PamiTC Job Board - Posting Details

Title: PhD Positions and Internships at MPI TübingenPosted: January 26, 2016
Company/Institution: MPI Tübingen
Location: Tübingen
Department: Perceiving Systems

Description: The MPI for Intelligent Systems in Tübingen is looking for several PhD students and graduate interns interested in computer vision and machine learning. The positions are fully funded. 1) PhD Position: Accurate Light Field Estimation http://www.cvlibs.net/downloads/flyer_2016_phd_lightfield.pdf The goal of this PhD project is to develop efficient probabilistic models of light, materials, and geometry in order to provide high-fidelity reconstructions of the user environment from RGB and RGB-D video: Given several images as observations, the task of inverting the image formation process is to recover the 3D geometry of the scene, recognizing the objects and materials they are composed of, identify light sources present in the scene and reason about light transport including reflection, refraction and shadows. As inverse graphics is an ill-posed inference problem, appropriate prior assumptions about the world must be made. 2) PhD Position: Efficient Invariant Deep Models for Computer Vision http://www.cvlibs.net/downloads/flyer_2016_phd_invariances.pdf The goal of this PhD project is to investigate physical invariances in deep convolutional neural networks for computer vision and to develop models which can generalize to novel domains such as new sensor arrangements. Towards this goal, a generative forward model of the image formation process using physically-based rendering techniques shall be developed and applied to several different tasks (e.g., scene understanding, semantic segmentation, object detection) in the context of multiple practically relevant applications (e.g., autonomous driving, advanced driver assistant systems). 3) Graduate Internship: Deep Learning for Computational Photography http://www.cvlibs.net/downloads/flyer_2016_internship_photography.pdf The goal of this project is to make deep neural networks applicable to very large imagery (10-50 Megapixels) while still allowing for efficient learning and inference. Furthermore, novel models shall be developed which are able to enhance images by learning the physics of the underlying image formation process and camera optics. This position is available to students currently enrolled in a PhD program who like to visit us for a period of 6 months. Andreas Geiger MPI Tübingen

Application Instructions: See: http://www.cvlibs.net/applications.php