search

drprojects / superpoint_transformer

Official PyTorch implementation of Superpoint Transformer introduced in [ICCV'23] "Efficient 3D Semantic Segmentation with Superpoint Transformer" and SuperCluster introduced in [3DV'24 Oral] "Scalable 3D Panoptic Segmentation As Superpoint Graph Clustering"
MIT License
601 stars 75 forks source link

How to train my own dataset #3

Closed kaka-ice closed 1 year ago

kaka-ice commented 1 year ago

Hello Professor！ My point cloud only has three properties of xyz, and the fourth column is the label value, how to train my own dataset using SPT network

drprojects commented 1 year ago

Hello, thanks for your interest in the project !

Although radiometric information can benefit model performance, SPT can work without it.

Actually, one can easily remove the radiometric features from the provided datasets by removing 'rgb' (for S3DIS), 'hsv' (for KITTI-360), 'intensity' (for DALES), from datamodule.partition_hf and datamodule.point_hf.

However, from what I gather from your message, I think you may want to create your own dataset. To this end, I invite you to have a look at the instructions here.

Best,

Damien

drprojects commented 1 year ago

I consider this issue closed. Feel free to re-open it if necessary.