Question about using MS3D with Multi-Lidar Setup

[melihyazgan]

Hi @darrenjkt,

Thank you for your great effort in open-sourcing such a tool!

I have a setup with multiple lidars mounted in different locations on the car, and I’m working with rosbags from this setup. Based on the MS3D pipeline and the focus on single-lidar sequences, I wanted to get your intuition on the following:

• Do you think it’s better to fuse the point clouds from all lidars first and then create sequences for MS3D to process? OR
• Would it be more effective to create a separate dataset for each lidar and run MS3D labeling on them individually?

Best Regards, Melih