YOLO图像检测

YOLO(You Only Look Once)是一种流行的物体检测和图像分割模型，因其高速度和高精度而迅速受到欢迎。

目前最新版本是YOLO11，项目地址：https://github.com/ultralytics/ultralytics

安装

pip install ultralytics yolo detect predict model=yolo11n.pt source='bus.jpg'

物体检测

from ultralytics import YOLO

def yolo_detect():
    model = YOLO('yolo11n.pt')
    results = model.predict('bus.jpg', save=True)
    for result in results:
        result.show()

if __name__ == '__main__':
    yolo_detect()

分段

def yolo_segment():
    model = YOLO('yolo11n-seg.pt')
    results = model.predict('bus.jpg', save=True)
    for result in results:
        result.show()

if __name__ == '__main__':
    yolo_segment()

分类

def yolo_classify():
    model = YOLO('yolo11n-cls.pt')
    results = model.predict('bus.jpg', save=True)
    for result in results:
        result.show()

if __name__ == '__main__':
    yolo_segment()

姿态

def yolo_pose():
    model = YOLO('yolo11n-pose.pt')
    results = model.predict('bus.jpg', save=True)
    for result in results:
        result.show()

if __name__ == '__main__':
    yolo_pose()

定向边缘检测

def yolo_obb():
    model = YOLO('yolo11n-obb.pt')
    results = model.predict('boats.jpg', save=True)
    for result in results:
        result.show()

if __name__ == '__main__':
    yolo_obb()

多目标跟踪

def yolo_track():
    model = YOLO('yolo11n.pt')
    model.track('car.mp4', show=True, save=True)

if __name__ == '__main__':
    yolo_track()

可用模型

官方训练好的模型有五种尺寸，模型越越大，精度越越高，相对的速度越来越慢，从小到大分别是

• yolo11n
• yolo11s
• yolo11m
• yolo11l
• ylol11x