Improved multi-object tracking by associating all detection boxes
Improved multi-object tracking by associating all detection boxes
ByteTrack: Multi-Object Tracking by Associating Every Detection Box
arXiv paper abstract https://arxiv.org/abs/2110.06864
arXiv PDF paper https://arxiv.org/pdf/2110.06864.pdf
GitHub https://github.com/ifzhang/ByteTrack
Multi-object tracking (MOT) aims at estimating bounding boxes and identities of objects in videos.
Most methods … objects with low detection scores, e.g. occluded objects, are simply thrown away, which brings non-negligible true object missing and fragmented trajectories.
… present a simple, effective and generic association method, called BYTE, tracking BY associaTing Every detection box instead of only the high score ones.
… utilize their similarities with tracklets to recover true objects and filter out the background detections.
… achieve 80.3 MOTA, 77.3 IDF1 and 63.1 HOTA on the test set of MOT17 with 30 FPS running speed on a single V100 GPU. …
Stay up to date. Subscribe to my posts https://morrislee1234.wixsite.com/website/contact
Web site with my other posts by category https://morrislee1234.wixsite.com/website
LinkedIn https://www.linkedin.com/in/morris-lee-47877b7b
