Fully Distributed Multi-View 3D Tracking in Real-Time
About
Multi-camera tracking with overlapping fields of view typically relies on centralized fusion, which creates computational bottlenecks that prevent deployment at scale. We present MV3DT, a fully distributed framework for real-time multi-view 3D tracking that achieves accurate identity propagation and occlusion recovery through peer-to-peer coordination, eliminating the need for central aggregation. Each camera node executes a lightweight modular pipeline comprising monocular 3D perception, distributed multi-view association, and collaborative fusion via lightweight messaging. MV3DT achieves 96.5% IDF1, 93.1% MOTA, and 94.6% MOTP on WILDTRACK, competitive with state-of-the-art centralized methods, and unprecedented 41.7% IDF1 and 50.9% MOTA on SCOUT while demonstrating superior scalability: sustaining 30 FPS on 100 cameras with <10ms inter-camera latency and only 2.2% communication overhead. MV3DT operates in a zero-shot regime given camera calibrations, requiring no scene-specific learning and making it directly deployable in new environments. These results establish MV3DT as a practical solution for real-time multi-view tracking in large-scale overlapping camera networks.
Key insights
What problem does MV3DT solve?
Multi-camera 3D tracking with overlapping views usually needs a central fusion server, creating bottlenecks. MV3DT makes each camera an independent node that shares lightweight peer-to-peer messages for distributed 3D tracking without central aggregation.
How does MV3DT differ from prior work?
Unlike prior distributed trackers operating in 2D image space, MV3DT uses 3D ground-plane geometry. On WILDTRACK it ties best IDF1 at 96.5% and on SCOUT it reaches 41.7% IDF1 versus 27.0% for UMPN+SP.
What are the practical scalability findings?
On a 100-camera warehouse simulation, MV3DT sustained 30 FPS with 5 ms average inter-camera latency and 2.2% communication overhead, demonstrating a scalable real-time prototype for large networks.
What limitations do the authors note?
MV3DT requires calibrated overlapping cameras and nominal anthropometric assumptions. Its accuracy was evaluated only on specific benchmark scenes, trading marginal MOTA for deployability versus learned methods.