Tracking Any Point with Frame-Event Fusion Network at High Frame Rate
FE-TAP is an image-event fusion point tracker that combines image frame context with high temporal resolution event data for robust tracking. It introduces EvoFusion to model image evolution guided by events and uses a transformer-based refinement strategy to update point trajectories iteratively. This method achieves high frame rates and outperforms state-of-the-art trackers, improving feature age by 24% on EDS datasets.