Autoregression-free video prediction using diffusion model for mitigating error propagation

TL;DR

This paper introduces ARFree, a novel diffusion model-based framework for long-term video prediction that avoids error accumulation by directly predicting future frames from context, outperforming existing methods.

Contribution

It presents the first autoregression-free video prediction framework using diffusion models, enhancing long-term prediction accuracy and consistency.

Findings

Outperforms state-of-the-art methods on benchmark datasets

Reduces error propagation in long-term video prediction

Improves motion continuity and contextual consistency

Abstract

Existing long-term video prediction methods often rely on an autoregressive video prediction mechanism. However, this approach suffers from error propagation, particularly in distant future frames. To address this limitation, this paper proposes the first AutoRegression-Free (ARFree) video prediction framework using diffusion models. Different from an autoregressive video prediction mechanism, ARFree directly predicts any future frame tuples from the context frame tuple. The proposed ARFree consists of two key components: 1) a motion prediction module that predicts a future motion using motion feature extracted from the context frame tuple; 2) a training method that improves motion continuity and contextual consistency between adjacent future frame tuples. Our experiments with two benchmark datasets show that the proposed ARFree video prediction framework outperforms several state-of-the-art video prediction methods.