Torsten Sattler de l’Université de Technologique de Chalmers (Suède) visitera l'ERL ViBot pour la soutenance de thèse de Nathan Piasco. A cette occasion, il nous présentera ses travaux le mardi 26 novembre 2019 à 10 dans l’amphi 8 du Centre Condorcet (Le Creusot).
Titre : Visual under Seasonal and Illumination Changes
Robustly perceiving the scene, independently of seasonal, illumination, and other changes, is a core requirement for any autonomous system operating in the real-world. In this talk, we will consider two tasks that require robust scene perception: Visual Localization and Semantic Segmentation, as well as their connection to each other. Starting with the visual localization problem of determining the pose of an autonomous system with respect to a scene representation, we explore two paradigms that can be used to localize images under strong changes in the viewing conditions:
1) Rather than designing representations, e.g., in the form of local features, that are invariant / robust to changes in the scene, domain adaption (for example in the form of generative neural networks) can be used to transform the current viewing conditions to a state close to the conditions under which the scene representation was constructed.
2) Rather than trying to predict how scenes evolve over time, invariant representations try to encapsulate the gist of a scene. In this talk, we consider semantic invariance, i.e., a fact that the semantic meaning of a scene should be invariant to seasonal and illumination changes.
The second approach relies on the assumption that semantic segmentations can be computed robustly under scene changes, which is often not true in practice. We thus discuss approaches to improve semantic segmentation performance using 3D geometry as a weakly supervisory signal.