ECCV 2022 Tutorial on

Self-Supervised Representation Learning in Computer Vision

Tel Aviv
Location: Room Israel A
Dan Panorama Hotel, Tel Aviv (map)
Monday, October 24 (PM), 2022
Zoom Link



This tutorial covers popular approaches and recent advancements in the field of self-supervised visual representation learning. We will cover topics such as Masked Autoencoders and Contrastive Learning. We will show how such frameworks are successfully learning from 2D static image and dynamic video information. Finally, we will also discuss self-supervised learning from a machine learning perspective. Overall, we will show connections and distinctions between different techniques for self-supervised learning, and provide insights about popular approaches in the community.

Tentative schedule

2:05 - 2:15 Welcome and agenda - Xinlei Chen and Christoph Feichtenhofer, Meta AI

2:15 - 3:00 Opening remarks - Yann LeCun, Meta AI and NYU

3:00 - 3:45 Masked autoencoders as scalable vision learners (slides) - Xinlei Chen, Meta AI

3:45 - 4:30 Self-supervised learning from masked video and audio (slides) - Christoph Feichtenhofer, Meta AI

4:30 - 4:45 Coffee Break

4:45 - 5:30 The virtuous cycle of object discovery and representation learning (slides) - Olivier J. Hénaff, DeepMind

5:30 - 6:15 Contrastive learning of visual representations (slides) - Ting Chen, Google

Contact: Xinlei Chen, Christoph Feichtenhofer