Point Transformer V3 Extreme: 1st Place Solution for 2024 Waymo Open Dataset Challenge in Semantic Segmentation

TL;DR

This paper presents the winning solution for the 2024 Waymo Open Dataset Challenge in semantic segmentation, enhancing Point Transformer V3 with advanced training, inference techniques, and ensemble strategies to achieve top performance.

Contribution

The paper introduces Point Transformer V3 Extreme, a significantly improved model with multi-frame training, no-clipping policy, and ensemble methods, setting new state-of-the-art results.

Findings

Achieved 1st place in the 2024 Waymo Open Dataset Challenge.

Significant performance improvements over the original PTv3.

Effective use of multi-frame training and model ensembling.

Abstract

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.