2000 character limit reached
TAPL: Dynamic Part-based Visual Tracking via Attention-guided Part Localization (2110.13027v1)
Published 25 Oct 2021 in cs.CV
Abstract: Holistic object representation-based trackers suffer from performance drop under large appearance change such as deformation and occlusion. In this work, we propose a dynamic part-based tracker and constantly update the target part representation to adapt to object appearance change. Moreover, we design an attention-guided part localization network to directly predict the target part locations, and determine the final bounding box with the distribution of target parts. Our proposed tracker achieves promising results on various benchmarks: VOT2018, OTB100 and GOT-10k