import argparse import cv2 import os os.environ["OMP_NUM_THREADS"] = "1" os.environ["OPENBLAS_NUM_THREADS"] = "1" os.environ["MKL_NUM_THREADS"] = "1" os.environ["VECLIB_MAXIMUM_THREADS"] = "1" os.environ["NUMEXPR_NUM_THREADS"] = "1" import sys import numpy as np from pathlib import Path import torch FILE = Path(__file__).resolve() ROOT = FILE.parents[0] WEIGHTS = ROOT / 'weights' if str(ROOT) not in sys.path: sys.path.append(str(ROOT)) if str(ROOT / 'trackers' / 'strongsort') not in sys.path: sys.path.append(str(ROOT / 'trackers' / 'strongsort')) from ultralytics.nn.autobackend import AutoBackend from ultralytics.yolo.data.dataloaders.stream_loaders import LoadImages from ultralytics.yolo.data.utils import VID_FORMATS from ultralytics.yolo.utils import LOGGER, colorstr from ultralytics.yolo.utils.checks import check_file, check_imgsz from ultralytics.yolo.utils.files import increment_path from ultralytics.yolo.utils.torch_utils import select_device from ultralytics.yolo.utils.ops import Profile, non_max_suppression, scale_boxes from ultralytics.yolo.utils.plotting import Annotator, colors from trackers.multi_tracker_zoo import create_tracker from sender.jsonlogger import JsonLogger from sender.szmq import ZmqLogger @torch.no_grad() def run( source='0', yolo_weights=WEIGHTS / 'yolov8n.pt', reid_weights=WEIGHTS / 'osnet_x0_25_msmt17.pt', imgsz=(640, 640), conf_thres=0.7, iou_thres=0.45, max_det=1000, device='', show_vid=True, save_vid=True, project=ROOT / 'runs' / 'track', name='exp', exist_ok=False, line_thickness=2, hide_labels=False, hide_conf=False, half=False, vid_stride=1, enable_json_log=False, enable_zmq=False, zmq_ip='localhost', zmq_port=5555, ): source = str(source) is_file = Path(source).suffix[1:] in (VID_FORMATS) if is_file: source = check_file(source) device = select_device(device) model = AutoBackend(yolo_weights, device=device, dnn=False, fp16=half) stride, names, pt = model.stride, model.names, model.pt imgsz = check_imgsz(imgsz, stride=stride) dataset = LoadImages( source, imgsz=imgsz, stride=stride, auto=pt, transforms=getattr(model.model, 'transforms', None), vid_stride=vid_stride ) bs = len(dataset) tracking_config = ROOT / 'trackers' / 'strongsort' / 'configs' / 'strongsort.yaml' tracker = create_tracker('strongsort', tracking_config, reid_weights, device, half) save_dir = increment_path(Path(project) / name, exist_ok=exist_ok) (save_dir / 'tracks').mkdir(parents=True, exist_ok=True) # Initialize loggers json_logger = JsonLogger(f"{source}-strongsort.log") if enable_json_log else None zmq_logger = ZmqLogger(zmq_ip, zmq_port) if enable_zmq else None vid_path, vid_writer = [None] * bs, [None] * bs dt = (Profile(), Profile(), Profile()) for frame_idx, (path, im, im0s, vid_cap, s) in enumerate(dataset): with dt[0]: im = torch.from_numpy(im).to(model.device) im = im.half() if model.fp16 else im.float() im /= 255.0 if len(im.shape) == 3: im = im[None] with dt[1]: pred = model(im, augment=False, visualize=False) with dt[2]: pred = non_max_suppression(pred, conf_thres, iou_thres, None, False, max_det=max_det) for i, det in enumerate(pred): seen = 0 p, im0, _ = path, im0s.copy(), dataset.count p = Path(p) annotator = Annotator(im0, line_width=line_thickness, example=str(names)) if len(det): # Filter detections for 'car' class only (class 2 in COCO dataset) car_mask = det[:, 5] == 2 # car class index is 2 det = det[car_mask] if len(det): det[:, :4] = scale_boxes(im.shape[2:], det[:, :4], im0.shape).round() for *xyxy, conf, cls in reversed(det): c = int(cls) id = f'{c}' label = None if hide_labels else (f'{id} {names[c]}' if hide_conf else f'{id} {names[c]} {conf:.2f}') annotator.box_label(xyxy, label, color=colors(c, True)) t_outputs = tracker.update(det.cpu(), im0) if len(t_outputs) > 0: for j, (output) in enumerate(t_outputs): bbox = output[0:4] id = output[4] cls = output[5] conf = output[6] # Log tracking data if json_logger or