.01)
print("Shape of filters_2d: ({}, {}, {})".format(
len(filters_2d), len(filters_2d[0]),
len(filters_2d[0][0]))) # for test
for per in range(pred_pose_2d.shape[0]):
for kp in range(pred_pose_2d.shape[1]):
for coord in range(pred_pose_2d.shape[2]):
pred_pose_2d[per][kp][coord] = filters_2d[per][kp][
coord](pred_pose_2d[per][kp][coord])
t6 = time_synchronized()
# Print time (filter 2d human pose)
filter_time_2d = t6 - t5
filter_fps_2d = 1.0 / filter_time_2d
print("Filter 2D Pose Time: {}, Filter 2D Pose FPS: {}".format(
filter_time_2d, filter_fps_2d))
# Process detections and estimations in 2D
for i, box in enumerate(pred_boxes):
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 box is not None and len(box):
# Rescale boxes from img_size to im0 size
box[:, :4] = scale_coords(img.shape[2:], box[:, :4],
im0.shape).round()
# Print results
for c in box[:, -1].unique():
n = (box[:, -1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]) # add to string
# Write results
for *xyxy, conf, cls in reversed(box):
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 args.save_conf else (
cls, *xywh) # label format
with open(txt_path + '.txt', 'a') as f:
f.write(('%g ' * len(line) + '\n') % line)
# Add bbox to image
if save_img or view_img:
label = '%s %.2f' % (names[int(cls)], conf)
plot_one_box(xyxy,
im0,
label=label,
color=colors[int(cls)],
line_thickness=3)
# Draw joint keypoints, number orders and human skeletons for every detected people in 2D
for person in pred_pose_2d:
# draw the human keypoints
for idx, coord in enumerate(person):
x_coord, y_coord = int(coord[0]), int(coord[1])
cv2.circle(im0, (x_coord, y_coord), 1, (0, 0, 255), 5)
cv2.putText(im0, str(idx), (x_coord, y_coord),
cv2.FONT_HERSHEY_SIMPLEX, 0.6,
(255, 255, 255), 2, cv2.LINE_AA)
# draw the human skeletons in PACIFIC mode
for skeleton in PACIFIC_SKELETON_INDEXES:
cv2.line(im0, (int(person[skeleton[0]][0]),
int(person[skeleton[0]][1])),
(int(person[skeleton[1]][0]),
int(person[skeleton[1]][1])), skeleton[2], 2)
# Print time (inference + NMS + estimation)
print('%sDone. (%.3fs)' % (s, t4 - t1))
# Stream results
if view_img:
detect_text = "Detect FPS:{0:0>5.2f}/{1:0>6.2f}ms".format(
detect_fps, detect_time * 1000)
estimate_text = "Estimate FPS:{0:0>5.2f}/{1:0>6.2f}ms".format(
estimate_fps, estimate_time * 1000)
cv2.putText(im0, detect_text, (10, 30),
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 2,
cv2.LINE_AA)
cv2.putText(im0, estimate_text, (10, 60),
cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 0, 255), 2,
cv2.LINE_AA)
cv2.namedWindow("2D Human Pose Estimation", cv2.WINDOW_NORMAL)
cv2.imshow("2D Human Pose Estimation", im0)
if cv2.waitKey(1) & 0xFF == ord('q'): # q to quit
return
# goto .mainFunc
# 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)
# Print time (inference + NMS + estimation + 2d filtering)
all_process_time = t6 - t1
all_process_fps = 1.0 / all_process_time
print("All Process Time: {}, All Process FPS: {}".format(
all_process_time, all_process_fps))
print("-" * 30)
# Goto label
# label .mainFunc
# Print saving results
if save_txt or save_img:
print('Results saved to %s' % save_dir)
# Release video reader and writer, then destory all opencv windows
dataset.vid_cap.release()
vid_writer.release()
cv2.destroyAllWindows()
print('Present 2D Human Pose Inference Done. Total Time:(%.3f seconds)' %
(time.time() - t0))
def detect(
weights="yolov5s.pt",
source="yolov5/data/images",
img_size=640,
conf_thres=0.75,
iou_thres=0.45,
device="",
view_img=False,
save_txt=False,
save_conf=False,
classes=None,
agnostic_nms=False,
augment=False,
update=False,
project="runs/detect",
name="exp",
exist_ok=False,
save_img=False,
):
"""
Args:
weights: str
model.pt path(s)
source: str
file/folder, 0 for webcam
img_size: int
inference size (pixels)
conf_thres: float
object confidence threshold
iou_thres: float
IOU threshold for NMS
device: str
cuda device, i.e. 0 or 0,1,2,3 or cpu
view_img: bool
display results
save_txt: bool
save results to *.txt
save_conf: bool
save confidences in save_txt labels
classes: int
filter by class: [0], or [0, 2, 3]
agnostic-nms: bool
class-agnostic NMS
augment: bool
augmented inference
update: bool
update all models
project: str
save results to project/name
name: str
save results to project/name
exist_ok: bool
existing project/name ok, do not increment
"""
source, weights, view_img, save_txt, imgsz = (
source,
weights,
view_img,
save_txt,
img_size,
)
webcam = (
source.isnumeric()
or source.endswith(".txt")
or source.lower().startswith(("rtsp://", "rtmp://", "http://"))
)
# 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
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=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, 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, " # 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)
# 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.endswith(
'.txt') or source.lower().startswith(
('rtsp://', 'rtmp://',
'http://')) or source.lower().startswith('intel')
# Initialize
set_logging()
device = select_device(opt.device)
folder_main = out.split('/')[0]
if os.path.exists(out):
shutil.rmtree(out) # delete output folder
folder_features = folder_main + '/features'
