MOSO: Decomposing MOtion, Scene and Object for Video Prediction

TL;DR

MOSO introduces a novel two-stage framework decomposing videos into motion, scene, and object components for improved prediction, generation, and interpolation, achieving state-of-the-art results on multiple benchmarks.

Contribution

The paper presents MOSO, a new decomposition-based framework combining VQVAE and Transformer for enhanced video prediction and generation.

Findings

Achieves state-of-the-art performance on five benchmarks.

Effectively decomposes videos into meaningful components.

Enables realistic video synthesis by combining different objects and scenes.

Abstract

Motion, scene and object are three primary visual components of a video. In particular, objects represent the foreground, scenes represent the background, and motion traces their dynamics. Based on this insight, we propose a two-stage MOtion, Scene and Object decomposition framework (MOSO) for video prediction, consisting of MOSO-VQVAE and MOSO-Transformer. In the first stage, MOSO-VQVAE decomposes a previous video clip into the motion, scene and object components, and represents them as distinct groups of discrete tokens. Then, in the second stage, MOSO-Transformer predicts the object and scene tokens of the subsequent video clip based on the previous tokens and adds dynamic motion at the token level to the generated object and scene tokens. Our framework can be easily extended to unconditional video generation and video frame interpolation tasks. Experimental results demonstrate that our method achieves new state-of-the-art performance on five challenging benchmarks for video prediction and unconditional video generation: BAIR, RoboNet, KTH, KITTI and UCF101. In addition, MOSO can produce realistic videos by combining objects and scenes from different videos.