def main():
args = parse_args()
if mmdet.CPU_ONLY:
device = torch.device('cpu')
else:
device = torch.device('cuda', args.device)
model = init_detector(args.config, args.checkpoint, device=device)
camera = cv2.VideoCapture(args.camera_id)
print('Press "Esc", "q" or "Q" to exit.')
while True:
ret_val, img = camera.read()
result = inference_detector(model, img)
ch = cv2.waitKey(1)
if ch == 27 or ch == ord('q') or ch == ord('Q'):
break
show_result(img,
result,
model.CLASSES,
score_thr=args.score_thr,
wait_time=1)
def main():
#args = parse_args()
'''model = init_detector(
args.config, args.checkpoint, device=torch.device('cuda', args.device))
'''
model = init_detector(
'../configs/cityscapes/mask_rcnn_r50_fpn_1x_cityscapes.py',
'../checkpoints/mask_rcnn_r50_fpn_1x_city_20190727-9b3c56a5.pth',
device=torch.device('cuda', 0))
modelseg = ESPNet_Encoder(3, p=2, q=3).to(device)
modelseg.load_state_dict(
torch.load('bdd_weights_20_ESPNET_road.pth', map_location='cpu'))
modelseg.eval()
#camera = cv2.VideoCapture(args.camera_id)
camera = cv2.VideoCapture('umgt.avi')
print('Press "Esc", "q" or "Q" to exit.')
if camera.isOpened():
while True:
ret_val, img = camera.read()
imgroad = RoadSeg(img, modelseg)
#imgroad = evaluateModel(img,modelseg)
result = inference_detector(model, img)
ch = cv2.waitKey(1)
if ch == 27 or ch == ord('q') or ch == ord('Q'):
break
#show_result(img, result, model.CLASSES, score_thr=args.score_thr, wait_time=1)
show_result(imgroad,
result,
model.CLASSES,
score_thr=0.5,
wait_time=1)
cv2.destroyAllWindows()
def main(show_one_pic=False):
model = init_detector(config_file, checkpoint_file)
if show_one_pic:
img = 'data/coco/train2017/IMG_9372.MOV0035.jpg'
result = inference_detector(model, img)
print(result)
show_result(img, result, model.CLASSES)
return
# fourcc = cv2.VideoWriter_fourcc(*'XVID')
# out = cv2.VideoWriter('out.avi', fourcc, 25.0, (1280, 720))
camera = cv2.VideoCapture('/home/rex/mmdetection/IMG_9372~1.avi')
print('Press "Esc", "q" or "Q" to exit.')
while True:
ret_val, img = camera.read()
result = inference_detector(model, img)
ch = cv2.waitKey(1)
if ch == 27 or ch == ord('q') or ch == ord('Q'):
break
# frame = show_result(img, result, ('wing', 'drumstick', 'pai', 'tart', 'none'), score_thr=0.35, wait_time=1,
# show=False)
show_result(img, result, model.CLASSES, score_thr=0.35, wait_time=1)
def save_det_result_bbox(config_file,
out_dir,
checkpoint_file=None,
img_dir=None):
cfg = Config.fromfile(config_file)
data_test = cfg.data.test
dataset = build_dataset(data_test)
classnames = dataset.CLASSES
# import ipdb;ipdb.set_trace()
# use checkpoint path in cfg
if not checkpoint_file:
checkpoint_file = osp.join(cfg.work_dir, 'latest.pth')
# use testset in cfg
if not img_dir:
img_dir = data_test.img_prefix
model = init_detector(config_file, checkpoint_file, device='cuda:0')
img_list = os.listdir(img_dir)
for img_name in img_list:
img_path = osp.join(img_dir, img_name)
img_out_path = osp.join(out_dir, img_name)
result = inference_detector(model, img_path)
show_result(img_path,
result,
model.CLASSES,
show=False,
score_thr=0.2,
out_file=img_out_path)
print(img_out_path)
def main():
args = parse_args()
model = init_detector(args.config,
args.checkpoint,
device=torch.device('cuda', args.device))
camera = cv2.VideoCapture(args.camera)
print('Press "Esc", "q" or "Q" to exit.')
while True:
ret_val, img = camera.read()
tic = time.time()
result = inference_detector(model, img)
cost = time.time() - tic
print('cost: {}, fps: {}'.format(cost, 1 / cost))
ch = cv2.waitKey(1)
if ch == 27 or ch == ord('q') or ch == ord('Q'):
break
show_result(img,
result,
model.CLASSES,
score_thr=args.score_thr,
wait_time=1)
def main():
parser = ArgumentParser()
#parser.add_argument('img_dir',help='Image file Dir')
parser.add_argument('config', help='Config file')
parser.add_argument('checkpoint', help='Checkpoint file')
parser.add_argument(
'--device', default='cuda:0', help='Device used for inference')
parser.add_argument(
'--score-thr', type=float, default=0.7, help='bbox score threshold')
args = parser.parse_args()
import os
img_dir = r"/home/amax/anaconda3/envs/mmdetection/mmdetection/demo/demo_image"
out_dir = "/home/amax/anaconda3/envs/mmdetection/mmdetection/demo/"+args.config[8:-3]
if(os.path.exists(out_dir) is None):
os.makedirs(out_dir)
for img_name in os.listdir(img_dir):
img_path = img_dir + "/" +img_name
out_file = out_dir + "/" + img_name
img = cv2.imread(img_path)
# build the model from a config file and a checkpoint file
model = init_detector(args.config, args.checkpoint, device=args.device)
# test a single image
result = inference_detector(model, img)
# show the results
#show_result_pyplot(img, result, model.CLASSES,score_thr=args.score_thr)
show_result(
img, result, model.CLASSES, score_thr=args.score_thr, wait_time=1,out_file=out_file)
def main():
parser = ArgumentParser()
parser.add_argument('img', help='Image file')
parser.add_argument('config', help='Config file')
parser.add_argument('checkpoint', help='Checkpoint file')
parser.add_argument('--device',
default='cuda:0',
help='Device used for inference')
parser.add_argument('--score-thr',
type=float,
default=0.3,
help='bbox score threshold')
args = parser.parse_args()
# build the model from a config file and a checkpoint file
model = init_detector(args.config, args.checkpoint, device=args.device)
# test a single image
result = inference_detector(model, args.img)
# show the results
show_result(
args.img,
result,
model.CLASSES,
score_thr=args.score_thr,
)
def verify_images():
im_list, out_dir = get_imlist2()
detections = {}
for imf in tqdm(im_list, total=len(im_list)):
im = cv2.imread(imf)
reiszed_im = cv2.resize(im, (1280, 800))
inputs = preprocess(reiszed_im)
result = model(inputs.cuda(), return_loss=False, try_dummy=False)
labels = [i * np.ones((len(r), 1)) for i, r in enumerate(result, 1)]
bboxes = np.vstack(result)
labels = np.vstack(labels)
# nms for all classes
bboxes, idx = nms(bboxes, 0.7)
labels = labels[idx]
# detection above score threshold
idx = bboxes[:, -1] > thresh
bboxes = bboxes[idx]
labels = labels[idx]
detections[imf] = np.hstack([bboxes, labels])
continue
out_file = os.path.join(out_dir, os.path.basename(imf))
show_result(reiszed_im,
result,
class_names,
score_thr=thresh,
out_file=out_file)
out_file = os.path.join(out_dir, 'detections.pkl')
with open(out_file, 'wb') as f:
pkl.dump(detections, f)
def validate_epoch(config_path, work_dir, epoch_path, pathImg, csvPath):
model = init_detector(config_path, epoch_path, device='cuda:4')
resultCsv = open(csvPath, "a") # 打开result csv文件，检测的结果将存放在这里面，需要提前新建好
writer = csv.writer(resultCsv)
# 取testImg文件夹中的每一张图片
#pathSaveImg = csvPath.split('/')[-1][0:-4]
pathSaveImg = os.path.join(work_dir, 'val_result',
epoch_path.split('/')[-1][:-4])
if (os.path.exists(pathSaveImg) != 1):
os.makedirs(pathSaveImg)
for imgP in os.listdir(pathImg):
img = os.path.join(pathImg,
imgP) #args.pathImg + '/' + imgP # img的绝对路径
#print("#######img\n",img)
# 将img传给model，产生输出
# test a single image
result = inference_detector(model, img)
# save the visualization results to image files
csvOut = []
