https://arxiv.org/abs/2110.06864 ByteTrack: Multi-Object Tracking by Associating Every Detection Box Multi-object tracking (MOT) aims at estimating bounding boxes and identities of objects in videos. Most methods obtain identities by associating detection boxes whose scores are higher than a threshold. The objects with low detection scores, e.g. occluded arxiv.org 이 논문은 data association 기법을 제안하는..
https://arxiv.org/abs/2012.15460 TransTrack: Multiple Object Tracking with Transformer In this work, we propose TransTrack, a simple but efficient scheme to solve the multiple object tracking problems. TransTrack leverages the transformer architecture, which is an attention-based query-key mechanism. It applies object features from the previ arxiv.org 이번에 리뷰할 논문은 2D MOT 논문으로 transformer를 이용한 Joi..
MOTA(Multi-Object Tracking Accuracy) $m_t$ : number of misses(놓친 객체 수) == FN $fp_t$ : number of false positives(잘못 등록한 객체 수) == FP $mme_t$ : number of ID switches(ID가 잘못 부여된 객체 수) $g_t$ : number of objects(모든 객체 수) 범위 : -∞ ~ 1 threshold에 따른 MOTA를 평균을 취한 것 : AMOTA 추적 성능을 평가하는데 주로 사용 MOTP(Multi-Object Tracking Precision) $d^i_t$ : 추적된 객체와 실제 위치 간 거리 $c_t$ : 정확하게 추적된 객체 수 범위 : 0 ~ ∞ threshold에 따른 M..
Joint Detection and Embedding은 주로 이미지를 대상으로 하는 2D MOT에서 사용된다. 3D MOT에 JDE를 적용하는데 겪는 어려움 및 한계를 알아보자. 1. Poly-MOT: A Polyhedral Framework For 3D Multi-Object Tracking Due to the data-driven nature of JDT, it is generally less precise and robust than TBD, and consequently, the majority of 3D MOT approaches adhere to the TBD architecture. 2. SimpleTrack : Understanding and Rethinking 3D Multi-object..
