Semantic-aware DropSplat: Adaptive Pruning of Redundant Gaussians for 3D Aerial-View Segmentation

TL;DR

SAD-Splat is a novel method for 3D aerial-view segmentation that adaptively prunes redundant Gaussian points using semantic confidence and sparsity, enhanced by pseudo-labels from 2D models, and validated on a new challenging dataset.

Contribution

This work introduces SAD-Splat, a Gaussian point drop module with learnable sparsity and a pseudo-label pipeline, plus a new 3D aerial semantic dataset, advancing 3D aerial scene segmentation.

Findings

Achieves a good balance between accuracy and compactness.

Effectively reduces redundant and ambiguous points.

Improves segmentation with pseudo-labels from 2D foundation models.

Abstract

In the task of 3D Aerial-view Scene Semantic Segmentation (3D-AVS-SS), traditional methods struggle to address semantic ambiguity caused by scale variations and structural occlusions in aerial images. This limits their segmentation accuracy and consistency. To tackle these challenges, we propose a novel 3D-AVS-SS approach named SAD-Splat. Our method introduces a Gaussian point drop module, which integrates semantic confidence estimation with a learnable sparsity mechanism based on the Hard Concrete distribution. This module effectively eliminates redundant and semantically ambiguous Gaussian points, enhancing both segmentation performance and representation compactness. Furthermore, SAD-Splat incorporates a high-confidence pseudo-label generation pipeline. It leverages 2D foundation models to enhance supervision when ground-truth labels are limited, thereby further improving segmentation accuracy. To advance research in this domain, we introduce a challenging benchmark dataset: 3D Aerial Semantic (3D-AS), which encompasses diverse real-world aerial scenes with sparse annotations. Experimental results demonstrate that SAD-Splat achieves an excellent balance between segmentation accuracy and representation compactness. It offers an efficient and scalable solution for 3D aerial scene understanding.