使用 Python 和 OpenCV 进行实时目标检测的详解

import cv2
 
# 初始化摄像头和 OpenCV 窗口
cap = cv2.VideoCapture(0)
cv2.namedWindow('Realtime Object Detection', cv2.WINDOW_NORMAL)
 
# 加载预训练的深度学习目标检测模型
net = cv2.dnn.readNet('model_data/yolov3.weights', 'model_data/yolov3.cfg')
 
# 读取分类标签
with open('model_data/coco.names', 'r') as f:
    labels = [line.strip() for line in f.readlines()]
 
# 进行目标检测的循环
while True:
    ret, frame = cap.read()
    if not ret:
        break
    
    # 获取网络输入尺寸
    blob = cv2.dnn.blobFromImage(frame, 0.00392, (416, 416), (0, 0, 0), True, crop=False)
    
    # 设置网络输入并进行前向传播
    net.setInput(blob)
    outputs = net.forward(net.getUnconnectedOutLayersNames())
    
    # 解析检测结果
    for output in outputs:
        for detection in output:
            # 忽略置信度低的检测结果
            if detection[2] > 0.5:
                # 获取类别索引、置信度、坐标
                class_id = detection[0]
                confidence = detection[2]
                box = detection[3:7] * np.array([frame.shape[1], frame.shape[0], frame.shape[1], frame.shape[0]])
                start_x, start_y, end_x, end_y = box.astype(np.int)
                
                # 绘制矩形框和标签
                cv2.rectangle(frame, (start_x, start_y), (end_x, end_y), (255, 0, 0), 2)
                cv2.putText(frame, f"{labels[class_id]}: {confidence * 100:.2f}%", (start_x, start_y - 5), cv2.FONT_HERSHEY_SIMPLEX, 0.5, (255, 0, 0), 2)
    
    # 显示结果
    cv2.imshow('Realtime Object Detection', frame)
    
    # 按 'q' 退出循环
    if cv2.waitKey(1) & 0xFF == ord('q'):
        break
 
# 释放摄像头资源并关闭所有窗口
cap.release()
cv2.destroyAllWindows()

这段代码使用了OpenCV的深度学习模块来进行实时目标检测。首先，它初始化了摄像头和OpenCV窗口，然后加载了预训练好的YOLOv3模型和相关的配置文件。接着，它在一个循环中不断地从摄像头中读取帧，并通过YOLOv3模型进行目标检测。检测结果会被绘制在帧上，并且在窗口中实时显示。用户可以通过按'q'键退出程序。