Code open-source date?

[songwei100110]

Code open-source date?

[ldkong1205]

Hi @swsongwei, thanks for your interest in this work!

We have recently open-sourced the code for:

• Generating semantic superpixels using SLIC, SAM, and SEEM.
• Mapping 2D semantic superpixels to 3D semantic superpoints on nuScenes. See this file for more details.

Up till now, you should be able to reproduce our baseline results reported in the paper. That is, you can follow the SLidR pertaining pipeline, along with our semantic superpixels and superpoints, for camera-to-LiDAR representation learning on nuScenes. We will also incorporate the source code from SLidR in this repository to facilitate such implementation.

In the near future, we will release:

• The code for generating semantic superpixels using more efficient vision foundation models, such as FastSAM, Semantic-SAM, and MobileSAM.
• The code of our spatial and temporal consistency regularization framework.

Please stay tuned!

[rockywind]

@ldkong1205 您好，请教个问题，
这个分割出来的点云类别映射表是什么？打印出来的类别范围非常大，从-1到361。