#show_result(img, result, model.CLASSES,score_thr=0.0001, show=False, out_file=args.work_dir + '/result_image/' + imgP, out_csv=csvOut)
show_result(img, result, model.CLASSES, score_thr=0.0001, show=False,\
out_file=os.path.join(pathSaveImg,imgP), out_csv=csvOut)
# 将结果写进csv,result array中保存了很多bbox，一次只写一个
for aBbox in csvOut:
# 产成大于阈值的list并将该list命名为record
writer.writerow(aBbox)
# 关闭打开的两个csv
resultCsv.close()
def verify_video():
out_dir = '/private/ningqingqun/results/centernet_results/senyun'
if not os.path.isdir(out_dir):
os.makedirs(out_dir)
vfile = '/private/ningqingqun/datasets/videos/WIN_20191030_20_38_42_Pro.mp4'
cap = cv2.VideoCapture(vfile)
total_frame = cap.get(cv2.CAP_PROP_FRAME_COUNT)
# print(total_frame)
display_num = 20
frame_count = 0
base_frame_index = random.randint(0, total_frame - display_num)
cap.set(cv2.CAP_PROP_POS_FRAMES, base_frame_index)
print('process from index: {}'.format(base_frame_index))
while (cap.isOpened()):
ret, frame = cap.read()
reiszed_im = cv2.resize(frame, (1280, 800))
inputs = preprocess(reiszed_im)
result = model(inputs.cuda(), return_loss=False, try_dummy=False)
out_file = os.path.join(
out_dir, '{:04}.jpg'.format(base_frame_index + frame_count))
show_result(reiszed_im,
result,
class_names,
score_thr=thresh,
out_file=out_file)
frame_count += 1
if frame_count > display_num:
break
cap.release()
def mock_detector(model, image_name, output_dir):
image = cv2.imread(image_name)
results = inference_detector(model, image)
basename = os.path.basename(image_name).split('.')[0]
result_name = basename + "_result.jpg"
result_name = os.path.join(output_dir, result_name)
show_result(image, results, model.CLASSES, out_file=result_name)
def main():
fcsv = open('/DATA/zhanghui/underwater-obj_de/submit/test_yb.csv',
'w',
encoding='utf-8')
csv_writer = csv.writer(fcsv)
csv_writer.writerow(
["name", "image_id", "confidence", "xmin", "ymin", "xmax", "ymax"])
imgpath = '/DATA/zhanghui/underwater-obj_de/data/test-A-image/'
num = 0
# for imgname in glob.glob(imgpath + '000003.jpg'):
imgname = '/DATA/zhanghui/underwater-obj_de/data/test-A-image/000003.jpg'
if not os.path.exists('/DATA/zhanghui/underwater-obj_de/data/temp_path'):
os.mkdir('/DATA/zhanghui/underwater-obj_de/data/temp_path')
else:
files = glob.glob(
'/DATA/zhanghui/underwater-obj_de/data/temp_path/*.jpg')
for f in files:
os.remove(f)
result = cut_save_img(imgname)
end_result = result_sort(result)
show_result(imgname, end_result, model.CLASSES)
# save_csv(csv_writer, imgname, end_result)
print(num)
num = num + 1
fcsv.close()
def draw_boxes(input_dir,
output_dir,
config_file,
checkpoint_file,
threshold=0.5):
if not os.path.exists(output_dir):
os.mkdir(output_dir)
images = [
f for f in os.listdir(input_dir)
if os.path.isfile(os.path.join(input_dir, f)) and f.endswith('.jpg')
]
# build the model from a config file and a checkpoint file
model = init_detector(config_file, checkpoint_file, device='cuda:0')
progress_bar = mmcv.ProgressBar(len(images))
for img in images:
input_img = os.path.join(input_dir, img)
result = inference_detector(model, input_img)
show_result(input_img,
result,
model.CLASSES,
show=False,
out_file=os.path.join(output_dir, img),
score_thr=threshold)
progress_bar.update()
def detect_masks(img, out_file):
result = inference_detector(model, img)
show_result(img,
result,
model.CLASSES,
score_thr=score_thr,
show=False,
out_file=out_file)
def main():
args = parse_args()
model = init_detector(
args.config, args.checkpoint, device=torch.device('cuda', args.device))
# camera = cv2.VideoCapture(args.camera_id)
camera = LoadStreams(args.camera_id)
# tracker
tracker = Sort(10, 4)
frame_rate = 0
start_time = time.time()
frame_count = 0
print('Press "Esc", "q" or "Q" to exit.')
