Question About Camera Only Inference with NuScenes Data

[NathanZhu8]

Hi! Thank you for your amazing work on this project!

I have a question about using camera data only. I am trying to perform inference using only the NuScenes camera data within pretrained checkpoint in readme directly. However, it seems the framework still requires LiDAR data.

Could this be due to a config issue, or does the code need modification to support camera-only inference? Any guidance or suggestions would be greatly appreciated.

Thank you in advance!

[ziqipang]

@NathanZhu8 Yes, You can disable LiDAR in the config or code, I think. I haven't worked on this direction for a while, so I cannot remember the specifics.

However, I do suggest adding LiDAR to run if it does not take tremendous effort. From my experience, LiDAR files take much smaller space than images, and they are useful for visualization.

Please let me know how these options work out for you.