Eaphan
LiDAR-only training and integration with non-pretrained baseline models
Description: Hello, I'm trying to train GLENet using a LiDAR-only configuration, but I’m having some trouble understanding how to correctly set it up and how to use it with other baseline models such as PointPillars.
Specifically, I would like some guidance on:
How to configure the model and dataset pipeline for LiDAR-only training.
Whether there are recommended changes to the config files or model architecture when not using camera data, or when using other models apart from the pretrained ones.
What I’ve Tried: I tried using the KITTI format setup with dummy data in the camera and image folders.
Questions:
Is there an official or recommended way to train GLENet with LiDAR-only data?
Are there any example configs or scripts I can refer to for GLENet's integration with other OpenPCDet models?
(Maintainers: Please tag this issue with 'help wanted' and 'question' if appropriate.)
Thank you!
