Description »

One of the main research topics at CVLab is the 3D reconstruction of deformable surfaces from a monocular camera, and the goal of this project is to use the image contours of a deforming surface to reconstruct its shape.

State-of-the-art methods in monocular surface reconstruction rely on texture only. Consequently, the quality of the reconstruction degrades if there is little texture on the surface. One valuable source of information which is currently not exploited comes from the object boundaries. For this work, we want to introduce this boundary information in a framework similar to [Salzmann et al., ECCV 2008] to estimate the 3D deformations.