Selective Frame Analysis for Efficient Object Tracking: Balancing Speed with Accuracy in MOT Systems

Yubraj Jung Shah, Yi Guo, Laurence A. F. Park, Oliver Obst

Abstract: Applications such as autonomous driving and video surveillance rely on Multiple Object Tracking (MOT) technology to accurately identify objects in video data. Real-time MOT systems are often challenged for continuously improving computational efficiency while maintaining the acceptable level of accuracy. For every frame, advanced algorithms for detection and tracking is used to identify and track objects. However, it is still unclear if they should be used across all frames, or whether it would be better to use the algorithm only for selected frames. This is an empirical question, best answered by experimental research.

External IDs:doi:10.1007/978-981-96-0692-4_20