SPOT-Occ: Sparse Prototype-guided Transformer for Camera-based 3D Occupancy Prediction

TL;DR

SPOT-Occ introduces a prototype-guided sparse transformer decoder for efficient, accurate 3D occupancy prediction from camera data, enhancing autonomous vehicle safety with real-time performance.

Contribution

The paper proposes a novel prototype-based sparse transformer decoder with a guided feature selection mechanism and denoising paradigm, improving efficiency and accuracy in 3D occupancy prediction.

Findings

Outperforms previous methods in speed and accuracy

Uses a two-stage prototype-guided feature aggregation

Leverages ground-truth masks for stable query-prototype association

Abstract

Achieving highly accurate and real-time 3D occupancy prediction from cameras is a critical requirement for the safe and practical deployment of autonomous vehicles. While this shift to sparse 3D representations solves the encoding bottleneck, it creates a new challenge for the decoder: how to efficiently aggregate information from a sparse, non-uniformly distributed set of voxel features without resorting to computationally prohibitive dense attention. In this paper, we propose a novel Prototype-based Sparse Transformer Decoder that replaces this costly interaction with an efficient, two-stage process of guided feature selection and focused aggregation. Our core idea is to make the decoder's attention prototype-guided. We achieve this through a sparse prototype selection mechanism, where each query adaptively identifies a compact set of the most salient voxel features, termed prototypes, for focused feature aggregation. To ensure this dynamic selection is stable and effective, we introduce a complementary denoising paradigm. This approach leverages ground-truth masks to provide explicit guidance, guaranteeing a consistent query-prototype association across decoder layers. Our model, dubbed SPOT-Occ, outperforms previous methods with a significant margin in speed while also improving accuracy. Source code is released at https://github.com/chensuzeyu/SpotOcc.