search

dvlab-research / Stratified-Transformer

Stratified Transformer for 3D Point Cloud Segmentation (CVPR 2022)
MIT License
376 stars 40 forks source link

Segmentation fault during testing #65

Open daylight111 opened 2 years ago

daylight111 commented 2 years ago

I have already set up the environment, compiled pointops successfully and change the model_path, save_folder and data_root_val accordingly. But, when i ran command "python3 test.py", the terminal printed "Segmentation fault".

X-Lai commented 2 years ago

Which line throws this error? Can you share more details of this error?

daylight111 commented 2 years ago 
   谢谢回复！我尝试在WSL2，Ubuntu 20.04，torch 1.10.1 + cu113的环境下，利用预训练模型对我自己的数据集进行语义分割。当我在Terminal运行python3 test.py语句后，出现报错`Segmentation fault`，针对这个错误python并不直接显示是哪条语句的问题，当我利用faulthandler包进行调试后，发现错误在第21行：`from util.voxelize import voxelize`。
   完整报错信息如下图所示：

Segmentation fault 另外，我想请问一下您有没有利用预训练模型对自定义数据集运行Stratified-Transformer的教程，本人是个科研新手，单看ReadMe文档还是不太会用您的模型，希望接受大神的指点，我也好教给课题组的其他同学，更好的传播您的文章。 感激！

daylight111 commented 2 years ago

By the way，这是我其他和torch相关的包的配置，我的cudatookit=11.3.1 torch相关包的配置