Course details


Deep Learning for Image Generation

SS 2023 Ulf Krumnack OFFLINE
B.Sc modules:
CS-BWP-NI - Neuroinformatics
KOGW-WPM-NI - Neuroinformatics
M.Sc modules:
CC-MWP-NI - Neuroinformatics
CS-MWP-NI - Neuroinformatics

CS-BW - Bachelor elective course
CS-MW - Master elective course
Tue: 10-12

Image generating systems like Stable-Diffusion, Imagen, and Dall-E have caught a lot of media attention during the last months. In this course, we will explore the techniques utilized by these and other systems, including generative adversarial networks, different types of autoencoders, autoregressive and recurrent models, flow-based models, as well as diffusion models. We will also cover several extensions and applications like superresolution, in- and outpainting, style-transfer, and deep-fakes. Beyond the discussion of different approaches, this course also aims at investigating the practical applicability of existing systems, allowing the participants to explore the state of the art, experimenting with different models and understanding limitations of current techniques. The ideal participant of this course would have some background in deep networks, such as it is covered in the course "Artificial Neural Networks with TensorFlow". Additional knowledge in computer vision and/or computer graphics is helpful but not strictly required.