def main(args): device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') if os.path.isfile(args.config): config.merge_from_file(args.config) config.freeze() model = build_tracker(config.MODEL) if os.path.isfile(args.model): model.load_state_dict(torch.load(args.model, map_location='cpu')) model.cuda().eval() h, w = [int(s) for s in args.insize.split('x')] tracker = JDETracker() if os.path.isfile(args.img_path): dataloader = dataset.VideoLoader(args.img_path, (h,w,3)) else: dataloader = dataset.ImagesLoader(args.img_path, (h,w,3), formats=['*.jpg', '*.png']) strs = re.split(r'[\\, /]', args.img_path) imgpath = os.path.join(args.workspace, 'result', strs[-3], 'img') mkdir(imgpath) traj_path = os.path.join(args.workspace, 'result', '{}.txt'.format(strs[-3])) os.system('rm -f {}'.format(os.path.join(imgpath, '*'))) for i, (path, im, lb_im) in enumerate(dataloader): input = torch.from_numpy(lb_im).unsqueeze(0).to(device) with torch.no_grad(): outputs = model(input) print('{} {} {} {}'.format(path, im.shape, lb_im.shape, outputs.size()), end=' ') outputs = nonmax_suppression(outputs, args.score_thresh, args.iou_thresh)[0] if outputs is None: print('no object detected!') segments = re.split(r'[\\, /]', path) cv2.imwrite(os.path.join(imgpath, segments[-1]), im) continue print('{}'.format(outputs.size()), end=' ') outputs[:, :4] = ltrb_net2img(outputs[:, :4], (h,w), im.shape[:2]) if not args.only_detect: trajectories = tracker.update(outputs.numpy()) print('{}'.format(len(trajectories))) result = overlap_trajectory(trajectories, im) save_trajectories(traj_path, trajectories, i + 1) else: print('') result = overlap(outputs, im) segments = re.split(r'[\\, /]', path) cv2.imwrite(os.path.join(imgpath, segments[-1]), result) os.system('ffmpeg -f image2 -i {} {}.mp4 -y'.format(os.path.join(imgpath, '%06d.jpg'), os.path.join(args.workspace, 'result', strs[-3])))
def main(): args = parse_args() if os.path.isfile(args.config): config.merge_from_file(args.config) config.merge_from_list(args.opts) config.freeze() print(config) model = build_tracker(config.MODEL) classes = (nn.Conv2d, nn.BatchNorm2d, nn.Linear, nn.ReLU, nn.MaxPool2d) with open(args.save_name, 'w') as fd: for name, module in model.named_modules(): if isinstance(module, classes): fd.write('{}\n'.format(name))
def main(): args = parse_args() if os.path.isfile(args.config): config.merge_from_file(args.config) config.merge_from_list(args.opts) config.freeze() print(config) model = build_tracker(config.MODEL) if os.path.isfile(args.weight): model.load_state_dict(torch.load(args.weight, map_location='cpu')) model.cuda().eval() print(model) input = torch.rand(64, 3, 320, 576) with torch.no_grad(): output = model(input.cuda()) print('output size: {}'.format(output.size()))
def main(): # Parse configurations. args = parse_args() if os.path.isfile(args.config): config.merge_from_file(args.config) config.merge_from_list(args.opts) torch.backends.cudnn.benchmark = True mkdirs(config.SYSTEM.TASK_DIR) # Build dataset dataset = build_dataset(config.DATASET) # Build model. num_ide = int(dataset.max_id + 1) config.MODEL.ARGS.HEAD.ARGS[1]['num_ide'] = num_ide model = build_tracker(config.MODEL) # Train tracker now. train_tracker(model, dataset, config)
def main(args): device = torch.device('cuda' if torch.cuda.is_available() else 'cpu') # if args.backbone == 'darknet': # model = darknet.DarkNet(np.random.randint(0, 100, (12, 2))).to(device) # elif args.backbone == 'shufflenetv2': # model = shufflenetv2.ShuffleNetV2(np.random.randint(0, 100, (12, 2)), # model_size=args.thin).to(device) # else: # print('unknown backbone architecture!') # sys.exit(0) # # # load state dict except the classifier layer # model_dict = model.state_dict() # trained_model_dict = torch.load(os.path.join(args.model), map_location='cpu') # trained_model_dict = {k : v for (k, v) in trained_model_dict.items() if k in model_dict} # trained_model_dict = collections.OrderedDict(trained_model_dict) # model_dict.update(trained_model_dict) # model.load_state_dict(model_dict) # # model.eval() if