MobileNet_UNet语义分割模型训练 - 地平线开发者社区

用户您好，请详细描述您所遇到的问题，这会帮助我们快速定位问题~

1.芯片型号：X3派

2.天工开物开发包OpenExplorer版本：XJ3_OE_2.5.2

3.问题定位：模型训练

4.问题具体描述：我想用VM虚拟机来训练，

/horizon_xj3_open_explorer_v2.5.2-py38_20230331/ddk/samples/ai_toolchain/horizon_model_train_samples/scripts目录下

的unet模型

输入以下命令

python3 tools/calops.py --config configs/segmentation/unet.py --input-shape "1,3,1024,2048"

测试一下网络的计算量和参数数量

报错

Traceback (most recent call last):

File "tools/calops.py", line 5, in <module>

import horizon_plugin_pytorch as horizon

File "/home/qlf/.local/lib/python3.8/site-packages/horizon_plugin_pytorch/__init__.py", line 8, in <module>

from . import extension, nn, quantization # noqa: F401

File "/home/qlf/.local/lib/python3.8/site-packages/horizon_plugin_pytorch/extension.py", line 40, in <module>

_register_extensions()

File "/home/qlf/.local/lib/python3.8/site-packages/horizon_plugin_pytorch/extension.py", line 35, in _register_extensions

raise RuntimeError(

RuntimeError: Fail to load libhorizon_ops, origin msg is

libcuda.so.1: cannot open shared object file: No such file or directory

VM虚拟机环境Ubuntu20.04，搜索VM是不能装CUDA的，也就是不能用GPU，所以我想先用CPU试试，可是一直报这个错误