官方文檔地址：https://gitee.com/ascend/ModelZoo-PyTorch/tree/master/ACL_PyTorch/built-in/cv/Yolov5_for_Pytorch
問題匯總
| 序號 | 問題 | 原因 | 解決辦法 |
| 1 | ModuleNotFoundError: No module named 'torch' | 缺少依賴包 | |
| 2 | RuntimeError: Unable to open /dev/urandom | 應該還是缺陷問題，不能打開文件，導致運行時間出錯。 | 修改文件的權限 chmod 777 /dev/urandom |
| 3 | ModuleNotFoundError: No module named '_bz2' | python下缺包 | 添加“”文件包 拷貝python3.6下的該類型包，並修改名字 scp /usr/lib/python3.6/lib-dynload/_bz2.cpython-36m-aarch64-linux-gnu.so 用戶名@localhost:/usr/local/python3.7.5/lib/python3.7/lib-dynload/_bz2.cpython-37m-aarch64-linux-gnu.so |
| 4 | ImportError: libGL.so.1: cannot open shared object file: No such file or directory | 缺少文件 | 若需要添加用戶sudo權限，則root用戶下執行添加權限操作： usermod libaoshan -G sudo 安裝相關依賴： sudo apt-get install libgl1-mesa-glx |
| 5 | ModuleNotFoundError: No module named '_lzma' | 缺少包 | https://zhuanlan.zhihu.com/p/404162713 |
| 6 | Sorry, user libaoshan is not allowed to execute '/usr/bin/vim /usr/local/python3.7.5/lib/python3.7/lzma.py' as root on davinci-mini. | 缺少執行權限 | |
| 7 | AttributeError: ‘Upsample‘ object has no attribute ‘recompute_scale_factor‘ | https://blog.csdn.net/qq_42578742/article/details/123582773 | |
| 8 | ONNX: export failure: No module named 'onnx' | 缺包 | pip install onnx |
| 9 | /usr/local/python3.7.5/bin/python3.7: No module named onnxsim | 缺包 | pip install onnx-simplifier |
| ModuleNotFoundError: No module named 'onnx.optimizer' | onnx包和onnx-simplifier版本不匹配 | pip3 install skl2onnx pip install onnx==1.8.1 https://blog.csdn.net/sinat_38685910/article/details/119329283 或者使用源碼來解決問題，安裝onnx和如下包。 https://github.com/onnx/optimizer |
|
參考：
昇騰論壇 https://bbs.huaweicloud.com/forum/thread-173871-1-1.html