KPConv-PyTorch icon indicating copy to clipboard operation
KPConv-PyTorch copied to clipboard

About multi-scans task on SemanticKITTI

Open liamlin5566 opened this issue 3 years ago • 5 comments

Thanks for sharing your work. I find that you have the result of multiple scan task on SemanticKITTI. I wonder know what number you set for "n_frames" to merge multiple scans for this task.

liamlin5566 avatar Feb 21 '23 08:02 liamlin5566

I didn't experiment a lot with SemanticKitti actually. I only tried 3 and 5 and did not optimize all parameters. 5 was slightly better, but also slower and more memory-consuming.

HuguesTHOMAS avatar Feb 21 '23 21:02 HuguesTHOMAS

OK, thanks for your replying

liamlin5566 avatar Mar 09 '23 03:03 liamlin5566

Sorry to bother again, I wonder if the number you set for "n_frames" is 5 on the leaderboard, because I want a fair comparison.

liamlin5566 avatar Mar 22 '23 17:03 liamlin5566

No, on the leaderboard, I was using n_frames=3

HuguesTHOMAS avatar Mar 22 '23 18:03 HuguesTHOMAS

How about your validation mIoU? Thanks.

liamlin5566 avatar Mar 23 '23 12:03 liamlin5566