while True:
# ret_val, img = camera.read()
img = camera.__next__()[0]
result = inference_detector(model, img)
if isinstance(result, tuple):
bbox_result, segm_result = result
else:
bbox_result, segm_result = result, None
bboxes = np.vstack(bbox_result)
bboxes = bboxes[bboxes[:, 4] > args.score_thr]
trackers = tracker.update(bboxes)
trackers = trackers.astype(np.int32)
for d in trackers:
cv2.rectangle(img, (d[0], d[1]), (d[2], d[3]), (255, 255, 255))
cv2.putText(img, '{}'.format(d[4]), (d[0], d[1]),
cv2.FONT_HERSHEY_SIMPLEX, 1, (255, 255, 255))
ch = cv2.waitKey(1)
if ch == 27 or ch == ord('q') or ch == ord('Q'):
break
end_time = time.time()
if (end_time - start_time) >= 1:
frame_rate = int(frame_count / (end_time - start_time))
start_time = time.time()
frame_count = 0
cv2.putText(img, str(frame_rate) + " fps", (10, 30),
cv2.FONT_HERSHEY_SIMPLEX, 1, (0, 255, 0),
thickness=2, lineType=2)
frame_count += 1
show_result(
img, result, model.CLASSES, score_thr=args.score_thr, wait_time=1)
def detect_images(config_file, checkpoint_file, imgs_path):
#imgs_path 包含image 路径的txt文件
f = open(img_path)
imgs_list = f.readlines(imgs_path)
f.close()
# test a list of images and write the results to image files
for i, result in enumerate(inference_detector(model, imgs_list)):
show_result(imgs[i],
result,
model.CLASSES,
out_file='result_{}.jpg'.format(i))
def verify_image():
imf = '/private/ningqingqun/bags/crane/howo1_2019_11_05_09_06_22_6.msg/front_right/201911050907/201911050907_00000417_1572916023857.jpg'
im = cv2.imread(imf)
reiszed_im = cv2.resize(im, (1280, 800))
inputs = preprocess(reiszed_im)
result = model(inputs.cuda(), return_loss=False, try_dummy=False)
show_result(reiszed_im,
result,
class_names,
score_thr=thresh,
out_file='centernet_verify.jpg')
def main():
args = parse_args()
model = init_detector(args.config, args.checkpoint, device='cuda:0')
result = inference_detector(model, args.input)
result = result[:-1] # ignore dummy
show_result(
args.input,
result,
CLASS_NAMES,
show=False,
out_file=args.output,
)
def main():
args = parse_args()
model = init_detector(
args.config, args.checkpoint, device=torch.device('cuda', args.device))
camera = cv2.VideoCapture(args.camera_id)
while True:
ret_val, img = camera.read()
result = inference_detector(model, img)
show_result(
img, result, model.CLASSES, score_thr=args.score_thr, wait_time=1)
def test_and_draw_from_single_file(sample_file, ext_name, bgr_file, out_file, config_file, checkpoint_file, score_threhold):
model = init_detector(config_file, checkpoint_file, device='cuda:0')
sample = None
if ext_name == 'bgr':
sample = mmcv.imread(sample_file)
elif ext_name == 'tiff':
sample = load_pol_sub_image(sample_file)
else:
sample = LoadPolNPZImageFromFile(sample_file)
print(sample)
result = inference_detector(model, sample)
img = mmcv.imread(bgr_file)
show_result(img, result, model.CLASSES, out_file=out_file, score_thr=score_threhold)
def main():
args = parse_args()
# build the model from a config file and a checkpoint file
model = init_detector(args.config_file, args.checkpoint, device='cuda:0')
# test a single image and show the results
img = args.input
result = inference_detector(model, img)
# visualize the results in a new window
# or save the visualization results to image files
show_result(
