Point Transformer V3 Extreme: 1st Place Solution for 2024 Waymo Open Dataset Challenge in Semantic Segmentation
This work addresses the problem of 3D semantic segmentation for autonomous driving systems, representing an incremental improvement on existing methods for a specific benchmark.
The authors tackled the semantic segmentation task on the 2024 Waymo Open Dataset Challenge by enhancing Point Transformer V3 with multi-frame training and a no-clipping-point policy, achieving first place on the leaderboard with substantial performance gains over the original model.
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.