yolov8 deepsort行人跟踪

### 使用YOLOv8和DeepSort实现行人跟踪 #### 安装依赖库 为了使用YOLOv8和DeepSort进行行人跟踪，首先需要安装必要的Python包。可以按照以下命令来设置开发环境： ```bash pip install torch torchvision torchaudio --extra-index-url https://download.pytorch.org/whl/cu117 pip install ultralytics opencv-python-headless numpy scipy filterpy lap ``` #### 下载预训练模型 下载YOLOv8的预训练权重文件以及DeepSort所需的资源。通常这些可以在GitHub仓库或者其他公开的数据集中找到。 #### 初始化YOLOv8检测器 创建一个脚本来加载YOLOv8模型并对输入图像执行目标检测操作。 ```python from ultralytics import YOLO def initialize_yolov8(): model = YOLO('path/to/yolov8n.pt') # 加载YOLOv8 nano模型 return model ``` #### 配置DeepSort参数 配置DeepSort的相关参数，比如最大消失时间、NN预算等，并初始化DeepSort实例。 ```python import deep_sort.nn_matching as nn_matching from deep_sort.detection import Detection from deep_sort.tracker import Tracker def setup_deepsort(max_age=30, n_init=3, nn_budget=None): metric = nn_matching.NearestNeighborDistanceMetric( "cosine", max_cosine_distance=0.2, budget=nn_budget) tracker = Tracker(metric, max_age=max_age, n_init=n_init) return tracker ``` #### 实现跟踪逻辑 编写主循环，在每一帧上运行YOLOv8得到边界框预测结果，再利用DeepSort完成实际的目标跟踪过程。 ```python import cv2 def run_tracking(video_path='input_video.mp4', output_path='output_video.avi'): cap = cv2.VideoCapture(video_path) fourcc = cv2.VideoWriter_fourcc(*'XVID') out = cv2.VideoWriter(output_path, fourcc, 20.0, (int(cap.get(3)), int(cap.get(4)))) yolov8_model = initialize_yolov8() deepsort_tracker = setup_deepsort() while True: ret, frame = cap.read() if not ret: break results = yolov8_model(frame)[0].boxes.data.cpu().numpy()[:, :4] detections = [] for bbox in results: confidences = [1.] * len(results) # 假设所有检测都具有相同的高置信度 class_ids = [-1] * len(results) # 如果只需要处理特定类别，则应调整此处 detection = Detection(bbox=bbox.tolist(), confidence=confidences[i], feature=np.random.rand(128)) detections.append(detection) deepsort_tracker.predict() matches = deepsort_tracker.update(detections=detections) for track_id, bbox in enumerate(matches): x1, y1, x2, y2 = map(int, bbox[:4]) label = f'Track ID: {track_id}' cv2.rectangle(frame, pt1=(x1,y1), pt2=(x2,y2), color=(0,255,0), thickness=2) cv2.putText(img=frame,text=label,org=(x1,y1-10), fontFace=cv2.FONT_HERSHEY_SIMPLEX, fontScale=0.9,color=(0,255,0),thickness=2,lineType=cv2.LINE_AA) out.write(frame) cap.release() out.release() ``` 此段代码展示了如何结合YOLOv8与DeepSort来进行基本的人群行为分析任务，包括但不限于行人重识别、轨迹绘制等功能[^1]。