Deflickering Vision-Based Occupancy Networks through Lightweight Spatio-Temporal Correlation

TL;DR

This paper introduces OccLinker, a lightweight plugin for vision-based occupancy networks that enhances temporal consistency by consolidating historical cues and learning sparse correlations, significantly reducing flickering artifacts in 3D environment reconstructions.

Contribution

The paper presents a novel plugin framework, OccLinker, that improves temporal coherence in VONs with minimal computational cost by integrating historical information through a dual cross-attention mechanism.

Findings

Achieves superior flickering reduction on benchmark datasets.

Maintains high computational efficiency compared to existing methods.

Effectively consolidates static and motion cues for better 3D reconstruction.

Abstract

Vision-based occupancy networks (VONs) provide an end-to-end solution for reconstructing 3D environments in autonomous driving. However, existing methods often suffer from temporal inconsistencies, manifesting as flickering effects that degrade temporal coherence and adversely affect downstream decision-making. While recent approaches incorporate historical information to alleviate this issue, they often incur high computational costs and may introduce misaligned or redundant features that interfere with object detection. We propose OccLinker, a novel plugin framework that can be easily integrated into existing VONs to improve performance. Our method efficiently consolidates historical static and motion cues, learns sparse latent correlations with current features through a dual cross-attention mechanism, and generates correction occupancy components to refine the base network predictions. In addition, we introduce a new temporal consistency metric to quantitatively measure flickering effects. Extensive experiments on two benchmark datasets demonstrate that our method achieves superior performance with minimal computational overhead while effectively reducing flickering artifacts.