Co-Fusion4D: Spatio-temporal Collaborative Fusion for Robust 3D Object Detection

2026-05-19

AI Summary

This work addresses the spatiotemporal inconsistency in bird’s-eye-view (BEV) object detection for autonomous driving, caused by the motion of both the ego-vehicle and surrounding objects across frames. To tackle this challenge, the authors propose Co-Fusion4D, a novel framework that centers on the current frame and integrates historical information through spatiotemporal alignment and filtering. The method introduces an innovative leader-complement fusion mechanism and a dual-attention fusion (DAF) module, which adaptively enhance regions with consistent motion while suppressing spurious associations and accumulated errors—departing from conventional uniform fusion paradigms. Evaluated on the nuScenes benchmark, Co-Fusion4D achieves state-of-the-art performance with a single model, attaining 74.9% mAP and 75.6% NDS, significantly improving the temporal stability and discriminative power of BEV representations.

Abstract

In autonomous driving, 3D object detection is essential for accurate perception and reliable decision-making. However, object motion and ego-motion often induce cross-frame spatiotemporal inconsistencies in BEV-based detectors, leading to temporal BEV feature misalignment and degraded spatiotemporal consistency. To address these challenges, we propose Co-Fusion4D, a unified framework that explicitly preserves cross-frame spatiotemporal consistency and suppresses temporal feature drift. Co-Fusion4D adopts a current-frame-centric strategy, treating the current frame as the primary source of information while selectively incorporating historical frames after spatiotemporal filtering and alignment. This dominant-complementary mechanism effectively mitigates cumulative alignment errors, suppresses noisy feature propagation, and exploits reliable temporal cues for a more consistent BEV representation. In addition, Co-Fusion4D integrates a Dual Attention Fusion (DAF) module to further enhance spatiotemporal feature interaction. DAF jointly leverages intra-frame spatial attention and inter-frame temporal attention to adaptively align and fuse multi-frame features, emphasizing motion-consistent regions while suppressing spurious correlations. By departing from conventional uniform fusion paradigms, this design substantially improves the temporal stability and discriminative capability of BEV representations. Extensive experiments on the nuScenes benchmark demonstrate that Co-Fusion4D achieves state-of-the-art performance, with 74.9% mAP and 75.6% NDS, without relying on test-time augmentation or external data.

Problem

• 3D object detection
• spatiotemporal consistency
• BEV feature alignment
• temporal feature drift
• autonomous driving

Innovation

• Spatio-temporal Fusion
• BEV Representation
• Dual Attention Fusion
• 3D Object Detection
• Temporal Consistency