Everest is a collaborative Indo-French project between research teams at Inria Grenoble (Thoth project-team) and IIIT Hyderabad (CVIT), and is funded by CEFIPRA. The team started its activities in April 2016.
The aim of this project is to enable the use of rich, complex models that are required to address the challenges of high-level computer vision. The work plan for the project will follow three directions. First, we will develop a learning framework that can handle weak annotations, in contrast to current models relying on expensive, fully-annotated data. Second, we will build on formulations such as self-paced learning to solve the non-convex optimization problem resulting from the learning framework. Third, we will develop efficient and accurate energy minimization algorithms, in order to make the optimization computationally feasible. The methodologies developed as part of this project will be thoroughly evaluated on challenging visual recognition problems, e.g., object detection and semantic segmentation.