B-on-Track: A Novel Evaluation Metric for Multiple Object Tracking
2023 18TH INTERNATIONAL JOINT SYMPOSIUM ON ARTIFICIAL INTELLIGENCE AND NATURAL LANGUAGE PROCESSING, ISAI-NLP(2023)
Abstract
Multi-object tracking (MOT) is essential in various applications. Still, more effective evaluation metrics are often needed to measure a tracker's ability to consistently follow an object without losing track and minimize re-identification issues. Addressing this precise problem, we present B-on-Track, a novel evaluation metric formulated to gauge a tracker's competence in maintaining continuous tracking without loss and reducing re-identification challenges. By implementing and evaluating the B-on-Track metric on the MOTChallenge Benchmark, this paper thoroughly compares with existing evaluation metrics, elucidating the distinctive benefits and potential applications of the B-on-Track metric in the MOT field. Detailed interpretations of the B-on-Track metric further demonstrate its robustness and innovation in filling the existing gaps in MOT evaluation.
MoreTranslated text
Key words
Multiple Object Tracking (MOT),Evaluation Metric,Object Detection
AI Read Science
Must-Reading Tree
Example
![](https://originalfileserver.aminer.cn/sys/aminer/pubs/mrt_preview.jpeg)
Generate MRT to find the research sequence of this paper
Chat Paper
Summary is being generated by the instructions you defined