Whareformer: Online Object Tracking for Long Egocentric Videos

dimadamen · x · 2026-07-10

This post introduces Whareformer, an upcoming paper accepted by ECCV 2026, which presents an online learning tracking method tailored for long egocentric videos. The approach combines appearance and 3D position information for inference, leveraging tracking memory and explicit new tracking tokens to handle dynamic objects.

Related event: Whareformer: A New Approach for Online Object Tracking in Long Egocentric Videos(6 posts)→

Original post →

More from Research

Research channel →