how to train my own dataset?

[SakuraTau]

Thanks for your great work! I have some questions for you, I noticed that the format of the for-instance dataset is: x,y,z,intensity,semantic_seg,treeID, if I want to train my own data, since my data doesn't have the intensity information, is it possible for me to mimic the format of the data as x,y,z,semantic_seg ,treeID?Looking forward to your reply, thank you very much!

[bxiang233]

Thanks for your great work! I have some questions for you, I noticed that the format of the for-instance dataset is: x,y,z,intensity,semantic_seg,treeID, if I want to train my own data, since my data doesn't have the intensity information, is it possible for me to mimic the format of the data as x,y,z,semantic_seg ,treeID?Looking forward to your reply, thank you very much!

Hi, sorry I somehow ignore your question before... Have you already solved this problem? If not, please try to modify this part of the code:

https://github.com/prs-eth/PanopticSegForLargeScalePointCloud/blob/b151bcb0785c41802cae091a02e960724c2da069/torch_points3d/datasets/segmentation/treeins.py#L59

We did not not read intensity either.