search

Pointcept / PointTransformerV3

[CVPR'24 Oral] Official repository of Point Transformer V3 (PTv3)
MIT License
787 stars 46 forks source link

Assertion Error: assert depth <= 16 #78

Open mbendjilali opened 3 months ago

mbendjilali commented 3 months ago

Hello, I'm trying to run PTv3 on a custom dataset, but I unfortunately run into the assertion error mentioned in the title, in the serialization function, line 74 of the structure.py script. Could you explain what it means, and also how to alleviate it please ? :)

AxelCole commented 3 months ago

Hey, I had the same problem. In my case, it was because my point cloud coordinates were too big to be encoded with only 16 bits (depth) during serialization. Consider normalizing your point cloud coordinates. Hope it will help !

xiaosa269 commented 2 months ago

Hello, I have encountered similar problems in training nuScenes and SemanticKITTI datasets. Have you solved this problem?

Gofinge commented 2 months ago

Hey, I had the same problem. In my case, it was because my point cloud coordinates were too big to be encoded with only 16 bits (depth) during serialization. Consider normalizing your point cloud coordinates. Hope it will help !

For real-word point cloud, I suggest not normalizing the coordinates. Just increase the grid_size or cropping your point cloud.