Problem
Research questions and friction points this paper is trying to address.
Cooperative 3D tracking
Multi-object tracking
Identity management
Cross-view appearance differences
Spatial misalignment
FFBL-Coop: Association-Decoupled Cooperative 3D Multi-Object Tracking
🤖 AI Summary
This study addresses the degradation of feature fusion and trajectory continuity in collaborative tracking caused by the coupling of evidence integration and identity inheritance. To overcome this, we propose a "fuse-then-bind" framework that introduces a novel association decoupling mechanism to separate query assignment from identity establishment. Additionally, a shared codebook is incorporated to compress transmission overhead while preserving accuracy. By integrating Transformer decoders, confidence-ranked slot admission, unified representation aggregation, and collaboration-prioritized identity anchoring, the proposed method achieves robust multi-agent tracking. Experimental results demonstrate that our approach attains AMOTA scores of 0.611 and 0.688 on the V2X-Seq and Griffin-25M datasets, respectively, significantly outperforming existing baselines.
📝 Abstract
Cooperative 3D tracking must integrate complementary observations across agents and time while maintaining consistent identities. When evidence integration and identity inheritance share a matching decision, errors arising from cross-view appearance differences and spatial misalignment can compromise both feature fusion and track continuity. We propose FFBL-Coop, a fuse first, bind later framework that separates instance admission from identity management. Confidence-ranked Slot Admission (CSA) allocates cooperative queries to available ego slots using confidence and spatial proximity. Unified Representation Aggregation (URA) uses cooperative semantic features and aligned anchors to guide ego-feature retrieval, refining the augmented query bank within a shared transformer decoder. After refinement, Cooperative-Priority Identity Anchoring (CPIA) combines learned association with persistent mappings to establish accepted identity assignments across frames. A shared codebook reduces transmitted payload while retaining AP and AMOTA close to the uncompressed variant. FFBL-Coop achieves AMOTA/AP of 0.611/0.548 on V2X-Seq and 0.688/0.653 on Griffin-25M. Code will be released.
Innovation
Methods, ideas, or system contributions that make the work stand out.
Cooperative 3D Tracking
Association-Decoupled
Slot Admission
Representation Aggregation
Shared Codebook
🔎 Similar Papers
No similar papers found.
