Computer Vision

Level

Foundation, Intermediate, Broad, Specialised, Theory, Algorithmic

This topic concerns computer vision for image/video analysis, with its courses ranging from foundational image processing up to specialised deep learning approaches. It covers both theoretical knowledge and its real-world applications (e.g., in autonomous systems perception, medical imaging, etc.).

Learning outcomes

Content /
Knowledge

Students should be able to:

Understand/describe common algorithms for image segmentation, filtering, registration, search and retrieval, compression/storage, 2D shape description and recognition, as well as algorithms for video motion estimation, description, search, indexing, retrieval, streaming and compression.
Understand/describe the operation of various visual sensors, the image/video digitization process and the most common existing image/video formats.
Compare between various handcrafted and convolutional feature extraction methods for image/video description / representation.
Understand/describe common algorithms for visual SLAM, face/person/pedestrian detection, object detection/tracking/pose estimation, semantic/instance segmentation, facial expression recognition and activity recognition.
Understand/describe the process of image acquisition by visual sensors and the mathematical modelling of image formation.
Understand/describe common algorithms for camera calibration, stereoscopic 3D vision, depth estimation, surface geometry description and object 3D localization, topology description, landmark extraction and registration.

Methodological
skills

Students should be able to:

Analyse and develop (in C/C++, MATLAB or Python) the taught computer vision algorithms, by practically applying their gained knowledge in a systematic manner.
Evaluate the accuracy of computer vision algorithm implementations on suitable datasets, by employing common and appropriate task-specific metrics.