windows下yolov8训练环境配置

安装

anaconda3

conda create -n yolov8 python=3.8
## 进入yolov8 虚拟环境
activate yolov8 

pip install ultralytics

conda install chardet

安装pytorch、cuda

https://pytorch.org/get-started/locally/

conda install pytorch torchvision torchaudio pytorch-cuda=11.8 -c pytorch -c nvidia

安装onnx格式支持

conda install onnx

导出pt模型为onnx格式：

yolo export model=best.pt format=onnx

https://docs.ultralytics.com/tasks/detect/#models

训练：

yolo detect train data=my_coco.yaml model=yolov8s.pt epochs=150 imgsz=640 batch=64 workers=0 device=0