zmq_logger: track_data = { 'bbox': bbox.tolist() if hasattr(bbox, 'tolist') else list(bbox), 'id': int(id), 'cls': int(cls), 'conf': float(conf), 'frame_idx': frame_idx, 'source': source, 'class_name': names[int(cls)] } if json_logger: json_logger.send(track_data) if zmq_logger: zmq_logger.send(track_data) if save_vid or show_vid: c = int(cls) id = int(id) label = f'{id} {names[c]}' if not hide_labels else f'{id}' if not hide_conf: label += f' {conf:.2f}' annotator.box_label(bbox, label, color=colors(c, True)) im0 = annotator.result() if show_vid: cv2.imshow(str(p), im0) if cv2.waitKey(1) == ord('q'): break if save_vid: if vid_path[i] != str(save_dir / p.name): vid_path[i] = str(save_dir / p.name) if isinstance(vid_writer[i], cv2.VideoWriter): vid_writer[i].release() if vid_cap: 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: fps, w, h = 30, im0.shape[1], im0.shape[0] vid_writer[i] = cv2.VideoWriter(vid_path[i], cv2.VideoWriter_fourcc(*'mp4v'), fps, (w, h)) vid_writer[i].write(im0) LOGGER.info(f"{s}{'' if len(det) else '(no detections), '}{dt[1].dt * 1E3:.1f}ms") for i, vid_writer_obj in enumerate(vid_writer): if isinstance(vid_writer_obj, cv2.VideoWriter): vid_writer_obj.release() cv2.destroyAllWindows() LOGGER.info(f"Results saved to {colorstr('bold', save_dir)}") def xyxy2xywh(x): # Convert nx4 boxes from [x1, y1, x2, y2] to [x, y, w, h] where xy1=top-left, xy2=bottom-right y = x.clone() if isinstance(x, torch.Tensor) else np.copy(x) y[:, 0] = (x[:, 0] + x[:, 2]) / 2 # x center y[:, 1] = (x[:, 1] + x[:, 3]) / 2 # y center y[:, 2] = x[:, 2] - x[:, 0] # width y[:, 3] = x[:, 3] - x[:, 1] # height return y def parse_opt(): parser = argparse.ArgumentParser() parser.add_argument('--source', type=str, default='0', help='file/dir/URL/glob, 0 for webcam') parser.add_argument('--yolo-weights', nargs='+', type=str, default=WEIGHTS / 'yolov8n.pt', help='model path') parser.add_argument('--reid-weights', type=str, default=WEIGHTS / 'osnet_x0_25_msmt17.pt') parser.add_argument('--imgsz', '--img', '--img-size', nargs='+', type=int, default=[640], help='inference size h,w') parser.add_argument('--conf-thres', type=float, default=0.7, help='confidence threshold') parser.add_argument('--iou-thres', type=float, default=0.45, help='NMS IoU threshold') parser.add_argument('--max-det', type=int, default=1000, help='maximum detections per image') parser.add_argument('--device', default='', help='cuda device, i.e. 0 or 0,1,2,3 or cpu') parser.add_argument('--show-vid', action='store_true', help='display results') parser.add_argument('--save-vid', action='store_true', help='save video tracking results') parser.add_argument('--project', default=ROOT / 'runs' / 'track', help='save results to project/name') parser.add_argument('--name', default='exp', help='save results to project/name') parser.add_argument('--exist-ok', action='store_true', help='existing project/name ok, do not increment') parser.add_argument('--line-thickness', default=2, type=int, help='bounding box thickness (pixels)') parser.add_argument('--hide-labels', default=False, action='store_true', help='hide labels') parser.add_argument('--hide-conf', default=False, action='store_true', help='hide confidences') parser.add_argument('--half', action='store_true', help='use FP16 half-precision inference') parser.add_argument('--vid-stride', type=int, default=1, help='video frame-rate stride') parser.add_argument('--enable-json-log', action='store_true', help='enable JSON file logging') parser.add_argument('--enable-zmq', action='store_true', help='enable ZMQ messaging') parser.add_argument('--zmq-ip', type=str, default='localhost', help='ZMQ server IP') parser.add_argument('--zmq-port', type=int, default=5555, help='ZMQ server port') opt = parser.parse_args() opt.imgsz *= 2 if len(opt.imgsz) == 1 else 1 return opt def main(opt): run(**vars(opt)) if __name__ == "__main__": opt = parse_opt() main(opt)