Efficient 3D Multi-View Multi-Object Tracking with Labeled Multi - Bernoulli Approximation

In this paper, we propose an efficient algorithm for 3D multi-view multi-object tracking. The method is based on labeled multi-Bernoulli approximation of a state-of-the-art 3D visual tracking based on generalized labeled multi-Bernoulli filter. Compare to the baseline, our method demonstrates the improved efficiency on real 3D pedestrian tracking datasets. If the detection quality is high, the tracking accuracy of our method is comparable to the baseline. Our source codes will be published at https://github.com/TranThienDat-Nguyen/3D-VisualTracking.