Content

Modern computer vision systems can recognize, localize, segment, and track objects in complex visual scenes. Deep learning frameworks such as PyTorch and GPU acceleration have become standard tools for developing these systems, making it possible to work directly with large models and realistic datasets. In the Vision Systems Lab, we focus on the practical development and analysis of modern computer vision methods. Students will implement, train, adapt, and evaluate state-of-the-art approaches for tasks such as object detection, image segmentation, visual recognition, and related vision applications. A central goal of the lab is to connect the concepts introduced in lectures with realistic computer vision problems. Students will gain hands-on experience with current architectures, datasets, and development workflows, while learning how to compare methods experimentally and critically assess their strengths and limitations. (Bi-)weekly meetings provide an opportunity to discuss concepts, implementation challenges, experimental results, and progress on the practical assignments.