1. 创建环境

conda create --name yolo_new python=3.10

2. 安装 yolo

pip install -U ultralytics

3. 编写获取视频代码

1. 环境：MacOS + iphone 摄像头

2. 导入相应的头文件

import cv2
from ultralytics import solutions

2. 获取摄像头

def open_iphone_camera_with_cv():
    """
    主函数：找到并打开iPhone摄像头
    """
    print("正在检测iPhone摄像头...\n")
        
        # for i in range(5):
    cap = cv2.VideoCapture(0)
    if cap.isOpened():
        print("打开默认摄像头 (索引0)")

        while True:
            ret, frame = cap.read()
            if ret:
                frame_corrected = cv2.flip(frame, 1)

                cv2.imshow(f'Camera {0} - 按q退出', frame_corrected)

            if cv2.waitKey(1) & 0xFF == ord('q'):
                break

        cv2.destroyAllWindows()
        cap.release()
        print("摄像头已关闭")
    else:
        print("无法打开摄像头，请检查连接。")

# 运行
if __name__ == "__main__":
    # 方法1：自动检测并打开iPhone摄像头
    open_iphone_camera_with_cv()

3. 设置视频长和宽

cap.set(cv2.CAP_PROP_FRAME_WIDTH, 2048)
cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 2048)

4. 设置yolo 代码

region_points = [(20, 400), (1080, 400), (1080, 360), (20, 360)]
        # w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))
        # video_writer = cv2.VideoWriter("object_counting_output.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))
        counter = solutions.ObjectCounter(
            show=True,  # display the output
            region=region_points,  # pass region points
            model="yolo26s.pt",  # model="yolo26n-obb.pt" for object counting with OBB model.
            # classes=[0, 2],  # count specific classes, e.g., person and car with the COCO pretrained model.
            # tracker="botsort.yaml",  # choose trackers, e.g., "bytetrack.yaml"
        )

会自动下载模型，region_points 为识别的可视野的大小

5. 处理视频中的一帧图像

results = counter(frame_corrected)

6. 完整代码

import cv2
from ultralytics import solutions


def open_iphone_camera_with_cv():
    """
    主函数：找到并打开iPhone摄像头
    """
    print("正在检测iPhone摄像头...\n")
        
        # for i in range(5):
    cap = cv2.VideoCapture(0)
    if cap.isOpened():
        print("打开默认摄像头 (索引0)")
        cap.set(cv2.CAP_PROP_FRAME_WIDTH, 2048)
        cap.set(cv2.CAP_PROP_FRAME_HEIGHT, 2048)
        region_points = [(20, 400), (1080, 400), (1080, 360), (20, 360)]
        # w, h, fps = (int(cap.get(x)) for x in (cv2.CAP_PROP_FRAME_WIDTH, cv2.CAP_PROP_FRAME_HEIGHT, cv2.CAP_PROP_FPS))
        # video_writer = cv2.VideoWriter("object_counting_output.avi", cv2.VideoWriter_fourcc(*"mp4v"), fps, (w, h))
        counter = solutions.ObjectCounter(
            show=True,  # display the output
            region=region_points,  # pass region points
            model="yolo26s.pt",  # model="yolo26n-obb.pt" for object counting with OBB model.
            # classes=[0, 2],  # count specific classes, e.g., person and car with the COCO pretrained model.
            # tracker="botsort.yaml",  # choose trackers, e.g., "bytetrack.yaml"
        )

        while True:
            ret, frame = cap.read()
            if ret:
                frame_corrected = cv2.flip(frame, 1)

                results = counter(frame_corrected)

                cv2.imshow(f'Camera {0} - 按q退出', results.plot_im)

            if cv2.waitKey(1) & 0xFF == ord('q'):
                break

        cv2.destroyAllWindows()
        cap.release()
        print("摄像头已关闭")
    else:
        print("无法打开摄像头，请检查连接。")

    


# 运行
if __name__ == "__main__":
    # 方法1：自动检测并打开iPhone摄像头
    open_iphone_camera_with_cv()

4. 总结

1. yolo 26 安装方便

2. 占用cpu更低

3. 不同模型识别速度不同，模型越大识别越慢

yolo26n.pt 最小(5.5M)，识别最快， 在 50ms左右

yolo26s.pt 20.4M, 在75ms左右

yolo26m.pt 44.3M, 在120ms左右

yolo26l.pt 53.2M，在150ms左右

yolo26x.pt 118.7M，在170ms左右