CropTrack: A Tracking with Re-Identification Framework for Precision Agriculture
Researchers have developed CropTrack, a new multiple-object tracking framework designed for precision agriculture. This system addresses challenges like similar object appearances and occlusions by integrating both appearance and motion information, unlike traditional methods that rely solely on motion. CropTrack employs enhanced appearance association strategies and a feature bank to improve identity preservation, demonstrating superior accuracy and fewer identity switches on agricultural datasets compared to existing state-of-the-art trackers. AI