Chris Liebold
SEASONAL aims to create intelligent deep learning models capable of transforming outdoor images across varying environmental conditions such as seasons, weather, and lighting.
By leveraging cutting-edge machine learning techniques, particularly in image-to-image translation, SEASONAL will enable the seamless transformation of images between different visual domains with high quality and plausibility. One of the key challenges is handling diverse environmental changes (multi-domain) while ensuring the generated results remain coherent and visually realistic. The project also addresses the associated challenges such as ensuring the plausibility of transformations, improving image quality both qualitatively and subjectively by developing new no-reference quality assessment metrics to evaluate the results effectively.
The results, that are generated in SEASONAL, can be applied in various computer vision domains e.g., object detection and segmentation, but also have applications in the fields of extended reality (XR) and content generation.
