Regarding custom dataset

[pytholic]

Hi. Thank you for the repository.

I wanted to ask if it is possible to get any guidance regarding the custom dataset. I have my own dataset i.e. X, Y, Z points with labels (4 columns in total) in txt files. I wanted to know how I can create a data loader for custom data and train the network.

[odddozen]

I would also like to know how to train on custom data, any guidance to save hours reverse understanding the code would be great

[pytholic]

@odddozen not sure about this repos/code, but I made my data work with PointNet using some other repositories and writing custom DataLoader.

[odddozen]

@rajahaseeb147 thanks for the response, sorry I didnt see it at the time. Would you be kind enough to let me look at your data loader for inspiration?

[pytholic]

@odddozen Sure. Here are some examples:

https://github.com/rajahaseeb147/3dFacialPartSegmentation/blob/main/Deep%20Learning/Implementation_2/pointnet/dataset_custom.py

https://github.com/rajahaseeb147/3dFacialPartSegmentation/blob/main/Deep%20Learning/Implementation_3/data_utils/CustomDataLoader.py

https://github.com/rajahaseeb147/3dFacialPartSegmentation/blob/main/Deep%20Learning/Implementation_1/part_seg_nosampling.ipynb

I found some implementation repos, and in those repos I tried to follow the ShapeNet dataloader because it seemed most relevant to my dataset.

[pytholic]

@odddozen links of those repos:

https://github.com/yanx27/Pointnet_Pointnet2_pytorch

https://github.com/fxia22/pointnet.pytorch

https://github.com/nikitakaraevv/pointnet

Hope it helps.

[odddozen]

@rajahaseeb147 thanks a lot! it definitely helps understand the general structure/approach. I was stuck for ages trying to compile spconv but finally got past that to try my own dataloader.

[pytholic]

@odddozen I am glad!