if os.path.exists(folder_features):
shutil.rmtree(folder_features) # delete features output folder
folder_crops = folder_main + '/image_crops'
if os.path.exists(folder_crops):
shutil.rmtree(folder_crops) # delete output folder with object crops
os.makedirs(out) # make new output folder
os.makedirs(folder_features) # make new output folder
os.makedirs(folder_crops) # make new output folder
half = device.type != 'cpu' # half precision only supported on CUDA
# Load model
model = torch.load(weights[0],
map_location=device)['model'].float() # load to FP32
model.to(device).eval()
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
if source.lower().startswith('intel'):
dataset = LoadRealSense2()
save_img = True
else:
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))]
# frames per second
# TODO if use intel or if use given footage
fps = 30 # dataset.cap.get(cv2.CAP_PROP_FPS)
critical_time_frames = opt.time * fps
# COUNTER: initialization
counter = VoteCounter(critical_time_frames, fps)
print('CRITICAL TIME IS ', opt.time, 'sec, or ', counter.critical_time,
' frames')
# Find index corresponding to a person
idx_person = names.index("person")
# Deep SORT: initialize the tracker
cfg = get_config()
cfg.merge_from_file(opt.config_deepsort)
deepsort = DeepSort(cfg.DEEPSORT.REID_CKPT,
max_dist=cfg.DEEPSORT.MAX_DIST,
min_confidence=cfg.DEEPSORT.MIN_CONFIDENCE,
nms_max_overlap=cfg.DEEPSORT.NMS_MAX_OVERLAP,
max_iou_distance=cfg.DEEPSORT.MAX_IOU_DISTANCE,
max_age=cfg.DEEPSORT.MAX_AGE,
n_init=cfg.DEEPSORT.N_INIT,
nn_budget=cfg.DEEPSORT.NN_BUDGET,
use_cuda=True)
# AlphaPose: initialization
# args_p = update_config(opt.config_alphapose)
# cfg_p = update_config(args_p.ALPHAPOSE.cfg)
#
# args_p.ALPHAPOSE.tracking = args_p.ALPHAPOSE.pose_track or args_p.ALPHAPOSE.pose_flow
#
# demo = SingleImageAlphaPose(args_p.ALPHAPOSE, cfg_p, device)
# output_pose = opt.output.split('/')[0] + '/pose'
# if not os.path.exists(output_pose):
# os.mkdir(output_pose)
# 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()
# TODO => COUNTER: draw queueing ROI
# compute urn centoid (1st frame only) and plot a bounding box around it
# if dataset.frame == 1:
# counter.read_urn_coordinates(opt.urn, im0s, opt.radius)
# counter.plot_urn_bbox(im0s)
# 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
if source.lower().startswith('intel'):
p, s, im0, frame = path, '%g: ' % i, im0s[i].copy(
), dataset.count
else:
p, s, im0 = path[i], '%g: ' % i, im0s[i].copy()
else:
p, s, im0 = path, '', im0s
save_path = str(Path(out) / Path(p).name)
print(save_path)
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
# Deep SORT: person class only
idxs_ppl = (
det[:, -1] == idx_person
).nonzero(as_tuple=False).squeeze(
dim=1) # 1. List of indices with 'person' class detections
dets_ppl = det[idxs_ppl, :
-1] # 2. Torch.tensor with 'person' detections
print('\n {} people were detected!'.format(len(idxs_ppl)))
# Deep SORT: convert data into a proper format
xywhs = xyxy2xywh(dets_ppl[:, :-1]).to("cpu")
confs = dets_ppl[:, 4].to("cpu")
# Deep SORT: feed detections to the tracker
if len(dets_ppl) != 0:
trackers, features = deepsort.update(xywhs, confs, im0)
# tracks inside a critical sphere
trackers_inside = []
for i, d in enumerate(trackers):
plot_one_box(d[:-1],
im0,
label='ID' + str(int(d[-1])),
color=colors[1],
line_thickness=1)
# TODO: queue COUNTER
# d_include = counter.centroid_distance(d, im0, colors[1], dataset.frame)
# if d_include:
# trackers_inside.append(d)
# ALPHAPOSE: show skeletons for bounding boxes inside the critical sphere
# if len(trackers_inside) > 0:
# pose = demo.process('frame_'+str(dataset.frame), im0, trackers_inside)
# im0 = demo.vis(im0, pose)
# demo.writeJson([pose], output_pose, form=args_p.ALPHAPOSE.format, for_eval=args_p.ALPHAPOSE.eval)
#
# counter.save_features_and_crops(im0, dataset.frame, trackers_inside, features, folder_main)
cv2.putText(im0, 'Voted ' + str(len(counter.voters_count)),
(50, 50), cv2.FONT_HERSHEY_SIMPLEX, 1.0, (0, 0, 255),
2)
print('NUM VOTERS', len(counter.voters))
print(list(counter.voters.keys()))
# COUNTER
if len(counter.voters) > 0:
counter.save_voter_trajectory(dataset.frame, folder_main)
# 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 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
if type(vid_cap
) is dict: # estimate distance_in_meters
# TODO hard code
w, h, fps = 640, 480, 6
else:
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('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.startswith(
'rtsp') or source.startswith('http') or source.endswith('.txt')
# 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
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):
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))
# Stream results
if view_img:
# cv2.imshow(p, im0)
cv2.imwrite("C:/Users/lenovo/Desktop/server/output/camera.jpg",
im0)
if cv2.waitKey(1) == ord('q'): # q to quit
raise StopIterationq
# 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 = cv2.VideoWriter(
save_path,
cv2.VideoWriter_fourcc('X', '2', '6', '4'), 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('Done. (%.3fs)' % (time.time() - t0))
def detect(opt, device, 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')
