THIS IS A CONTINUATION OF N00014-14-1-0646 Deep Structured Learning
Abstract
Short Work Statement:Develop novel machine learning approaches for holistic scene understanding that include feature learning, scenerecognition, structured learning and inference.Objective:Investigate and develop a range of novel machine learning approaches for holistic scene understanding that include feature learning, scene recognition, structured learning and inference.Approach:The Pis will develop structured machine learning in the framework of deep learning with convolutional networks. Some key innovations include: (a) Fusion of deep feature learning methods with structured learning approaches, to enable end-to-end training of system; (b) Algorithms for the automatic and human-guided discovery of latent structure; and ~ Develop methods to integrate high-level knowledge and low-level image data in the convolutional network models. For algorithm development and experiments, the Pis will leverage several popular vision datasets including the NYU-Depthdataset V2.0. Together with images of realistic indoor scenes, it provides a dense depth map and human per-pixel annotations and support relationships.Overall Merits and ONR Relevance:This work is expected to substantially contribute to the methods for learning complex concepts including complex objects and their relationships, and scenes.This work is highly relevant to enabling Navy?s Autonomy and Information Dominance. Automated methods for recognizing objects and scene are of critical importance to naval missions that include perception for autonomous agents and understanding surveillance imagery.
Document Details
- Document Type
- DoD Grant Award
- Publication Date
- Sep 23, 2016
- Source ID
- N000141612698
Entities
People
- Rob Fergus
Organizations
- New York University
- Office of Naval Research
- United States Navy