Learning Object-Centric Video Models by Contrasting Sets
This work provides an incremental improvement for researchers working on self-supervised object-centric representation learning in video by refining the contrastive objective.
The paper addresses a limitation in contrastive self-supervised learning for object representations where existing methods struggle to differentiate between distinct objects in slots versus the same object replicated across slots. The authors propose a global, set-based contrastive loss that aggregates representations and contrasts joined sets, along with attention-based encoders, leading to improved reconstruction, future prediction, and object separation on synthetic video datasets.
Contrastive, self-supervised learning of object representations recently emerged as an attractive alternative to reconstruction-based training. Prior approaches focus on contrasting individual object representations (slots) against one another. However, a fundamental problem with this approach is that the overall contrastive loss is the same for (i) representing a different object in each slot, as it is for (ii) (re-)representing the same object in all slots. Thus, this objective does not inherently push towards the emergence of object-centric representations in the slots. We address this problem by introducing a global, set-based contrastive loss: instead of contrasting individual slot representations against one another, we aggregate the representations and contrast the joined sets against one another. Additionally, we introduce attention-based encoders to this contrastive setup which simplifies training and provides interpretable object masks. Our results on two synthetic video datasets suggest that this approach compares favorably against previous contrastive methods in terms of reconstruction, future prediction and object separation performance.