Temporally Consistent Object-Centric Learning by Contrasting Slots
This work addresses the need for stable and compositional video representations for downstream tasks like autonomous control, representing an incremental improvement over existing methods by explicitly enforcing temporal consistency.
The paper tackled the problem of achieving temporally consistent object-centric representations in unsupervised video learning by introducing an object-level temporal contrastive loss, which significantly improved temporal consistency and object discovery, leading to state-of-the-art results on synthetic and real-world datasets.
Unsupervised object-centric learning from videos is a promising approach to extract structured representations from large, unlabeled collections of videos. To support downstream tasks like autonomous control, these representations must be both compositional and temporally consistent. Existing approaches based on recurrent processing often lack long-term stability across frames because their training objective does not enforce temporal consistency. In this work, we introduce a novel object-level temporal contrastive loss for video object-centric models that explicitly promotes temporal consistency. Our method significantly improves the temporal consistency of the learned object-centric representations, yielding more reliable video decompositions that facilitate challenging downstream tasks such as unsupervised object dynamics prediction. Furthermore, the inductive bias added by our loss strongly improves object discovery, leading to state-of-the-art results on both synthetic and real-world datasets, outperforming even weakly-supervised methods that leverage motion masks as additional cues.