April 14, 2008
ECE graduate student Andrew Gallagher and Professor Tsuhan Chen won a Best Paper Award at the 2007 IEEE Conference on Computer Vision and Pattern Recognition (CVPR), Workshop on Semantic Learning Applications in Multimedia (SLAM). The paper, titled "Using Group Prior to Identify People in Consumer Images" describes a method for recognizing people in consumer images using more than just facial appearance.
People take pictures with their family and friends, and the authors form a probabilistic model of these social groups. The paper quantifies the benefit of combining this social information with traditional face recognition to improve the accuracy of people recognition in consumer image collections.
"This research enables applications for searching an image collection for images containing specific people, allowing you to find certain images quickly," said Gallagher. Other applications would include surveillance where authorities might want to document the associates of a person of interest. In each case, explains Gallagher, understanding the relationship between the people helps interpret the image collection.