ECCV 2020 Daily - Tuesday

2 Workshop Report 16 DAILY T u s d a y The 1st edition of the ECCV SHApe Recovery from Partial Textured 3D Scans (SHARP) was held virtually on two days ago on Sunday. The workshop was organized by Djamila Aouada (SnT, University of Luxembourg), Kseniya Cherenkova (Artec 3D), Alexandre Saint (SnT, University of Luxembourg), David Fofi (University of Burgundy), Gleb Gusev(Artec3D), andBjörnOttersten (SnT, University of Luxembourg). SHARP aims to promote concepts that exploit both shape and texture in the processing of 3D scans and specifically, this year, in the task of recovery from partial incomplete data . Indeed, recently, many successful learning-based approaches have been developed for 3D shape reconstruction. However, very little has been proposed for reconstructing full 3D objects with shape and texture at the same time . The workshop articulated around two challenges on the reconstruction of full 3D textured meshes from partial 3D scans, and it released two new datasets. The first one is 3DBodyTex.v2 , which is an extended version of 3DBodyTex released in 2018 captured with the Shapify Booth of Artec3D. 3DBodyTex. v2 contains over 3000 body scans: 800 in tight clothingand2500 incasual clothing with up to 6 various poses per subject . We note that, for privacy reasons, faces are blurred both on geometry and texture. This 3D human body data offers a unique high-quality texture which opens new possibilities for original developments. A second dataset was made available as a subset from Artec 3D’s ViewShape repository of 3D scans. This dataset is called 3DObjectTex.v1. It contains over 2000 various generic objects with various complexity in texture and geometry. The workshop featured two exciting plenary talks by Didier Stricker , Director of the German Research Center for Artificial Intelligence (DFKI), and by Hao Li , CEO and co-founder of Pinscreen, talks by Challenge finalists, a review of challenge results, and a live panel discussion. Artec 3D awarded a cash prize of 4k€ to the winners, Julian Chibane and Gerard Pons-Moll from Max Planck Institute