Labeling faces with names based on the name semantic network

In this study, we propose a method of labeling faces with names in a large number of news images with captions. Other works explored facial similarities to label faces with names that are sensitive to the intra-person appearance variations, and the captions can offer the cues for the correlations of candidate names. Our method combines textual similarity from image captions with visual similarity of face collections of candidate name to automatically recognize celebrities. It does not require any supervisory inputs. It includes two main steps. Firstly, we build a name semantic network based on textual and visual similarity. Secondly, we apply a name semantic network to label face images with names. We perform experiments on the data set which consists of approximate half a million news images from Yahoo news. The experimental results show that the performance of our method is better than the existing algorithms.