Self-Supervised Visual Learning and Synthesis
- document 1 document 2 document 3
- niveau 1 niveau 2 niveau 3
Computer vision has made impressive gains through the use of deep learning models, trained with large-scale labeled data. However, labels require expertise and curation and are expensive to collect. Can one discover useful visual representations without the use of explicitly curated labels? In this talk, I will present several case studies exploring the paradigm of self-supervised learning — using raw data as its own supervision. Several ways of defining objective functions in high-dimensional spaces will be discussed, including the use of General Adversarial Networks (GANs) to learn the objective function directly from the data. Applications of self-supervised learning will be presented, including colorization, on/off-screen source separation, image forensics, paired and unpaired image-to-image translation (aka pix2pix and cycleGAN), and curiosity-based exploration.
Sur le même thème
La synthèse d’images (parfois appelée « la 3D ») permet de créer dans l’ordinateur des mondes fictifs, ultra-réalistes ou de style cartoon selon l’envie des graphistes, des réalisateurs, des
Despite our great expressive skills, we humans lack an easy way of conveying the 3D worlds we imagine. While impressive advances were made in the last fifteen years to evolve digital modeling
In this talk I will draw upon my own experiences as a computer graphics researcher and as a tango dancer to provide a very personal perspective on how research works, and on how to create sublime