colorOrder = ['red', 'purple', 'blue', 'green', 'yellow', 'orange']
frame_num = 0
framestr = 'Frame {frame}'
fpses = []
frame_catch_pairs = []
ball_person_pairs = {}
for color in colorDict:
ball_person_pairs[color] = 0
# Read Class Name Yaml
with open(opt.data) as f:
data_dict = yaml.load(f, Loader=yaml.FullLoader)
names = data_dict['names']
# initialize deepsort
cfg = get_config()
cfg.merge_from_file(opt.config_deepsort)
deepsort = DeepSort(cfg.DEEPSORT.REID_CKPT,
max_dist=cfg.DEEPSORT.MAX_DIST,
min_confidence=cfg.DEEPSORT.MIN_CONFIDENCE,
nms_max_overlap=cfg.DEEPSORT.NMS_MAX_OVERLAP,
max_iou_distance=cfg.DEEPSORT.MAX_IOU_DISTANCE,
max_age=cfg.DEEPSORT.MAX_AGE,
n_init=cfg.DEEPSORT.N_INIT,
nn_budget=cfg.DEEPSORT.NN_BUDGET,
use_cuda=True)
# Initialize
if os.path.exists(out):
shutil.rmtree(out) # delete output folder
os.makedirs(out) # make new output folder
# 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)
else:
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 range(len(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 = 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()
bbox_xywh = []
confs = []
clses = []
# Write results
for *xyxy, conf, cls in det:
img_h, img_w, _ = im0.shape # get image shape
x_c, y_c, bbox_w, bbox_h = bbox_rel(img_w, img_h, *xyxy)
obj = [x_c, y_c, bbox_w, bbox_h]
bbox_xywh.append(obj)
confs.append([conf.item()])
clses.append([cls.item()])
xywhs = torch.Tensor(bbox_xywh)
confss = torch.Tensor(confs)
clses = torch.Tensor(clses)
# Pass detections to deepsort
outputs = []
global groundtruths_path
if not 'disable' in groundtruths_path:
# print('\nenabled', groundtruths_path)
groundtruths = solution.load_labels(
groundtruths_path, img_w, img_h, frame_num)
if (groundtruths.shape[0] == 0):
outputs = deepsort.update(xywhs, confss, clses, im0)
else:
# print(groundtruths)
xywhs = groundtruths[:, 2:]
tensor = torch.tensor((), dtype=torch.int32)
confss = tensor.new_ones((groundtruths.shape[0], 1))
clses = groundtruths[:, 0:1]
outputs = deepsort.update(xywhs, confss, clses, im0)
if frame_num >= 2:
for real_ID in groundtruths[:, 1:].tolist():
for DS_ID in xyxy2xywh(outputs[:, :5]):
if (abs(DS_ID[0] - real_ID[1]) / img_w < 0.005
) and (abs(DS_ID[1] - real_ID[2]) / img_h <
0.005) and (
abs(DS_ID[2] - real_ID[3]) /
img_w < 0.005) and (
abs(DS_ID[3] - real_ID[4]) /
img_w < 0.005):
id_mapping[DS_ID[4]] = int(real_ID[0])
else:
outputs = deepsort.update(xywhs, confss, clses, im0)
# draw boxes for visualization
if len(outputs) > 0:
bbox_xyxy = outputs[:, :4]
identities = outputs[:, 4]
clses = outputs[:, 5]
scores = outputs[:, 6]
#Temp solution to get correct id's
mapped_id_list = []
for ids in identities:
if (ids in id_mapping):
mapped_id_list.append(int(id_mapping[ids]))
else:
mapped_id_list.append(ids)
ball_detect, frame_catch_pairs, ball_person_pairs = solution.detect_catches(
im0, bbox_xyxy, clses, mapped_id_list, frame_num,
colorDict, frame_catch_pairs, ball_person_pairs,
colorOrder, save_img)
t3 = time_synchronized()
draw_boxes(im0, bbox_xyxy, [names[i] for i in clses],
scores, ball_detect, identities)
else:
t3 = time_synchronized()
#Draw frame number
tmp = framestr.format(frame=frame_num)
t_size = cv2.getTextSize(tmp, cv2.FONT_HERSHEY_PLAIN, 2, 2)[0]
cv2.putText(im0, tmp, (0, (t_size[1] + 10)),
cv2.FONT_HERSHEY_PLAIN, 2, [255, 255, 255], 2)
#Inference Time
fps = (1 / (t3 - t1))
fpses.append(fps)
print('FPS=%.2f' % fps)
# 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
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(*opt.fourcc),
fps, (w, h))
vid_writer.write(im0)
frame_num += 1
#t4 = time_synchronized()
avgFps = (sum(fpses) / len(fpses))
print('Average FPS = %.2f' % avgFps)
#print('Total Runtime = %.2f' % (t4 - t0))
outpath = os.path.basename(source)
outpath = outpath[:-4]
outpath = out + '/' + outpath + '_out.csv'
solution.write_catches(outpath, frame_catch_pairs, colorOrder)
if save_txt or save_img:
print('Results saved to %s' % os.getcwd() + os.sep + out)
if platform == 'darwin': # MacOS
os.system('open ' + save_path)
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')
# 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))]
# Find index corresponding to a person
idx_person = names.index("person")
# SORT: initialize the tracker
mot_tracker = sort_module.Sort(max_age=opt.max_age,
min_hits=opt.min_hits,
iou_threshold=opt.iou_threshold)
# 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[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
# SORT: number of people detected
idxs_ppl = (
det[:, -1] == idx_person
).nonzero(as_tuple=False).squeeze(
dim=1) # 1. List of indices with 'person' class detections
dets_ppl = det[idxs_ppl, :-1].to(
"cpu") # 2. Torch.tensor with 'person' detections
print('\n {} people were detected!'.format(len(idxs_ppl)))
# SORT: feed detections to the tracker
if len(dets_ppl) != 0:
trackers = mot_tracker.update(dets_ppl)
for d in trackers:
plot_one_box(d[:-1],
im0,
label='ID' + str(int(d[-1])),
color=colors[1],
line_thickness=1)
# 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 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('Done. (%.3fs)' % (time.time() - t0))
def detect(opt, 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')
# initialize deepsort
cfg = get_config()
