拉取基础镜像docker pull swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvidia/cuda:12.6.0-devel-ubuntu22.04docker tag swr.cn-north-4.myhuaweicloud.com/ddn-k8s/docker.io/nvidia/cuda:12.6.0-devel-ubuntu22.04 docker.io/nvidia/cuda:12.6.0-devel-ubuntu22.04docker run --shm-size“32g” --gpus all -it -v /media:/media -v /mnt:/mnt --name yolo_test0630 nvidia/cuda:12.6.0-devel-ubuntu22.04 /bin/bash以下在容器中运行安装 Miniforge3./Miniforge3-25.3.0-3-Linux-x86_64.shsource ~/miniforge3/bin/activatesource ~/.bashrcconda init bash创建虚拟环境conda create -n yolo_master python3.11 -yconda activate yolo_master查看 CUDA 版本nvcc -Vpip install torch2.6.0 torchvision0.21.0 torchaudio2.6.0 --index-url https://download.pytorch.org/whl/cu126cd YOLO-Master-main/pip install -r requirements.txtpip install -e .安装 flash-attention 加速库对 CUDA 版本的要求较高pip install packagingpip install ninjapip install opencv-python-headless4.13.0.90MAX_JOBS4 pip install flash-attn --no-build-isolation --no-cache-dir问题ImportError: libxcb.so.1: cannot open shared object file: No such file or directoryImportError: libxcb.so.1 是 Linux 无图形界面环境服务器/容器缺少 OpenCV 依赖的 X11 图形库cv2 默认绑定 GUI 模块找不到 libxcb 动态库。解决conda install -c conda-forge libxcb