Modern Deep Learning has fundamentally changed Artificial Intelligence. Novel applications as well as significant improvements to old problems continue to appear at a staggering rate. Especially the areas of image and video understanding, retrieval, and synthesis have seen tremendous improvements and even the human baseline has been outperformed in several difficult applications.

The algorithms and the fundamental research in deep Machine Learning and Computer Vision that are driving this revolution are improving at an ever-increasing rate. The goal of this practical lab is, therefore, to give students hands-on experience with the state-of-the-art in this field of research. We will work on current problems in Computer Vision and Machine Learning and build on current algorithms to practically implement novel solutions. Consequently, the practical is also a good opportunity to take a close look at this area of research and prepare for a potential future final thesis.

Topics include but are not limited to:

• Image & video synthesis
• Visual superresolution and Image completion
• Artistic style transfer
• Interpretability of deep models
• Modern deep learning approaches, such as transformers and self-attention, invertible neural networks, diffusion models, etc.

Over the past two years, there has been a surge in new AI-based image generation and editing tools that do not require special computer skills, but are usable by laypersons, artists, and designers. This has been largely influenced by the publicly available, open-source “Stable Diffusion” model (https://ommer-lab.com/stable-diffusion). Now many researchers, start-ups, and artists are investigating downstream tasks without the need for a high-performance GPU cluster to train a base model. Moreover, a number of closed-source services such as Midjourney and Open AI’s Dall-E 2 have also drawn a lot of attention.

The foundation of this technology is the task of generating a single image solely based on a textual description of what should be depicted in the image. Examples of this can be found on websites such as https://lexica.art/. This technology can be extended to include additional information, such as depth maps, and allows for flexible image editing by changing existing objects based on text or removing parts of the image and filling it while paying attention to the remaining image. The latest advances also allow for generation or modification of video as well as rendering 3D scenes.

These topics and more will be covered in our seminar, where we will investigate AI-based image and video editing and generation techniques. Each student will focus on a specific topic. The objective of this seminar is then to investigate the connection between these techniques and the students’ respective fields of study and the greater societal and research implications. We will explore potential applications and issues in applying this technology. Each student will give a presentation of their ideas and write a report about the technique and its potential applications and implications.

Machine learning is changing the world. It can be found in almost all segments including healthcare services, education, transport, food, entertainment and many more. Specifically, in the field of computer vision, machine learning techniques have enabled programs to interpret and understand the world, sometimes even better than humans.

In this course, we will work on several projects to solve several computer vision tasks using python. The students will have a chance to get their hands on the classical (or potentially deep) machine learning algorithms in teams, and apply those techniques to some of the most interesting topics in computer vision.

This lecture focuses on deep learning approaches in computer vision with a particular emphasis on generative approaches that not only analyze, but in particular synthesize novel images and video.

Modern deep learning has fundamentally changed artificial intelligence. Computer vision was at the forefront of many of these developments and has tremendously benefited over the last decade from this progress. Novel applications as well as significant improvements to old problems continue to appear at a staggering rate. Especially the areas of image and video synthesis and understanding have seen previously unthinkable improvements – and provided astounding visual results with wide-ranging implications (trustworthiness of AI, deep fakes).

We will discuss how a computer can learn to understand images and videos based on deep neural networks. The lecture will briefly review the necessary foundations of deep learning and computer vision and then cover the latest works from this quickly developing field. The practical exercises that accompany this course will provide hands-on experience and allow attendees to practice while building and experimenting with powerful image generation architectures.

Topics include but are not limited to:

• Image & video synthesis
• Visual superresolution and Image completion
• Artistic style transfer
• Interpretability, trustworthyness of deep models
• Self-supervised learning
• Modern deep learning approaches, such as transformers and self-attention, invertible neural networks, diffusion models, etc.

Modern Deep Learning has fundamentally changed Artificial Intelligence. Novel applications as well as significant improvements to old problems continue to appear at a staggering rate. Especially the areas of image and video understanding, retrieval, and synthesis have seen tremendous improvements and even the human baseline has been outperformed in several difficult applications.

The algorithms and the fundamental research in deep Machine Learning and Computer Vision that are driving this revolution are improving at an ever-increasing rate. The goal of this practical lab is, therefore, to give students hands-on experience with the state-of-the-art in this field of research. We will work on current problems in Computer Vision and Machine Learning and build on current algorithms to practically implement novel solutions. Consequently, the practical is also a good opportunity to take a close look at this area of research and prepare for a potential future final thesis.

Topics include but are not limited to:

• Image & video synthesis
• Visual superresolution and Image completion
• Artistic style transfer
• Interpretability of deep models
• Modern deep learning approaches, such as transformers and self-attention, invertible neural networks, diffusion models, etc.

Over the past two years, there has been a surge in new AI-based image generation and editing tools that do not require special computer skills, but are usable by laypersons, artists, and designers. This has been largely influenced by the publicly available, open-source “Stable Diffusion” model (https://ommer-lab.com/stable-diffusion). Now many researchers, start-ups, and artists are investigating downstream tasks without the need for a high-performance GPU cluster to train a base model. Moreover, a number of closed-source services such as Midjourney and Open AI’s Dall-E 2 have also drawn a lot of attention.

The foundation of this technology is the task of generating a single image solely based on a textual description of what should be depicted in the image. Examples of this can be found on websites such as https://lexica.art/. This technology can be extended to include additional information, such as depth maps, and allows for flexible image editing by changing existing objects based on text or removing parts of the image and filling it while paying attention to the remaining image. The latest advances also allow for generation or modification of video as well as rendering 3D scenes.

These topics and more will be covered in our seminar, where we will investigate AI-based image and video editing and generation techniques. Each student will focus on a specific topic. The objective of this seminar is then to investigate the connection between these techniques and the students’ respective fields of study and the greater societal and research implications. We will explore potential applications and issues in applying this technology. Each student will give a presentation of their ideas and write a report about the technique and its potential applications and implications.