cfg.merge_from_file(opt.config_deepsort)
deepsort = DeepSort(cfg.DEEPSORT.REID_CKPT,
max_dist=cfg.DEEPSORT.MAX_DIST, min_confidence=cfg.DEEPSORT.MIN_CONFIDENCE,
nms_max_overlap=cfg.DEEPSORT.NMS_MAX_OVERLAP, max_iou_distance=cfg.DEEPSORT.MAX_IOU_DISTANCE,
max_age=cfg.DEEPSORT.MAX_AGE, n_init=cfg.DEEPSORT.N_INIT, nn_budget=cfg.DEEPSORT.NN_BUDGET,
use_cuda=True)
# Initialize
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 = torch.load(weights, map_location=device)[
'model'].float() # load to FP32
model.to(device).eval()
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:
view_img = True
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
# run once
_ = model(img.half() if half else img) if device.type != 'cpu' else None
save_path = str(Path(out))
txt_path_raw = str(Path(out)) + '/results_raw.txt'
for frame_idx, (path, img, im0s, vid_cap) in enumerate(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)
print(pred)
# 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
s += '%gx%g ' % img.shape[2:] # print string
save_path = str(Path(out) / Path(p).name)
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
bbox_xywh = []
confs = []
clss = []
# Adapt detections to deep sort input format
for *xyxy, conf, cls in det:
x_c, y_c, bbox_w, bbox_h = bbox_rel(*xyxy)
obj = [x_c, y_c, bbox_w, bbox_h]
bbox_xywh.append(obj)
confs.append([conf.item()])
clss.append(cls.item())
bbox_xywh = bbox_xywh
cls_conf = confs
cls_ids = clss
# xywhs = torch.Tensor(bbox_xywh)
# confss = torch.Tensor(confs)
# cls_ids = clss
# if len(bbox_xywh) == 0:
# continue
# print("detection cls_ids:", cls_ids)
#filter cls id for tracking
# print("cls_ids")
# print(cls_ids)
# # select class
# mask = []
# lst_move_life = [0,1,2]
# # lst_for_track = []
# for id in cls_ids:
# if id in lst_move_life:
# # lst_for_track.append(id)
# mask.append(True)
# else:
# mask.append()
# # print("mask cls_ids:", mask)
# # print(bbox_xywh)
# bbox_xywh = list(compress(bbox_xywh,mask))
# bbox dilation just in case bbox too small, delete this line if using a better pedestrian detector
# bbox_xywh[:,3:] *= 1.2
# cls_conf = list(compress(cls_conf,mask))
# print(cls_conf)
bbox_xywh = torch.Tensor(bbox_xywh)
cls_conf = torch.Tensor(cls_conf)
# Pass detections to deepsort
outputs = deepsort.update(bbox_xywh, cls_conf, im0, cls_ids)
'''
TODO:
카운터 추가 요망
'''
# counting num and class
# draw boxes for visualization
if len(outputs) > 0:
bbox_xyxy = outputs[:, :4]
identities = outputs[:, 4:5]
cls_id = outputs[:,-1]
# print(outputs[:,-1]) #--> 문제 발견
# print("track res cls_id:", cls_id)
# cls_ids_show = [cls_ids[i] for i in cls_id]
draw_boxes(im0, bbox_xyxy, cls_id, identities)
# Write MOT compliant results to file
if save_txt and len(outputs) != 0:
for j, output in enumerate(outputs):
bbox_left = output[0]
bbox_top = output[1]
bbox_w = output[2]
bbox_h = output[3]
identity = output[4]
classname = output[5]
with open(txt_path_raw, 'a') as f: # Yolov5와 DeepSort를 통하여 만들어진 첫 결과물(원본결과물)
f.write(('%g ' * 6 +'%g' *1 +'%g ' * 3 + '\n') % (frame_idx, identity, bbox_left,
bbox_top, bbox_w, bbox_h, classname, -1, -1, -1)) # label format
else:
deepsort.increment_ages()
# 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:
print('saving img!')
if dataset.mode == 'images':
cv2.imwrite(save_path, im0)
else:
print('saving 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
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(*opt.fourcc), fps, (w, h))
vid_writer.write(im0)
if save_txt or save_img:
print('Results saved to %s' % os.getcwd() + os.sep + out)
if platform == 'darwin': # MacOS
os.system('open ' + save_path)
print('Done. (%.3fs)' % (time.time() - t0))
def detect(opt, 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')
# initialize deepsort
cfg = get_config()
cfg.merge_from_file(opt.config_deepsort)
deepsort = DeepSort(cfg.DEEPSORT.REID_CKPT,
max_dist=cfg.DEEPSORT.MAX_DIST,
min_confidence=cfg.DEEPSORT.MIN_CONFIDENCE,
nms_max_overlap=cfg.DEEPSORT.NMS_MAX_OVERLAP,
max_iou_distance=cfg.DEEPSORT.MAX_IOU_DISTANCE,
max_age=cfg.DEEPSORT.MAX_AGE,
n_init=cfg.DEEPSORT.N_INIT,
nn_budget=cfg.DEEPSORT.NN_BUDGET,
use_cuda=True)
# Initialize
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 = torch.load(weights,
map_location=device)['model'].float() # load to FP32
model.to(device).eval()
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = torch_utils.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
flag = 0 # 비디오 저장시 사용할 플래그
view_img = True
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
idx = -1
compare_dict = {}
# create a new figure or activate an exisiting figure
fig = plt.figure()
ax = fig.add_subplot(111, projection='polar') # 1,1,1그리드
for path, img, im0s, vid_cap in dataset:
#plt
# Plot origin (agent's start point) - 원점=보행자
ax.plot(0, 0, color='black', marker='o', markersize=20, alpha=0.3)
# Plot configuration
ax.set_rticks([])
ax.set_rmax(1)
ax.grid(False)
ax.set_theta_zero_location("S") # 0도가 어디에 있는지-S=남쪽
ax.set_theta_direction(-1) # 시계방향 극좌표
img = torch.from_numpy(img).to(device)
# img 프레임 자르기
# '''input 이미지 프레임 자르기'''
img = img[:, 100:260, :]
temp = img
add_img = temp[:, :, :32]
img = torch.cat((img, add_img), dim=2)
# 결과 이미지 프레임 자르기
#결과 프레임 자르기 (bouding box와 object 매칭 시키기 위해!!)
