Deep learning for shape reconstruction (C09*)

• Subject Area: Mathematics ; Image and Language Processing, Computer Graphics and Visualisation, Human Computer Interaction, Ubiquitous and Wearable Computing
• Term: since 2020
• Project identifier: Deutsche Forschungsgemeinschaft (DFG) - Project number 195170736

Shape reconstruction is of central interest in computer vision, arising in applications such as image segmentation or 3D reconstruction from multiple views. The focus of this project is to study the use of deep neural networks for solving the inverse problem of shape reconstruction. This will also require the development of generalization of neural networks to incorporate additional constraints as well as a theoretical analysis of certain aspects of neural networks.

• DFG Programme: CRC/Transregios
• Subproject of: TRR 109: Discretisation in Geometry and Dynamics
• Applicant Institution: Technische Universität Berlin
• Project Heads: Professor Dr. Daniel Cremers; Professorin Dr. Gitta Kutyniok