FitVid: Overfitting in Pixel-Level Video Prediction

TL;DR

This paper introduces FitVid, a new video prediction architecture that overfits on benchmarks to achieve high-quality predictions, demonstrating that overfitting can be beneficial and mitigated with augmentation techniques.

Contribution

We propose FitVid, a novel architecture capable of severe overfitting with similar parameters to existing models, improving performance across multiple benchmarks.

Findings

FitVid achieves state-of-the-art results on four benchmarks.

Overfitting can produce high-quality predictions by repeating training data.

Data augmentation techniques help mitigate overfitting effects.

Abstract

An agent that is capable of predicting what happens next can perform a variety of tasks through planning with no additional training. Furthermore, such an agent can internally represent the complex dynamics of the real-world and therefore can acquire a representation useful for a variety of visual perception tasks. This makes predicting the future frames of a video, conditioned on the observed past and potentially future actions, an interesting task which remains exceptionally challenging despite many recent advances. Existing video prediction models have shown promising results on simple narrow benchmarks but they generate low quality predictions on real-life datasets with more complicated dynamics or broader domain. There is a growing body of evidence that underfitting on the training data is one of the primary causes for the low quality predictions. In this paper, we argue that the inefficient use of parameters in the current video models is the main reason for underfitting. Therefore, we introduce a new architecture, named FitVid, which is capable of severe overfitting on the common benchmarks while having similar parameter count as the current state-of-the-art models. We analyze the consequences of overfitting, illustrating how it can produce unexpected outcomes such as generating high quality output by repeating the training data, and how it can be mitigated using existing image augmentation techniques. As a result, FitVid outperforms the current state-of-the-art models across four different video prediction benchmarks on four different metrics.