im0s = im0s[200:520, :, :]
temp = im0s
add_im0s = temp[:, :64, :]
print(add_im0s.shape)
im0s = np.concatenate((im0s, add_im0s), axis=1)
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()
idx += 1
if (idx % 10 != 0): # 동영상 길이 유지
if len(outputs) > 0:
ori_im = draw_boxes(im0s, bbox_xyxy, identities,
isCloser) # 이전 정보로 bbox 그리기
vid_writer.write(im0s)
continue
# 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 #우리는 여기로 감 - 파일경로 전까지의 출력문은 datasets.py에서 삭제해야함
save_path = str(Path(out) / Path(p).name)
#print(dataset.frame) #프레임 번호
#s += '%gx%g ' % img.shape[2:] # print string #영상 사이즈 출력 (예:640x320) - 삭제가능
gn = torch.tensor(im0.shape)[[1, 0, 1,
0]] # normalization gain whwh
#만약 차량이 detect된 경우
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 #개수와 클래스 출력(예: 5 cars) -삭제가능
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
#s += '%g %ss, ' % (n, names[int(c)]) # add to string
bbox_xywh = []
confs = []
# Adapt detections to deep sort input format
for *xyxy, conf, cls in det:
img_h, img_w, _ = im0.shape #결과프레임의 사이즈
x_c, y_c, bbox_w, bbox_h = bbox_rel(img_w, img_h,
*xyxy) #center좌표, w, h
obj = [x_c, y_c, bbox_w, bbox_h]
bbox_xywh.append(obj)
confs.append([conf.item()])
xywhs = torch.Tensor(bbox_xywh)
confss = torch.Tensor(confs)
# Pass detections to deepsort
outputs = []
if len(bbox_xywh) != 0: #뭔가 detect됐다면 deepsort로 보냄
outputs, coors, frame, bbox_size = deepsort.update(
xywhs, confss, im0, compare_dict, dataset.frame)
# draw boxes for visualization
if len(outputs) > 0:
print("!", outputs)
bbox_xyxy = outputs[:, :4]
identities = outputs[:, 4]
isCloser = outputs[:, -1]
print("isCloser:", isCloser)
print(compare_dict)
ori_im = draw_boxes(im0, bbox_xyxy, identities,
isCloser) # bbox 그리기
alert.show_direction(ax, coors, bbox_size,
isCloser) # 방향 display하는 함수 호출
# Print time (inference + NMS)
#print('%sDone. (%.3fs)' % (s, t2 - t1))
plt.show(block=False)
# plot 차트 저장
# idx가 10이면 1부터 9까지 10과 같은 이미지로 저장
'''
file = '/Users/wonyeong/Desktop/ewha/project/plotimgs/img%d.png' % idx
plt.savefig(file)
for j in range(9):
file = '/Users/wonyeong/Desktop/ewha/project/plotimgs/img%d.png' % (j + idx + 1)
plt.savefig(file)
# 차량이 detect된 경우에만 그린다..
'''
plt.pause(0.01)
plt.cla()
# Stream results
cv2.imshow('frame', im0)
if cv2.waitKey(1) & 0xFF == ord('q'):
break
print(im0s.shape)
# Save results (image with detections)
if save_img:
if dataset.mode == 'images':
cv2.imwrite(save_path, im0)
else:
if (flag == 0):
vid_path = save_path
if isinstance(vid_writer, cv2.VideoWriter):
vid_writer.release(
) # release previous video writer
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))
w = 1344
h = 320
flag = 1
vid_writer = cv2.VideoWriter(
save_path, cv2.VideoWriter_fourcc(*opt.fourcc),
fps, (w, h))
else:
vid_writer.write(im0)
if save_txt or save_img:
print('Results saved to %s' % os.getcwd() + os.sep + out)
if platform == 'darwin': # MacOS
os.system('open ' + save_path)
print('Done. (%.3fs)' % (time.time() - t0))
def detect(opt, 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')
# initialize deepsort
cfg = get_config()
cfg.merge_from_file(opt.config_deepsort)
deepsort = DeepSort(cfg.DEEPSORT.REID_CKPT,
max_dist=cfg.DEEPSORT.MAX_DIST,
min_confidence=cfg.DEEPSORT.MIN_CONFIDENCE,
nms_max_overlap=cfg.DEEPSORT.NMS_MAX_OVERLAP,
max_iou_distance=cfg.DEEPSORT.MAX_IOU_DISTANCE,
max_age=cfg.DEEPSORT.MAX_AGE,
n_init=cfg.DEEPSORT.N_INIT,
nn_budget=cfg.DEEPSORT.NN_BUDGET,
use_cuda=True)
# Initialize
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
# Read Yaml
with open(opt.data) as f:
data_dict = yaml.load(f, Loader=yaml.FullLoader)
names = data_dict['names']
print(names)
# Load model
#google_utils.attempt_download(weights)
model = torch.load(weights,
map_location=device)['model'].float() # load to FP32
#model = torch.save(torch.load(weights, map_location=device), weights) # update model if SourceChangeWarning
# model.fuse()
model.to(device).eval()
if half:
model.half() # to FP16
# Second-stage classifier
classify = False
if classify:
modelc = torch_utils.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
t_fps = time_synchronized()
frame_fps = 0
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[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
#print(det[:, -1].unique())
for c in det[:, -1].unique():
n = (det[:, -1] == c).sum() # detections per class
s += '%g %ss, ' % (n, names[int(c)]) # add to string
bbox_xywh = []
confs = []
#clses = []
# Write results
for *xyxy, conf, cls in det:
img_h, img_w, _ = im0.shape # get image shape
x_c, y_c, bbox_w, bbox_h = bbox_rel(img_w, img_h, *xyxy)
obj = [x_c, y_c, bbox_w, bbox_h]
bbox_xywh.append(obj)
confs.append([conf.item()])
#clses.append([cls.item()])
#outputs, clses2 = deepsort.update((torch.Tensor(bbox_xywh)), (torch.Tensor(confs)), (torch.Tensor(clses)), im0)
#outputs = deepsort.update((torch.Tensor(bbox_xywh)), (torch.Tensor(confs)), im0)
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)
xywhs = torch.Tensor(bbox_xywh)
confss = torch.Tensor(confs)
# Pass detections to deepsort
outputs = deepsort.update(xywhs, confss, im0)
# draw boxes for visualization
if len(outputs) > 0:
bbox_tlwh = []
bbox_xyxy = outputs[:, :4]
identities = outputs[:, -1]
