CenterTrack-EKF: 확장된 칼만 필터를 이용한 개선된 다중 객체 추적
(CenterTrack-EKF: Improved Multi Object Tracking with Extended Kalman Filter)
0
조회
0
다운로드
0
인용
0
참고문헌
키워드
초록
객체 궤적 모델링은 다중 객체 추적(Multi Object Tracking, MOT)의 주요 과제다. CenterTrack은 객체 중심 위치를 추적하는 Heatmap 기반의 방법으로 이를 해결하고자 했다. 하지만 복잡한 움직임과 비선형성을 가진 객체를 추적할 때 제한적인 성능을 보였다. 우리는 CenterTrack의 성능 저하 요인을 보행자의 동적 움직임으로 간주하여 확장된 칼만 필터(Extended Kalman Filter, EKF)를 CenterTrack에 통합했다. 우리가 제안하는 방법의 우수성을 입증하기 위해 기존 칼만 필터(Kalman Filter, KF)와 무향 칼만 필터(Unscented Kalman Filter, UKF)를 CenterTrack에 적용 후 다양한 데이터셋에 비교 평가했다. 실험결과, EKF를 CenterTrack에 통합했을 때 73.7% MOTA(Multiple Object Tracking Accuracy)를 달성하며 CenterTrack에 가장 적합한 필터임을 확인했다.
Abstract
Multi-Object trajectory modeling is a major challenge in MOT. CenterTrack tried to solve this problem with a Heatmap-based method that tracks the object center position. However, it showed limited performance when tracking objects with complex movements and nonlinearities. Considering the degradation factor of CenterTrack as the dynamic movement of pedestrians, we integrated the EKF into CenterTrack. To demonstrate the superiority of our proposed method, we applied the existing KF and UKF to CenterTrack and compared and evaluated it on various datasets. The experimental results confirmed that when EKF was integrated into CenterTrack, it achieved 73.7% MOTA, making it the most suitable filter for CenterTrack.
참고문헌 (19)
- "On the detection-to-track association for online multi-object tracking" (2021)[URL]
- "ReIDTrack : Multi-Object Track and Segmentation Without Motion" (2023)[URL]
- "TrajectoryFormer : 3D Object Tracking Transformer with Predictive Trajectory Hypotheses" (2023)[URL]
- Kaiwen Duan, "Centernet : Keypoint triplets for object detection", Proceedings of the IEEE, 2019, 6569~6578
- Xingyi Zhou, "Tracking objects as points", European Conference on Computer Vision Springer International Publishing, pp.474~490 (2020)
- RIBEIRO, "Kalman and Extended Kalman Filters : Concept, Derivation and Properties", Institute for Systems and Robotics, 43(46), pp.3736~3741 (2004)
- Simon Dan, "Kalman filtering", Embedded systems programming, 14(6), pp.72~79 (2001)
- Wan Eric, "The unscented Kalman filter for nonlinear estimation", Proceedings of the IEEE adaptive systems for signal processing, pp.153~158 (2000)
- Djuric Petar M, "Particle filtering", IEEE signal processing magazine, 5(20), pp.19~38 (2003)
- Jie Cao, "RetinaMOT : rethinking anchor-free YOLOv5 for online multiple object tracking", Complex & Intelligent Systems, pp.1~19 (2023)
- "TransReID : Transformer-based Object Re-Identification" (2021)[URL]
- "MOT16 : A benchmark for multi-object tracking" (2016)[URL]
- Geiger Andreas, "Are we ready for autonomous driving? the kitti vision benchmark suite", IEEE conference on computer vision and pattern recognition, pp.3354~3361 (2012)
- Bergmann Philipp, "Tracking without bells and whistles", Proceedings of the IEEE, 2019, 941~951
- Jialian Wu, "Track To Detect and Segment : An Online Multi-Object Tracker", Proceedings of the IEEE, 2021, 12352~12361
- Pang Jiangmiao, "Quasi-dense similarity learning for multiple object tracking", Proceedings of the IEEE, 2021, ~
- Fangao Zeng, "MOTR : End-to-End Multiple-Object Tracking with Transformer", European Conference on Computer Vision, pp.659~675 (2022)
- Yifu Zhang, "Fairmot : On the fairness of detection and re-identification in multiple object tracking", International Journal of Computer Vision, pp.3069~3087 (2021)
- Zhang Yifu, "Bytetrack : Multi-object tracking by associating every detection box", European Conference on Computer Vision, pp.1~21 (2022)