1 下载ONNX文件

ONNX下载地址

2 Python代码

import cv2
from ultralytics import YOLO


def check(yolo:str, path:str):
    # 加载 YOLOv11
    model = YOLO(yolo)
    # 读取图片
    img = cv2.imread(path)
    # 推理（可以传文件路径或图像数组）
    results = model(img)
    # 取第一个结果（图像级别）
    for result in results:
        for box in result.boxes:
            # 获取边框坐标并转换为整数
            x1, y1, x2, y2 = map(int, box.xyxy[0])
            # 获取类别编号和置信度
            id = int(box.cls[0])
            # 获取标签名称
            label = result.names[id]
            # 画矩形框
            cv2.rectangle(img, (x1, y1), (x2, y2), (0, 255, 0), 2)
            print('{}:{},\t {},{}\t {},{}'.format(id, label, x1, y1, x2, y2))
            # 绘制文字
            conf = float(box.conf[0])
            text = f'{label} {conf:.2f}'
            cv2.putText(img, text, (x1, y1 - 10), cv2.FONT_HERSHEY_SIMPLEX, 0.6, (0, 255, 0), 2)
    # 显示图像
    cv2.imshow("YOLOv11", img)
    cv2.waitKey(0)
    cv2.destroyAllWindows()
    # 可选：保存标注后的图像
    cv2.imwrite("output.jpg", img)


def video(yolo:str):
    # 加载 YOLO 模型（你也可以换成 yolov8n.pt / yolov8s.pt 等）
    model = YOLO(yolo)  # 更轻量，适合实时运行
    # 打开默认摄像头
    cap = cv2.VideoCapture(0)
    while True:
        ret, frame = cap.read()
        if not ret:
            break
        # 推理：BGR 图像直接传入
        results = model(frame, verbose=False)
        # 可视化结果：会自动绘制框和类别
        annotated_frame = results[0].plot()
        # 显示
        cv2.imshow("YOLO Detection", annotated_frame)
        # 按 Q 键退出
        if cv2.waitKey(1) & 0xFF == ord('q'):
            break
    cap.release()
    cv2.destroyAllWindows()


def convert(yolo:str):
    # 加载你的训练模型（.pt）
    model = YOLO("runs/detect/train/weights/best.pt")
    # 导出为 ONNX
    return model.export(format="onnx", dynamic=True)  # or simplify=True to simplify

check('yolo11n.pt','./11.png')

0: 320x640 6 persons, 11 cars, 650.3ms
Speed: 176.4ms preprocess, 650.3ms inference, 43.0ms postprocess per image at shape (1, 3, 320, 640)
2:car,	 234,542	 469,661
2:car,	 637,224	 795,324
2:car,	 1580,571	 1729,657
2:car,	 585,558	 714,647
2:car,	 1223,236	 1389,323
2:car,	 1249,508	 1510,626
0:person,	 1414,195	 1513,331
2:car,	 275,177	 509,314
0:person,	 382,850	 454,919
0:person,	 1765,554	 1868,659
0:person,	 1065,520	 1192,660
0:person,	 861,525	 952,657
2:car,	 1725,244	 1869,333
2:car,	 937,166	 1106,228
0:person,	 964,538	 1061,649
2:car,	 872,489	 1011,570
2:car,	 897,509	 1012,567

[Done] exited with code=0 in 39.085 seconds