os.path.isfile(args.config): config.merge_from_file(args.config) config.freeze() model = build_tracker(config.MODEL) if os.path.isfile(args.model): model.load_state_dict(torch.load(args.model, map_location='cpu')) model.cuda().eval() # if '320x576' in args.insize: # anchors = ((6,16), (8,23), (11,32), (16,45), # (21,64), (30,90), (43,128), (60,180), # (85,255), (120,360), (170,420), (340,320)) # elif '480x864' in args.insize: # anchors = ((6,19), (9,27), (13,38), (18,54), # (25,76), (36,107), (51,152), (71,215), # (102,305), (143,429), (203,508), (407,508)) # elif '608x1088' in args.insize: # anchors = ((8,24), (11,34), (16,48), (23,68), # (32,96), (45,135), (64,192), (90,271), # (128,384), (180,540), (256,640), (512,640)) h, w = [int(s) for s in args.insize.split('x')] # decoder = jde.JDEcoder((h, w), embd_dim=args.embedding) tracker = JDETracker() if os.path.isfile(args.img_path): dataloader = dataset.VideoLoader(args.img_path, (h, w, 3)) else: dataloader = dataset.ImagesLoader(args.img_path, (h, w, 3), formats=['*.jpg', '*.png']) strs = re.split(r'[\\, /]', args.img_path) imgpath = os.path.join(args.workspace, 'result', strs[-3], 'img') mkdir(imgpath) traj_path = os.path.join(args.workspace, 'result', '{}.txt'.format(strs[-3])) os.system('rm -f {}'.format(os.path.join(imgpath, '*'))) for i, (path, im, lb_im) in enumerate(dataloader): input = torch.from_numpy(lb_im).unsqueeze(0).to(device) with torch.no_grad(): outputs = model(input) # outputs = decoder(outputs) print('{} {} {} {}'.format(path, im.shape, lb_im.shape, outputs.size()), end=' ') outputs = nonmax_suppression(outputs, args.score_thresh, args.iou_thresh)[0] if outputs is None: print('no object detected!') segments = re.split(r'[\\, /]', path) cv2.imwrite(os.path.join(imgpath, segments[-1]), im) continue print('{}'.format(outputs.size()), end=' ') outputs[:, :4] = ltrb_net2img(outputs[:, :4], (h, w), im.shape[:2]) if not args.only_detect: trajectories = tracker.update(outputs.numpy()) print('{}'.format(len(trajectories))) result = overlap_trajectory(trajectories, im) save_trajectories(traj_path, trajectories, i + 1) else: print('') result = overlap(outputs, im) segments = re.split(r'[\\, /]', path) cv2.imwrite(os.path.join(imgpath, segments[-1]), result) os.system('ffmpeg -f image2 -i {} {}.mp4 -y'.format( os.path.join(imgpath, '%06d.jpg'), os.path.join(args.workspace, 'result', strs[-3])))
def main(): args = parse_args() if osp.isfile(args.config): config.merge_from_file(args.config) config.freeze() calibs = [parse_calibration_data(f) for f in args.calibration] print('calibration:\n{}'.format(calibs)) # Create input and output queue for each tracker. ncamera = len(args.inputs) images = [] # image queues tracklets = [] # tracklet queues for i in range(ncamera): images.append(Queue(maxsize=0)) tracklets.append(Queue(maxsize=0)) trajectories = [] # global trajectory list # Feature extractor. extractor = FeatureExtractor(model_name='osnet_x1_0', model_path=args.reid, device='cuda') # Create working threads. tid = 0 threads = [] exit = Value('i', 0) # shared thread exit switch for i in range(ncamera): # Datastore thread. tid += 1 threads.append(Datastore(tid, args.inputs[i], images[i])) # MTSCT thread. tid += 1 model = build_tracker(config.MODEL) model.load_state_dict(torch.load(args.tracker, map_location='cpu')) locator = ImageToWorldTsai(calibs[i]) threads.append( MTSCT(tid, images[i], tracklets[i], exit, model, locator)) # MTMCT thread. tid += 1 threads.append(MTMCT(tid, tracklets, trajectories, exit, extractor)) # Start all threads. for thread in threads: thread.start() # Waiting for Datastore finish. ndead = 0 while ndead != ncamera: ndead = sum([int(not t.is_alive()) for t in threads[:-1][0::2]]) time.sleep(1) print('Datastore done.') # Waiting for MTSCT finish. nempty = 0 while nempty != ncamera: nempty = sum([int(q.empty()) for q in images]) time.sleep(1) print('MTSCT done.') # Waiting for MTMCT finish. nempty = 0 while nempty != ncamera: nempty = sum([int(q.empty()) for q in tracklets]) time.sleep(1) print('MTMCT done.') exit.value = 1 for thread in threads: thread.join() print('All works done.')