ori_im = draw_boxes(im0, bbox_xyxy, identities)
# Print time (inference + NMS)
#print('%sDone. (%.3fs)' % (s, t2 - t1))
print('%sDone.' % s)
# 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
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(*opt.fourcc),
fps, (w, h))
vid_writer.write(im0)
frame_fps += 1
if ((time_synchronized() - t_fps) >= 1):
print('\n')
print('FPS=%.2f' % (frame_fps))
t_fps = time_synchronized()
frame_fps = 0
#print('FPS=%.2f' % (1/(time_synchronized() - t1)))
if save_txt or save_img:
print('Results saved to %s' % os.getcwd() + os.sep + out)
if platform == 'darwin': # MacOS
os.system('open ' + save_path)
def detect(save_img=False):
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://'))
# 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)
else:
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, 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)
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)
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(opt, save_img=False):
out, source, weights, view_img, save_txt, imgsz, GCP_list = \
opt.output, opt.source, opt.weights, opt.view_img, opt.save_txt, opt.img_size, opt.GCP_list
webcam = source == '0' or source.startswith('rtsp') or source.startswith(
'http') or source.endswith('.txt')
# initialize deepsort
cfg = get_config()
cfg.merge_from_file(opt.config_deepsort)
deepsort = DeepSort(cfg.DEEPSORT.REID_CKPT,
max_dist=cfg.DEEPSORT.MAX_DIST,
min_confidence=cfg.DEEPSORT.MIN_CONFIDENCE,
nms_max_overlap=cfg.DEEPSORT.NMS_MAX_OVERLAP,
max_iou_distance=cfg.DEEPSORT.MAX_IOU_DISTANCE,
max_age=cfg.DEEPSORT.MAX_AGE,
n_init=cfg.DEEPSORT.N_INIT,
nn_budget=cfg.DEEPSORT.NN_BUDGET,
use_cuda=True)
# Initialize
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 = torch.load(weights,
map_location=device)['model'].float() # load to FP32
model.to(device).eval()
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:
view_img = True
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
# run once
_ = model(img.half() if half else img) if device.type != 'cpu' else None
save_path = str(Path(out))
txt_path_raw = str(Path(out)) + '/results_raw.txt'
# 속도까지 붙여버린 데이터 따로 생성해서 비교해보자 : 수정수정
txt_path_raw2 = str(Path(out)) + '/results_raw2.txt'
# point load
with open('./mapdata/point.yaml') as f:
data = yaml.load(f.read())
frm_point = data['frm_point']
geo_point = data['geo_point']
Counter_1 = [(488, 589), (486, 859)]
Counter_2 = [(3463, 795), (3487, 1093)]
Counter_list = [Counter_1, Counter_2]
datum_dist = []
counter_dist = []
line_fileName = './mapdata/Busan1_IC_Polyline_to_Vertex.txt'
all_line = mapdata_load(line_fileName, frm_point, geo_point)
percep_frame = 5
from _collections import deque
pts = [deque(maxlen=percep_frame + 1) for _ in range(10000)]
ptsSpeed = [deque(maxlen=1) for _ in range(10000)]
frame_len = calc_dist(frm_point[1], frm_point[4])
geo_len = calc_dist(geo_point[1], geo_point[4])
# ----------------- fix val start
fixcnt = 1
# ----------------- fix val end
# ----------------- counter val start
memory_index = {}
memory_id = {}
cnt = np.zeros((len(Counter_list), 4))
# total_counter = 0 # 나중에 총 카운터를 만들어 넣으면 되겠지?
# count_1_total = 0
# count_1_veh_c0 = 0
# count_1_veh_c1 = 0
# count_1_veh_c2 = 0
# count_2_total = 0
# count_2_veh_c0 = 0
# count_2_veh_c1 = 0
# count_2_veh_c2 = 0
# ----------------- counter val end
for frame_idx, (path, img, im0s, vid_cap) in enumerate(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)
# Apply Classifier
if classify:
pred = apply_classifier(pred, modelc, img, im0s)
print(pred)
t2 = time_synchronized()
# 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
s += '%gx%g ' % img.shape[2:] # print string
save_path = str(Path(out) / Path(p).name)
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
bbox_xywh = []
confs = []
clss = []
# Adapt detections to deep sort input format
for *xyxy, conf, cls in det:
x_c, y_c, bbox_w, bbox_h = bbox_rel(*xyxy)
obj = [x_c, y_c, bbox_w, bbox_h]
bbox_xywh.append(obj)
confs.append([conf.item()])
clss.append(cls.item())
bbox_xywh = bbox_xywh
cls_conf = confs
cls_ids = clss
# xywhs = torch.Tensor(bbox_xywh)
# confss = torch.Tensor(confs)
# cls_ids = clss
# if len(bbox_xywh) == 0:
# continue
# print("detection cls_ids:", cls_ids)
#filter cls id for tracking
# print("cls_ids")
# print(cls_ids)
# # select class
# mask = []
# lst_move_life = [0,1,2]
# # lst_for_track = []
# for id in cls_ids:
# if id in lst_move_life:
# # lst_for_track.append(id)
# mask.append(True)
# else:
# mask.append()
# # print("mask cls_ids:", mask)
# # print(bbox_xywh)
# bbox_xywh = list(compress(bbox_xywh,mask))
# bbox dilation just in case bbox too small, delete this line if using a better pedestrian detector
# bbox_xywh[:,3:] *= 1.2
# cls_conf = list(compress(cls_conf,mask))
# print(cls_conf)
bbox_xywh = torch.Tensor(bbox_xywh)
cls_conf = torch.Tensor(cls_conf)
# Pass detections to deepsort
outputs = deepsort.update(bbox_xywh, cls_conf, im0, cls_ids)
"""
# output 형식
[[박스 좌측상단 x, 박스 좌측상단 y, 박스 우측하단 x, 박스 우측하단 y, 차량 id, 클래스 넘버],
[박스 좌측상단 x, 박스 좌측상단 y, 박스 우측하단 x, 박스 우측하단 y, 차량 id, 클래스 넘버],
[박스 좌측상단 x, 박스 좌측상단 y, 박스 우측하단 x, 박스 우측하단 y, 차량 id, 클래스 넘버],
[박스 좌측상단 x, 박스 좌측상단 y, 박스 우측하단 x, 박스 우측하단 y, 차량 id, 클래스 넘버],
...]
