jetson orin nano安装环境
更新软件、安装pip3 配置源等
下载软件
sudo apt install python3-pip nano -ysudo -H pip3 install jetson-statssudo apt-get install libopenblas-devsudo jtop 查看jetpack版本设置pip源
mkdir -p ~/.pipnano ~/.pip/pip.conf安装cuda
sudo apt-get install python3-pip -yrebootjtopsudo apt install nvidia-jetpacksudo vim ~/.bashrc 添加
export PATH=/usr/local/cuda/bin:$PATH
export LD_LIBRARY_PATH=/usr/local/cuda/lib64:$LD_LIBRARY_PATH更改cuda为11.7
cuda下载链接https://developer.nvidia.com/cuda-12-0-0-download-archive?target_os=Linux&target_arch=aarch64-jetson&Compilation=Native&Distribution=Ubuntu&target_version=20.04&target_type=deb_local
安装paddlepaddle
paddle下载地址
sudo pip install paddlepaddle_gpu-2.4.2-cp38-cp38-linux_aarch64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple安装yolo8、torch-gpu
sudo pip install ultralytics==8.1.0 -i https://pypi.tuna.tsinghua.edu.cn/simplesudo pip uninstall torchsudo pip uninstall torchvisionsudo apt-get install libopenblas-devpip install torch-2.1.0a0+41361538.nv23.06-cp38-cp38-linux_aarch64.whl -i https://pypi.tuna.tsinghua.edu.cn/simple安装torchvision
cd visionsudo export BUILD_VERSION=0.16.0sudo python3 setup.py install --user注意torch和torchvision的对应关系，安装ultralytics时可能覆盖安装torch和torchvision，需重新安装torchvision
pip install torchvision==0.16.0已验证版本：
| jetpack | cuda | cudnn | paddlepaddle | yolo | torch | torchvision |
|---|---|---|---|---|---|---|
| 5.1.4 | 11.4 | 8.6 | 2.4.2 | 8.1.0 | 2.1.0 | 0.16.0 |
| 5.1.2 | 11.4 | 8.6 | 2.4.2 | 8.1.0 | 2.0.0 | 0.15.1 |
设置服务自启动(方式一)
1. 编译服务文件 nano /etc/systemd/system/video_hit.service
2. 启动服务
systemctl start laser-fulingsystemctl enable laser-fuling修改于 2025-07-25 06:06:12