conda install torchvision==0.12.0了后测试detectron2仍显示ModuleNotFoundError: No module named 'torchvision'

安装完torchvision之后，仍然出现“ModuleNotFoundError: No module named 'torchvision'”的错误，可能是因为detectron2无法找到torchvision的路径。可以尝试以下两种方法解决该问题：

1. 确认torchvision是否已经正确安装。可以在终端中输入以下命令来检查：

   python -c "import torchvision; print(torchvision.__version__)"

如果能够正确输出torchvision的版本号，则说明torchvision已经成功安装。否则，需要重新安装torchvision。

2. 将torchvision的路径添加到PYTHONPATH环境变量中。可以在终端中输入以下命令：

   export PYTHONPATH=$PYTHONPATH:/path/to/torchvision

其中，/path/to/torchvision是torchvision的安装路径。这样，detectron2就能够找到torchvision的路径了。

相关问题

conda install pytorch==1.0.0 torchvision==0.2.1 -c pytorch 和conda install pytorch==1.0.0 torchvision==0.2.1 cuda90 -c pytorch有什么区别

`conda install pytorch==1.0.0 torchvision==0.2.1 -c pytorch` 是安装 PyTorch 和 torchvision 的 CPU 版本，而 `conda install pytorch==1.0.0 torchvision==0.2.1 cuda90 -c pytorch` 是安装 PyTorch 和 torchvision 的 CUDA 版本，需要与 NVIDIA 的 GPU 驱动程序和 CUDA 工具包一起使用。如果你不确定你的 GPU 是否支持 CUDA，可以使用第一种方式安装 CPU 版本。

ModuleNotFoundError: No module named 'torch' PS G:\NJFU\YOLO\yolov5-corrosion> conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cpuonly Collecting package metadata (current_repodata.json): done Solving environment: failed with initial frozen solve. Retrying with flexible solve. Collecting package metadata (repodata.json): done Solving environment: failed with initial frozen solve. Retrying with flexible solve. PackagesNotFoundError: The following packages are not available from current channels: - cpuonly - pytorch==1.8.0 - torchvision==0.9.0 - torchaudio==0.8.0

It seems that you are encountering some issues with installing the required packages for PyTorch. The error message suggests that the packages you are trying to install are not available from the current channels.

To resolve this issue, you can try the following steps:

1. Create a new conda environment:

conda create -n myenv python=3.8

2. Activate the new environment:

conda activate myenv

3. Install PyTorch and torchvision using conda:

conda install pytorch==1.8.0 torchvision==0.9.0 torchaudio==0.8.0 cpuonly -c pytorch

By specifying the `-c pytorch` flag, conda will search for the packages in the PyTorch channel.

Please note that the `cpuonly` variant installs PyTorch without CUDA support, which means it will not utilize GPU acceleration. If you have a compatible GPU and want to enable GPU acceleration, you should install the appropriate version of PyTorch with CUDA support.

I hope this helps! Let me know if you have any further questions.