"""
# ------------------------------------------------------------------------------------------------------ img fix start
t3 = time_synchronized()
match_mid_point_list = matcher_BRISK_BF(im0, GCP_list)
t4 = time_synchronized()
# ---------------------------------------------------------------------------------------------------------------------- line start
# 기준점 위치 갱신을 위한 삼변측량의 거리 정의 및 고정
if frame_idx == 0:
for pointNum in range(len(frm_point)):
for GCP_num in range(len(match_mid_point_list)):
datum_dist.append(
point_dist(match_mid_point_list[GCP_num],
frm_point[pointNum]))
datum_dist = np.reshape(
datum_dist,
(len(frm_point), len(match_mid_point_list)))
for Ct_list in Counter_list:
for Ctpoint_num in range(len(Ct_list)):
for GCP_num in range(len(match_mid_point_list)):
counter_dist.append(
point_dist(match_mid_point_list[GCP_num],
Ct_list[Ctpoint_num]))
counter_dist = np.reshape(counter_dist,
(len(Counter_list), len(Ct_list),
len(match_mid_point_list)))
t5 = time_synchronized()
pre_P = (0, 0)
for line_num, eachline in enumerate(all_line):
for newpoint in eachline['frmPoint']:
if line_num == 0:
im0 = cv2.circle(im0, newpoint, 5, (0, 0, 255),
-1) # 차선_실선
if calc_dist(pre_P, newpoint) < 390:
im0 = cv2.line(im0, pre_P, newpoint,
(0, 0, 255), 2, -1)
elif line_num == 1:
im0 = cv2.circle(im0, newpoint, 5, (0, 255, 0),
-1) # 도로 경계
if calc_dist(pre_P, newpoint) < 420:
im0 = cv2.line(im0, pre_P, newpoint,
(0, 255, 0), 2, -1)
elif line_num == 2:
im0 = cv2.circle(im0, newpoint, 5, (255, 0, 0),
-1) # 차선_겹선
if calc_dist(pre_P, newpoint) < 350:
im0 = cv2.line(im0, pre_P, newpoint,
(255, 0, 0), 2, -1)
else:
im0 = cv2.circle(im0, newpoint, 5, (100, 100, 0),
-1) # 차선_점선
if calc_dist(pre_P, newpoint) < 600:
im0 = cv2.line(im0, pre_P, newpoint,
(100, 100, 0), 2, -1)
pre_P = newpoint
t6 = time_synchronized()
for pointNum in range(len(frm_point)):
im0 = cv2.circle(im0, frm_point[pointNum], 10, (0, 0, 0),
-1)
newPoint = intersectionPoint(match_mid_point_list,
datum_dist[pointNum])
frm_point[pointNum] = newPoint
t7 = time_synchronized()
#---------------------------------------------------------------------------------------------------------------------- line end
# ------------------------------------------------------------------------------------------------------ img fix end
# ------------------------------------------------------------------------------------------------------ counting num and class start
Counter_newpoint = []
for Ct_num in range(len(Counter_list)):
Ct_list = Counter_list[Ct_num]
for Ctpoint_num in range(len(Ct_list)):
Counter_newpoint.append(
intersectionPoint(
match_mid_point_list,
counter_dist[Ct_num][Ctpoint_num]))
Counter_newpoint = np.reshape(
Counter_newpoint, (len(Counter_list), len(Ct_list), 2))
for CountNum in Counter_newpoint:
im0 = cv2.line(im0, tuple(CountNum[0]), tuple(CountNum[1]),
(0, 0, 0), 5, -1)
boxes = []
indexIDs = []
classIDs = []
previous_index = memory_index.copy()
previous_id = memory_id.copy()
memory_index = {}
memory_id = {}
COLORS = np.random.randint(0,
255,
size=(200, 3),
dtype="uint8")
if save_txt and len(outputs) != 0:
for j, output in enumerate(outputs):
boxes.append(
[output[0], output[1], output[2], output[3]])
indexIDs.append(int(output[4]))
classIDs.append(int(output[5]))
memory_index[indexIDs[-1]] = boxes[
-1] # 인덱스 아이디와 박스를 맞춰줌
memory_id[indexIDs[-1]] = classIDs[
-1] # 인덱스 아이디와 클레스 아이디를 맞춰줌
if len(pts[output[4]]) == 0:
pts[output[4]].append(frame_idx)
center = (int(((output[0]) + (output[2])) / 2),
int(((output[1]) + (output[3])) / 2))
pts[output[4]].append(center)
if len(pts[output[4]]) == percep_frame + 1:
frmMove_len = np.sqrt(
pow(
pts[output[4]][-1][0] -
pts[output[4]][-percep_frame][0], 2) + pow(
pts[output[4]][-1][1] -
pts[output[4]][-percep_frame][1], 2))
geoMove_Len = geo_len * frmMove_len / frame_len
speed = geoMove_Len * vid_cap.get(
cv2.CAP_PROP_FPS) * 3.6 / (pts[output[4]][0] -
frame_idx)
ptsSpeed[output[4]].append(speed)
pts[output[4]].clear()
if len(boxes) > 0:
i = int(0)
for box in boxes:
# 현 위치와 이전 위치를 비교하여 지나갔는지 체크함
(x, y) = (int(box[0]), int(box[1])) # Output 0 1
(w, h) = (int(box[2]), int(box[3])) # Output 2 3 과 같다.
