SUG-Occ: Explicit Semantics and Uncertainty Guided Sparse Learning for Efficient 3D Occupancy Prediction

TL;DR

SUG-Occ introduces a sparse learning framework for 3D occupancy prediction that leverages explicit semantics and uncertainty to improve efficiency and accuracy in autonomous driving scenarios.

Contribution

The paper presents a novel sparse learning approach utilizing semantic and uncertainty priors, explicit unsigned distance encoding, and an object contextual representation for efficient 3D occupancy prediction.

Findings

Outperforms baselines on SemanticKITTI and Occ3D-Nuscenes datasets.

Achieves better accuracy and efficiency in 3D occupancy prediction.

Effectively exploits scene sparsity to reduce computational cost.

Abstract

3D semantic occupancy prediction has emerged as a critical perception task for autonomous driving due to its ability to offer voxel-level semantic and geometric understanding of the environment. However, such a refined representation for large-scale scenes incurs prohibitive computation, posing a significant challenge to practical real-time deployment. To address this, we propose SUGOcc, an explicit semantics and uncertainty guided sparse learning framework for efficient occupancy prediction, which exploits the inherent sparsity of 3D scenes to reduce redundant computation while maintaining geometric and semantic integrity. Specifically, we first utilize semantic and uncertainty priors to suppress image projections from free space while employing explicit unsigned distance encoding to enhance geometric consistency, thereby producing a structurally sparse representation. Secondly, we introduce a cascade sparse completion module to enable efficient coarse-to-fine reasoning over the sparse representation via hyper cross sparse convolution, generative upsampling and adaptive pruning. Finally, we propose an object contextual representation (OCR) based mask decoder that refines the voxel-wise predictions through lightweight query-context interactions, thereby avoiding expensive attention operations over volumetric features. Extensive experiments on SemanticKITTI and Occ3D-Nuscenes benchmark demonstrate that the proposed approach outperforms the baselines, achieving notable improvements in both accuracy and efficiency across datasets.