{"type":"ScholarlyArticle","title":"BankTweak: Adversarial Attack Against Multi-Object Trackers by Manipulating Feature Banks","authors":[{"name":"Woojin Shin","slug":"woojin-shin","given":"Woojin","family":"Shin","affiliation":""},{"name":"Donghwa Kang","slug":"donghwa-kang","given":"Donghwa","family":"Kang","affiliation":""},{"name":"Daejin Choi","slug":"daejin-choi","given":"Daejin","family":"Choi","affiliation":""},{"name":"Brent Byunghoon Kang","slug":"brent-byunghoon-kang","given":"Brent Byunghoon","family":"Kang","affiliation":""},{"name":"Jinkyu Lee","slug":"jinkyu-lee","given":"Jinkyu","family":"Lee","affiliation":""},{"name":"Hyeongboo Baek","slug":"hyeongboo-baek","given":"Hyeongboo","family":"Baek","affiliation":""}],"abstract":"Modern multi-object tracking (MOT) predominantly relies on the tracking-by-detection paradigm to construct object trajectories.  Traditional MOT attacks primarily degrade detection quality in specific frames only, lacking efficiency, while state-of-the-art (SOTA) approaches induce persistent identity (ID) switches by manipulating object positions during the association phase, even after the attack ends. In this paper, we reveal that these SOTA attacks can be easily counteracted by adjusting distance-related parameters in the association phase, exposing their lack of robustness. To overcome these limitations, we propose BankTweak, a novel adversarial attack targeting feature-based MOT systems to induce persistent ID switches (efficiency) without modifying object positions (robustness). BankTweak exploits a critical vulnerability in the Hungarian matching algorithm of MOT systems by strategically injecting altered features into feature banks during the association phase.  Extensive experiments on MOT17 and MOT20 datasets, combining various detectors, feature extractors, and trackers, demonstrate that BankTweak significantly outperforms SOTA attacks up to 11.8 times, exposing fundamental vulnerabilities in the tracking-by-detection framework.","doi":"10.24963/ijcai.2025/206","url":"https://ijcai.org/proceedings/2025/206","pdf":"https://ijcai.org/proceedings/2025/0206.pdf","markdown":"https://ijcai.org/proceedings/2025/206.md","bibtex":"https://ijcai.org/proceedings/2025/bibtex/206","volume":{"slug":"2025","title":"Proceedings of the Thirty-Fourth International Joint Conference on Artificial Intelligence","short_title":"IJCAI 2025","url":"https://ijcai.org/proceedings/2025/","year":2025,"location":"Montreal, Canada","dates":"16-22 August 2025","editors":"James Kwok","isbn":"978-1-956792-06-5","publisher":"International Joint Conferences on Artificial Intelligence Organization","doi_prefix":"10.24963/ijcai.2025"},"section":"Computer Vision","track":"Main Track","number":206,"pages":{"first":1847,"last":1855},"keywords":["Multidisciplinary Topics and Applications: MTA: Security and privacy","Computer Vision: CV: Recognition (object detection, categorization)","Computer Vision: CV: Adversarial learning, adversarial attack and defense methods"],"videos":[],"references":[],"license_note":"Copyright International Joint Conferences on Artificial Intelligence."}