Time-Conditioned Generative Modeling of Object-Centric Representations for Video Decomposition and Prediction

TL;DR

This paper introduces a time-conditioned generative model for videos that improves object shape reconstruction and novel view prediction by disentangling object and view representations, using Transformers and Gaussian processes.

Contribution

It presents a novel approach combining Transformers and Gaussian processes to enhance object-centric video modeling without requiring viewpoint annotations.

Findings

Accurately reconstructs complete object shapes even when occluded.

Enables novel view prediction without explicit viewpoint labels.

Demonstrates superior performance on multiple datasets.

Abstract

When perceiving the world from multiple viewpoints, humans have the ability to reason about the complete objects in a compositional manner even when an object is completely occluded from certain viewpoints. Meanwhile, humans are able to imagine novel views after observing multiple viewpoints. Recent remarkable advances in multi-view object-centric learning still leaves some unresolved problems: 1) The shapes of partially or completely occluded objects can not be well reconstructed. 2) The novel viewpoint prediction depends on expensive viewpoint annotations rather than implicit rules in view representations. In this paper, we introduce a time-conditioned generative model for videos. To reconstruct the complete shape of an object accurately, we enhance the disentanglement between the latent representations of objects and views, where the latent representations of time-conditioned views are jointly inferred with a Transformer and then are input to a sequential extension of Slot Attention to learn object-centric representations. In addition, Gaussian processes are employed as priors of view latent variables for video generation and novel-view prediction without viewpoint annotations. Experiments on multiple datasets demonstrate that the proposed model can make object-centric video decomposition, reconstruct the complete shapes of occluded objects, and make novel-view predictions.