knowledge_map/linux/nvidia jetson 部署torch和torchvision.md

当前部署环境

Jetpack 4.6 Python 3.6.9 cuda 10.2

安装版本

torch 1.8.0 torchvision 0.9.0

部署步骤

资源下载

参考链接 https://forums.developer.nvidia.com/t/pytorch-for-jetson/72048

实际操作

sudo apt-get install python3-pip libopenblas-base libopenmpi-dev libomp-dev
sudo apt-get install libjpeg-dev zlib1g-dev libpython3-dev libopenblas-dev libavcodec-dev libavformat-dev libswscale-dev

pip3 install -U pip -i https://mirrors.aliyun.com/pypi/simple/  # 换源
pip3 config set global.index-url https://mirrors.aliyun.com/pypi/simple/

pip3 install Cython scipy uff protobuf tqdm Pillow numpy==1.19.4 seaborn==0.9.1 
pip3 install -U scipy
pip3 install torch-1.8.0-cp36-cp36m-linux_aarch64.whl   # 安装torch
cd torchvision  # 编译安装torchvision
export BUILD_VERSION=0.9.0
python3 setup.py install --user    
mkdir  ~/.config/Ultralytics/  # 安装字体
cp yolov5-6.1/fonts/* ~/.config/Ultralytics/    

cd yolov5-6.1/  # 测试
python3 detect_rectify.py