adding Superpoint Transformer (ICCV'23 with 3DV'24 Oral follow-up)
Hello,
Thanks for maintaining this awesome repo ! I would like to suggest a new method for the point cloud processing section.
Superpoint Transformer is an active project for efficient 3D deep learning. The project implements two papers for semantic and panoptic segmentation, repsectively. This project is from the same authors as superpoint_graph and improves it in all regards: attention-based, faster, better performance, better codebase, more datasets, more pretrained models, new panoptic task...
Besides, this active project involves both academic and industrial users. It counts several undergoing extension projects that will likely offer more features in the near future: better cross-platform support, improved 3D pipeline, additional modalities.
Thanks for considering this PR :pray: