2407.15282

Total: 1

#1 Point Transformer V3 Extreme: 1st Place Solution for 2024 Waymo Open Dataset Challenge in Semantic Segmentation [PDF6] [Copy] [Kimi2] [REL]

Authors: Xiaoyang Wu, Xiang Xu, Lingdong Kong, Liang Pan, Ziwei Liu, Tong He, Wanli Ouyang, Hengshuang Zhao

In this technical report, we detail our first-place solution for the 2024 Waymo Open Dataset Challenge's semantic segmentation track. We significantly enhanced the performance of Point Transformer V3 on the Waymo benchmark by implementing cutting-edge, plug-and-play training and inference technologies. Notably, our advanced version, Point Transformer V3 Extreme, leverages multi-frame training and a no-clipping-point policy, achieving substantial gains over the original PTv3 performance. Additionally, employing a straightforward model ensemble strategy further boosted our results. This approach secured us the top position on the Waymo Open Dataset semantic segmentation leaderboard, markedly outperforming other entries.

Subject: Computer Vision and Pattern Recognition

Publish: 2024-07-21 22:08:52 UTC