color = compute_color_for_labels(indexIDs[i])
if indexIDs[i] in previous_index:
previous_box = previous_index[indexIDs[i]]
# print()
# print('previous_box : ')
# print(previous_box)
(x2, y2) = (int(previous_box[0]),
int(previous_box[1]))
(w2, h2) = (int(previous_box[2]),
int(previous_box[3]))
p0 = (int(x + (w - x) / 2), int(y + (h - y) / 2)
) # 현재 박스
p1 = (int(x2 + (w2 - x2) / 2),
int(y2 + (h2 - y2) / 2)) # 이전 박스
cv2.line(
im0, p0, p1, color, 3
) # 이전 정보와 비교 : 중앙에 점을 찍어 가면서 (이전 데이터와 검지 데이터의 점)
# 클레스 구분
previous_class_id = previous_id[
indexIDs[i]] # 어차피 인덱스 같기 때문에 그냥 넣어줘도 됨 개꿀ㅋ
# Yolov5와 DeepSort를 통하여 만들어진 첫 결과물(내가 맨든 결과물)
# 프레임 수, 인덱스 아이디, 클레스 이름, x좌표, y좌표, w값, h값, 속도값, null, null
# with open(txt_path_raw2, 'a') as f:
# f.write(('%g ' * 10+ '\n') % (frame_idx, indexIDs[i], previous_class_id,
# p0[0], p0[1], box[2], box[3], -1, -1)) # label format
for cntr in range(len(Counter_newpoint)):
if intersect(p0, p1, Counter_newpoint[cntr][0],
Counter_newpoint[cntr]
[1]): # 실질적으로 체크함
if previous_class_id == 0:
cnt[cntr][1] += 1
elif previous_class_id == 1:
cnt[cntr][2] += 1
elif previous_class_id == 2:
cnt[cntr][3] += 1
cnt[cntr][0] += 1
i += 1 # 다음 인덱스와 비교하게 만들기 위하여
# draw counter
for cntr in range(len(Counter_newpoint)):
cv2.putText(im0, 'count_{}_total : {}'.format(
cntr + 1, cnt[cntr][0]), (100 + 400 * cntr, 110),
cv2.FONT_HERSHEY_DUPLEX, 1.0, (0, 0, 0),
2) # 카운팅 되는거 보이게
cv2.putText(im0, 'count_{}_{} : {}'.format(
cntr + 1, names[0],
cnt[cntr][1]), (100 + 400 * cntr, 140),
cv2.FONT_HERSHEY_DUPLEX, 0.5, (0, 0, 0),
2) # 카운팅 되는거 보이게
cv2.putText(im0, 'count_{}_{} : {}'.format(
cntr + 1, names[1],
cnt[cntr][2]), (100 + 400 * cntr, 170),
cv2.FONT_HERSHEY_DUPLEX, 0.5, (0, 0, 0),
2) # 카운팅 되는거 보이게
cv2.putText(im0, 'count_{}_{} : {}'.format(
cntr + 1, names[2],
cnt[cntr][3]), (100 + 400 * cntr, 200),
cv2.FONT_HERSHEY_DUPLEX, 0.5, (0, 0, 0),
2) # 카운팅 되는거 보이게
t8 = time_synchronized()
# ---------------------------------------------------------------------------------------------------------------------- counter end
# draw boxes for visualization
if len(outputs) > 0:
bbox_xyxy = outputs[:, :4]
identities = outputs[:, 4:5]
cls_id = outputs[:, -1]
draw_boxes(im0, bbox_xyxy, cls_id, identities, names,
ptsSpeed)
t9 = time_synchronized()
# Write MOT compliant results to file
if save_txt and len(outputs) != 0:
for j, output in enumerate(outputs): # 한 라인씩 쓰는 구조
bbox_left = output[0]
bbox_top = output[1]
bbox_w = output[2]
bbox_h = output[3]
identity = output[4]
classname = output[5]
with open(
txt_path_raw, 'a'
) as f: # Yolov5와 DeepSort를 통하여 만들어진 첫 결과물(원본결과물)
f.write(('%g ' * 6 + '%g' * 1 + '%g ' * 3 + '\n') %
(frame_idx, identity, bbox_left, bbox_top,
bbox_w, bbox_h, classname, -1, -1,
-1)) # label format
# else:
# deepsort.increment_ages()
t10 = time_synchronized()
# Print time (inference + NMS + classify)
#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
t11 = time_synchronized()
# Save results (image with detections)
# dataset.mode = 'images'
# save_path = './track_result/output/{}.jpg'.format(i)
if save_img:
print('saving img!')
if dataset.mode == 'images':
cv2.imwrite(save_path, im0)
else:
print('saving 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
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(*opt.fourcc),
fps, (w, h))
vid_writer.write(im0)
t12 = time_synchronized()
print('inference + NMS + classify (%.3fs)' % (t2 - t1))
print('Yolo + DeepSORT (%.3fs)' % (t3 - t2))
print('find mid point (%.3fs)' % (t4 - t3))
print('삼변측량을 위한 기준거리 산정 (%.3fs)' % (t5 - t4))
print('draw line (%.3fs)' % (t6 - t5)
) # 현재는 정밀도로지도에 있는 모든 점들을 대상 계산중 -> 추후 화면에 표시될 점만 계산하는 작업 필요
print('GCP 점 계산 (%.3fs)' % (t7 - t6))
print('Count & speed (%.3fs)' % (t8 - t7))
print('각차량별 그리기 (%.3fs)' % (t9 - t8))
print('txt 데이터 저장 (%.3fs)' % (t10 - t9))
print('스크린에 표시하기 (%.3fs)' % (t11 - t10))
print('비디오파일로 저장하기 (%.3fs)' % (t12 - t11))
print('one frame done (%.3fs)' % (t12 - t1))
if save_txt or save_img:
print('Results saved to %s' % os.getcwd() + os.sep + out)
if platform == 'darwin': # MacOS
os.system('open ' + save_path)
print('Done. (%.3fs)' % (time.time() - t0))