Rethinking Object-Centric Representations for Video Dynamics Modeling


This is a companion discussion topic for the original entry at https://arxiv.org/abs/2606.23436