Robust Multi-Object Tracking With Local Appearance and Stable Motion Models
Multi-object tracking (MOT) has been steadily studied for video understanding in computer vision. However, existing MOT frameworks usually employ straightforward appearance or motion models and may struggle in dynamic environments with similar appearance and complex motion. In this paper, we present...
Main Authors: | , , , , |
---|---|
Format: | Article |
Language: | English |
Published: |
IEEE
2023-01-01
|
Series: | IEEE Access |
Subjects: | |
Online Access: | https://ieeexplore.ieee.org/document/10187121/ |
_version_ | 1797768419025616896 |
---|---|
author | Jubi Hwang Kyujin Shim Kangwook Ko Namkoo Ha Changick Kim |
author_facet | Jubi Hwang Kyujin Shim Kangwook Ko Namkoo Ha Changick Kim |
author_sort | Jubi Hwang |
collection | DOAJ |
description | Multi-object tracking (MOT) has been steadily studied for video understanding in computer vision. However, existing MOT frameworks usually employ straightforward appearance or motion models and may struggle in dynamic environments with similar appearance and complex motion. In this paper, we present a robust MOT framework with local appearance and stable motion models to overcome these two hindrances. The framework incorporates object and local part detectors, a feature extractor, a keypoint extractor, and a data association method. For the data association, we utilize five types of similarity metrics and a cascaded matching strategy. The local appearance model is suggested to be used additionally with global appearance features of full bounding boxes to obtain discriminative features even for objects with a similar appearance. At the same time, the stable motion model considers the core of the body as the central point of the object and subdivides the body using a novel 12-tuple Kalman state vector to analyze complex motion. As a result, our new tracker achieves state-of-the-art performance on the DanceTrack test set, surpassing all other listed tracking systems in terms of both detection and tracking quality metrics, obtaining 61.3 HOTA, 82.3 DetA, 45.8 AssA, and 91.7 MOTA. The source code is available at <uri>https://github.com/Jubi-Hwang/Robust-MOT-with-Local-Appearance-and-Stable-Motion-Models</uri>. |
first_indexed | 2024-03-12T20:53:16Z |
format | Article |
id | doaj.art-88884750e56949168a95f35af9f4b1b6 |
institution | Directory Open Access Journal |
issn | 2169-3536 |
language | English |
last_indexed | 2024-03-12T20:53:16Z |
publishDate | 2023-01-01 |
publisher | IEEE |
record_format | Article |
series | IEEE Access |
spelling | doaj.art-88884750e56949168a95f35af9f4b1b62023-07-31T23:01:21ZengIEEEIEEE Access2169-35362023-01-0111770237703310.1109/ACCESS.2023.329673110187121Robust Multi-Object Tracking With Local Appearance and Stable Motion ModelsJubi Hwang0https://orcid.org/0009-0003-0639-8806Kyujin Shim1https://orcid.org/0000-0003-3015-8725Kangwook Ko2Namkoo Ha3Changick Kim4https://orcid.org/0000-0001-9323-8488School of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of KoreaSchool of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of KoreaSchool of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of KoreaEO/IR Systems Research and Development Laboratory, LIG Nex1 Company Ltd., Yongin, Republic of KoreaSchool of Electrical Engineering, Korea Advanced Institute of Science and Technology (KAIST), Daejeon, Republic of KoreaMulti-object tracking (MOT) has been steadily studied for video understanding in computer vision. However, existing MOT frameworks usually employ straightforward appearance or motion models and may struggle in dynamic environments with similar appearance and complex motion. In this paper, we present a robust MOT framework with local appearance and stable motion models to overcome these two hindrances. The framework incorporates object and local part detectors, a feature extractor, a keypoint extractor, and a data association method. For the data association, we utilize five types of similarity metrics and a cascaded matching strategy. The local appearance model is suggested to be used additionally with global appearance features of full bounding boxes to obtain discriminative features even for objects with a similar appearance. At the same time, the stable motion model considers the core of the body as the central point of the object and subdivides the body using a novel 12-tuple Kalman state vector to analyze complex motion. As a result, our new tracker achieves state-of-the-art performance on the DanceTrack test set, surpassing all other listed tracking systems in terms of both detection and tracking quality metrics, obtaining 61.3 HOTA, 82.3 DetA, 45.8 AssA, and 91.7 MOTA. The source code is available at <uri>https://github.com/Jubi-Hwang/Robust-MOT-with-Local-Appearance-and-Stable-Motion-Models</uri>.https://ieeexplore.ieee.org/document/10187121/Multi-object trackingtracking-by-detectionsimilarity metricsmatching strategy |
spellingShingle | Jubi Hwang Kyujin Shim Kangwook Ko Namkoo Ha Changick Kim Robust Multi-Object Tracking With Local Appearance and Stable Motion Models IEEE Access Multi-object tracking tracking-by-detection similarity metrics matching strategy |
title | Robust Multi-Object Tracking With Local Appearance and Stable Motion Models |
title_full | Robust Multi-Object Tracking With Local Appearance and Stable Motion Models |
title_fullStr | Robust Multi-Object Tracking With Local Appearance and Stable Motion Models |
title_full_unstemmed | Robust Multi-Object Tracking With Local Appearance and Stable Motion Models |
title_short | Robust Multi-Object Tracking With Local Appearance and Stable Motion Models |
title_sort | robust multi object tracking with local appearance and stable motion models |
topic | Multi-object tracking tracking-by-detection similarity metrics matching strategy |
url | https://ieeexplore.ieee.org/document/10187121/ |
work_keys_str_mv | AT jubihwang robustmultiobjecttrackingwithlocalappearanceandstablemotionmodels AT kyujinshim robustmultiobjecttrackingwithlocalappearanceandstablemotionmodels AT kangwookko robustmultiobjecttrackingwithlocalappearanceandstablemotionmodels AT namkooha robustmultiobjecttrackingwithlocalappearanceandstablemotionmodels AT changickkim robustmultiobjecttrackingwithlocalappearanceandstablemotionmodels |