segfault (core dumped)：段错误 (核心已转储)

[Huotianci]

Dear author, I am very interested in your work.
But some problems occurred while reproducing your code. My hardware and software parameters are: 3090, cuda12.1, python=3.8, pytorch=2.1.2. The deployment of the environment has been completed, but an error occurred during inference: 段错误 (核心已转储)

Here's what I do:
Step 1: Enter the command in the terminal: python infer.py --config config.yaml data.image_list=[test.jpg,] --image_preprocess, but an error is reported: 段错误 (核心已转储), as shown in the figure:

Step 2: Print where there may be problems, trying to find which line of code has the problem. Finally, it is found that the error caused by this line of code : self.point_encoder = tgs.find(self.cfg.pointcloud_encoder_cls)(self.cfg.pointcloud_encoder) . Further exploration revealed that it was a problem with the module = importlib.import_module(module_string, package=None) code in tgs.find.
The code is located as follows:

Could you give me some advice on this issue?

I would be very grateful if I could receive any help from you! ! !
Thank you again for such a great job! ! !

[zouzx]

Could you provide more information? E.g., could you locate the specific line in __init__ function of LocalPoolPointnet that causes this?

[LucioC]

I had the same issue. It seems it got fixed after I have installed pytorch-scatter using conda.

conda install pytorch-scatter -c pyg

[Huotianci]

I had the same issue. It seems it got fixed after I have installed pytorch-scatter using conda.

conda install pytorch-scatter -c pyg

Thank you for your reply, I have solved it successfully!