DeepRemaster: Temporal Source-Reference Attention Networks for Comprehensive Video Enhancement

TL;DR

DeepRemaster introduces a unified neural network framework that enhances vintage videos by combining super-resolution, noise removal, contrast adjustment, and colorization, maintaining temporal consistency and improving with more reference images.

Contribution

It presents a novel source-reference attention mechanism within a temporal CNN for comprehensive video remastering, handling multiple references without segmentation.

Findings

Outperforms existing methods in quality metrics.

Performance improves with longer videos and more reference images.

Effectively handles diverse remastering sub-tasks in a single framework.

Abstract

The remastering of vintage film comprises of a diversity of sub-tasks including super-resolution, noise removal, and contrast enhancement which aim to restore the deteriorated film medium to its original state. Additionally, due to the technical limitations of the time, most vintage film is either recorded in black and white, or has low quality colors, for which colorization becomes necessary. In this work, we propose a single framework to tackle the entire remastering task semi-interactively. Our work is based on temporal convolutional neural networks with attention mechanisms trained on videos with data-driven deterioration simulation. Our proposed source-reference attention allows the model to handle an arbitrary number of reference color images to colorize long videos without the need for segmentation while maintaining temporal consistency. Quantitative analysis shows that our framework outperforms existing approaches, and that, in contrast to existing approaches, the performance of our framework increases with longer videos and more reference color images.