Deep machine learning applied to moving image restoration

Online presentation at the SEAPAVAA Conference Navigating New Horizons in Audiovisual Archiving in Solo, Indonesia, on the 12^th of June 2024.

The panel Navigating the Future: AI/ML Applications in Audiovisual Archiving has been organised and moderated by Joshua Ng (Archives New Zealand). This was the first presentation, followed by Tasdik Eko Pramono (Digital Forensic and Artificial Intelligence Lab, Indonesia) and Shahab Qamar (National Film and Sound Archive, Australia).

Table of Contents

• Abstract
• Resources
• Presenter

Abstract

AV Preservation by reto.ch started restoring multispectral moving images using deep machine learning in 2017. The goal of this presentation is to share lessons learned in this field over the past seven years.

The presentation begins by recalling the ethical principles of conservation and restoration. After the basics of what is called artificial intelligence, the main part is devoted to an introduction to both multispectral image processing and supervised training of artificial neural networks. The final part summarises our positive and negative experiences in the field of machine learning, specifically applied to the restoration of moving image.

Resources

• The presentation’s slides, PDF, 8.8 MB.
• The video recording, MP4, 32.5 MB.

Presenter

Reto Kromer

2024-06-13