img, result, model.CLASSES, out_file=img.split('.')[0] + '_result.jpg')
def predict_with_txt(txt_name):
path = '/data2/yeliang/data/leakage_data/line_10_719/Camera7'
files = os.listdir(path)
imgs = [os.path.join(path, i) for i in files]
print("start inference")
for i, result in enumerate(inference_detector(model, imgs)):
show_result(
imgs[i],
result,
model.CLASSES,
out_path=
'/data2/yeliang/data/tunnel_camera/20190719_line10_camera/20190719_model/Camera7/',
out_file='{}'.format(files[i][:-3] + 'jpg'),
txt=txt_name)
def mask_rcnn():
cfg_file = 'configs/mask_rcnn_r50_fpn_1x.py'
ckp_file = '../../checkpoints/' \
'mask_rcnn_r50_fpn_1x_20181010-069fa190.pth'
model = apis.init_detector(cfg_file, ckp_file, device='cuda:0')
img_file = '/home/lhhuang/data/coco/val2017/000000397133.jpg'
img = mmcv.imread(img_file)
result = apis.inference_detector(model, img)
out_file = 'out.jpg'
apis.show_result(
img, result, model.CLASSES,
show=False, out_file=out_file)
def visualize():
args = parse_args()
model = load_model(args)
root = '/home/fengyao/MSCOCO2017dataset/test/test2017/'
save_path = '/home/fengyao/MSCOCO2017dataset/test/test2017/output/'
imgs = ['000000000001.jpg', '000000000016.jpg']
for index in range(0, len(imgs)):
result = inference_detector(model, root + imgs[index])
show_result(root + imgs[index],
result,
model.CLASSES,
show=False,
out_file=save_path + imgs[index])
print('[INFO]Done.')
def visual_image_reulsts(images_files, cfg_file, model_file, result_dir, device="cuda:0"):
if not osp.exists(result_dir):
os.makedirs(result_dir)
if not osp.exists(cfg_file) or not osp.exists(model_file):
print("Error: no such file: %s or %s"%(cfg_file, model_file))
return None
cfg = mmcv.Config.fromfile(cfg_file)
cfg.model.pretrained = None
model = build_detector(cfg.model, test_cfg=cfg.test_cfg)
_ = load_checkpoint(model, model_file)
#set specified score thresh for better results visualization
cfg.test_cfg["score_thr"] = 0.35
for i, result in enumerate(tqdm(inference_detector(model, images_files, cfg, device=device))):
show_result(images_files[i], result, dataset="hangkongbei", out_file=osp.join(result_dir, osp.basename(images_files[i])))
def faster_rcnn():
cfg_file = 'configs/faster_rcnn_r50_fpn_1x.py'
ckp_file = '../../checkpoints/' \
'faster_rcnn_r50_fpn_2x_20181010-443129e1.pth'
model = apis.init_detector(cfg_file, ckp_file, device='cuda:0')
img_file = '/home/lhhuang/data/VOCdevkit/VOC2007/JPEGImages/007663.jpg'
img = mmcv.imread(img_file)
result = apis.inference_detector(model, img)
out_file = 'out.jpg'
apis.show_result(
img, result, model.CLASSES,
show=False, out_file=out_file)
def test_and_draw_from_xmls(xml_dir, ext_name, sample_dir, bgr_dir, out_dir,
config_file, checkpoint_file, score_threhold):
model = init_detector(config_file, checkpoint_file, device='cuda:0')
xlms = os.listdir(xml_dir)
sample_ids = [
i.split('_')[0] + '_' + i.split('_')[1] for i in xlms
if i.endswith('.xml')
]
for xlm_filename in xlms:
sample_file = ''
sample_path = ''
sample = None
sample_id = xlm_filename.split('.')[0]
if ext_name == 'bgr':
sample_file = xlm_filename.split('.')[0] + '.tiff'
elif ext_name == 'pol' or ext_name == 'sub':
sample_file = sample_id.split('_')[0] + '_' + sample_id.split(
'_')[1] + '.' + 'tiff'
else:
sample_file = sample_id.split('_')[0] + '_' + sample_id.split(
'_')[1] + '.' + 'ext_name' + '.npz'
