开始训练

以yolov10 or yolov8 为例 (官方文档 )

备注

• yolov5可参考北极熊战队的教程

• 配置数据集

备注

yolo系列模型支持coco或coco txt数据集标注格式

数据集目录结构应为

├─dataset
    ├─test # test数据集可有可无
    │  ├─images
    │  └─labels
    ├─train
    │  ├─images
    │  └─labels
    └─valid
        ├─images
        └─labels

• 创建data.yaml,比如

# Train/val/test sets as 1) dir: path/to/imgs, 2) file: path/to/imgs.txt, or 3) list: [path/to/imgs1, path/to/imgs2, ..]

path: dataset  # dataset root dir
train: train/images  # train images (relative to 'path')
val: val/images  # val images (relative to 'path')
test:  # test images (optional)

# Classes
nc: 1
names:
0: car

• 创建训练脚本

if __name__ == '__main__':
    from ultralytics import YOLOv10
    model = YOLOv10(r'G:\yolov10\runs\detect\train2\weights\last.pt')
    model.train(data='data.yaml', epochs=100, batch=64, imgsz=480)

• 开始训练

在python控制台中输入

import torch
print(torch.cuda.is_available)

输出若为true则可以用显卡进行训练，若为false则转到常见问题查看解决方案.

• 创建预测脚本

from ultralytics import YOLOv10

# Load a model
model = YOLOv10(r"G:\yolov10\runs\detec.. note::t\train2\weights\best.pt")  # load a custom model
source = r"your.mp4"  # predict on a video
# results = model(source, stream=True)  # generator of Results objects
results = model.predict(source, conf=0.6,device='0',half=False,imgsz=480,show=True)  # generator of Results objects

• 模型导出请看官方教程

2024.9.2 123456dfg edit