ReIDTracker_Sea: Multi-Object Tracking in Maritime Computer Vision

Kaer Huang, Weitu Chong, Hui Yang, Kanokphan Lertniphonphan, Jun Xie, Feng Chen

Abstract: In recent years, we've witnessed the remarkable growth of computer vision applications in both maritime and fresh water domains. These technologies have played pivotal roles in search and rescue (SaR), the detection of illegal fishing, airborne water surface reconnaissance, offshore wind farm and oil rig inspections, animal population monitoring, and beyond. Multi-Object Tracking is one of the most important technologies in maritime computer vision. Our paper tries to explore Multi-Object Tracking in maritime Unmanned Aerial vehicles (UAVs) and Unmanned Surface Vehicles (USVs). Most of the current Multi-Object Tracking algorithms require complex association strategies and association information (2D location and motion, 3D motion, 3D depth, 2D appearance) to achieve better performance, which makes the entire tracking system extremely complex and heavy. At the same time, most of the current Multi-Object Tracking algorithms still require video annotation data which is costly to obtain for training. Our paper tries to explore Multi-Object Tracking in a completely unsupervised way. The scheme accomplishes instance representation learning by using self-supervision on ImageNet. Then, by cooperating with high-quality detectors, the multitarget tracking task can be completed simply and efficiently. The scheme achieved state-of-the-art performance on both UAV-based Multi-Object Tracking with Reidentification and USV-based Multi-Object Tracking benchmarks, and the solution won the championship in many multiple Multi-Object Tracking competitions. They include the championship of CVPR2022 WAD BDD MOT Challenges, 4 Championships of ECCV2022 SSLAD BDD MOT/MOTS/SSMOT/SSMOTS, and the latest CVPR2023 WAD BDD MOTS Challenges.