RT-GS2: Real-Time Generalizable Semantic Segmentation for 3D Gaussian Representations of Radiance Fields

TL;DR

RT-GS2 is the first real-time, generalizable semantic segmentation method for 3D Gaussian radiance fields, significantly improving segmentation accuracy and speed over previous scene-specific approaches.

Contribution

Introduces RT-GS2, a novel method that enables real-time, scene-generalizable semantic segmentation using Gaussian Splatting with view-independent features and a new fusion technique.

Findings

Achieves 8.01% higher mIoU on the Replica dataset.

Runs at 27.03 FPS, 901 times faster than prior methods.

Outperforms state-of-the-art in semantic segmentation quality.

Abstract

Gaussian Splatting has revolutionized the world of novel view synthesis by achieving high rendering performance in real-time. Recently, studies have focused on enriching these 3D representations with semantic information for downstream tasks. In this paper, we introduce RT-GS2, the first generalizable semantic segmentation method employing Gaussian Splatting. While existing Gaussian Splatting-based approaches rely on scene-specific training, RT-GS2 demonstrates the ability to generalize to unseen scenes. Our method adopts a new approach by first extracting view-independent 3D Gaussian features in a self-supervised manner, followed by a novel View-Dependent / View-Independent (VDVI) feature fusion to enhance semantic consistency over different views. Extensive experimentation on three different datasets showcases RT-GS2's superiority over the state-of-the-art methods in semantic segmentation quality, exemplified by a 8.01% increase in mIoU on the Replica dataset. Moreover, our method achieves real-time performance of 27.03 FPS, marking an astonishing 901 times speedup compared to existing approaches. This work represents a significant advancement in the field by introducing, to the best of our knowledge, the first real-time generalizable semantic segmentation method for 3D Gaussian representations of radiance fields.