MARRAUD , Denis

– Nous étudions le problème de la segmentation sémantique d'images de profondeur pour clusteriser les pixels d'une image qui appar-tiennent à une même classe d'objet afin d'étendre les approches de localisation et cartographie (SLAM) à des cartes labellisées. Nous nous focaliserons sur les approches denses et directes qui exploitent l'ensemble...

This paper presents methods for performing real-time semantic SLAM aimed at autonomous navigation and control of a humanoid robot in a manufacturing scenario. A novel multi-keyframe approach is proposed that simultaneously minimizes a semantic cost based on class-level features in addition to common photometric and geometric costs. The approach...

This paper proposes a novel approach called Semantic Visual Odometry (SemVO) which incorporates class-level consistency priors into the problem of 6-DoF Visual Odometry. Dense class-level labels are learnt for each pixel of the image using a CNN trained for semantic segmentation. A semantic error is formulated penalising the sum of squared...