sample_path = os.path.join(sample_dir, sample_file)
if ext_name == 'bgr':
sample = mmcv.imread(sample_path)
elif ext_name == 'pol' or ext_name == 'sub':
sample = load_pol_sub_image(sample_path)
else:
sample = load_pol_npz_image(sample_path)
img_path = os.path.join(bgr_dir, xlm_filename.split('.')[0] + '.tiff')
img = mmcv.imread(img_path)
result = inference_detector(model, sample)
out_file = sample_id.split('_')[0] + '_' + sample_id.split(
'_')[1] + '.' + ext_name + '.jpg'
out_path = os.path.join(out_dir, out_file)
show_result(img,
result,
model.CLASSES,
show=False,
out_file=out_path,
score_thr=score_threhold)
print(out_path)
def main():
args = parse_args()
if not os.path.exists(args.test_folder):
os.makedirs(args.test_folder)
config_file = os.path.join(args.model_folder, args.config_filename)
checkpoint_file = os.path.join(args.model_folder, args.checkpoint_filename)
# build the model from a config file and a checkpoint file
model = init_detector(config_file, checkpoint_file, device='cuda:0')
# test a single image and show the results
imgs = [os.path.join(args.test_folder, img) for img in os.listdir(args.test_folder)
if (img.endswith('.jpg') or img.endswith('.jpeg'))]
print(imgs)
for i, result in enumerate(inference_detector(model, imgs)):
show_result(imgs[i], result, model.CLASSES, out_file=os.path.join(args.prediction_folder, 'result_{}.jpg'.format(i)))
def result():
cfg = mmcv.Config.fromfile(config2make_json)
cfg.model.pretrained = None
# construct the model and load checkpoint
model = build_detector(cfg.model, test_cfg=cfg.test_cfg)
_ = load_checkpoint(model, model2make_json)
# test a single image
imgs = os.listdir(pic_path)
meta = []
from tqdm import tqdm
for im in tqdm(imgs):
img = pic_path + im
img = mmcv.imread(img)
result = inference_detector(model, img)
re,img = show_result(img, result, dataset='cloths', show=False,score_thr = 0.5)
if len(re):
for box in re:
anno = {}
anno['name'] = im
anno['category'] = int(box[5])
anno['bbox'] = [round(float(i), 2) for i in box[0:4]]
anno['score'] = float(box[4])
meta.append(anno)
with open(json_path, 'w') as fp:
json.dump(meta, fp, cls=MyEncoder,indent=4, separators=(',', ': '))
def vis():
cfg = [mmcv.Config.fromfile(config_) for config_ in config]
for cfg_ in cfg:
cfg_.model.pretrained = None
# construct the model and load checkpoint
model = [build_detector(cfg_.model, test_cfg=cfg_.test_cfg) for cfg_ in cfg]
_ = [load_checkpoint(model[i], model_path[i]) for i in range(len(model))]
# test a single image
imgs = os.listdir(pic_path)
from tqdm import tqdm
for im in tqdm(imgs):
img = pic_path+im
img = mmcv.imread(img)
for i in range(len(model)):
result = inference_detector(model[i], img, cfg[i])
re,image = show_result(img,result,dataset='cloths',show = False,score_thr = 0.5)
if im in anno.keys():
bboxes = anno[im]
for bbox in bboxes:
box = bbox[0]
label = bbox[1]
x1 = int(box[0])
y1 = int(box[1])
x2 = int(box[2])
y2 = int(box[3])
cv2.rectangle(image, (x1, y1), (x2, y2), (0, 0, 255), 2)
# cv2.putText(image, '%d' % cat[label], (x2, y2), cv2.FONT_HERSHEY_COMPLEX, 0.8, (0, 0, 255), 1)
cv2.putText(image, '%d' % label, (x2, y2), cv2.FONT_HERSHEY_COMPLEX, 0.8, (0, 0, 255), 1)
name = config[i].split('_')[-1].split('.')[0]
cv2.namedWindow(str(name),0)
cv2.resizeWindow(str(name),1920,1080)
cv2.imshow(str(name),image)
cv2.waitKey(0)
