First NSF grant under CRII
The National Science Foundation has awarded $175,000 to the University of Central Florida (UCF) for a project entitled "CRII: RI: Multi-Source Domain Generalization Approaches to Visual Attribute Detection" under the direction of Boqing Gong. The award starts May 1, 2016 and ends April 30, 2018. This is the first NSF award of its kind ever granted to UCF.
Under this project, Dr. Gong plans to develop new feature extraction tools tailored to account for the middle-level attributes, as opposed to the traditional visual features which were primarily designed and tested for high-level visual recognition. To this end he proposed three major thrusts. He begins by learning a "shallow" feature mapping to distill attribute-discriminative signals while to eliminate category-centric cues. He then investigates "deeper" into the feature extraction frameworks---Fisher vectors and convolutional neural networks|to revise them for the purpose of attribute detection. For more information please visit the corresponding NSF page: http://www.nsf.gov/awardsearch/showAward?AWD_ID=1566511