Presentation

DescriptionWe present the TensorFlow Graphics library, containing a set of graphics-inspired differentiable components that allow us to build structured neural networks to solve various two- and three-dimensional graphics and vision problems. We augment the talks with real-time demos.

Prerequisites Some familiarity with deep learning, neural networks, and graphics pipelines would be beneficial.

Intended Audience Graphics and vision researchers, practitioners, and undergraduate/graduate students with some exposure to graphics and neural networks.

Lecturers

Cengiz Oztireli

Google Inc.

University of Cambridge