Advanced Deep Learning in Computer Vision

• Select, implement, and utilize state-of-the-art deep learning architectures for classical computer vision tasks such as classification, segmentation, and recognition
• Utilize and examine deep learning models that combine image data with other modalities, such as text
• Implement deep learning models for temporal image data, such as videos
• Explain and implement deep generative models for image synthesis
• Describe, implement and compare alternative methods for training deep learning models in limited data settings
• Assess the quality of deep learning models for computer vision from viewpoints of both performance and responsibility/ethics.
• Present projects and associated results in writing as well as oral presentation.
• Discuss strengths, weaknesses and societal implications of state-of-the-art deep learning models.

The course gives an introduction to advanced topics within deep learning for computer vision. Therefore, focus in the exercises is on implementing algorithms and using these for solving practical computer vision problems within the following topics: image recognition, sequential data, generative models, video understanding, explainability, and fairness.

02450/02456/02516, 02450, 02456, 02516 or their equivalents (in particular, we expect that you are well experienced with training neural networks for computer vision on custom data using PyTorch on a remote server)

Lectures, practical exercises and projects.

Minimum: 8.

Please be aware that this course will only be held if the required minimum number of participants is met. You will be informed 8 days before the start of the course, whether the course will be held.