Recognizing Human Races through Machine Learning—A Multi-Network, Multi-Features Study

The human face holds a privileged position in multi-disciplinary research as it conveys much information—demographical attributes (age, race, gender, ethnicity), social signals, emotion expression, and so forth. Studies have shown that due to the distribution of ethnicity/race training datasets, biometric algorithms suffer from “cross race effect”—their performance is better on subjects closer “country origin” algorithm. contributions this paper are two-fold: (a) first, we gathered, annotated made public large-scale database (over 175,000) facial images by automatically crawling Internet for celebrities’ belonging various ethnicity/races, (b) trained compared four state art convolutional neural networks problem ethnicity classification. To best our knowledge, largest, data-balanced, publicly-available with information. We also studied impact traits image characteristics race/ethnicity deep learning classification methods obtained results ones extracted psychological studies anthropomorphic studies. Extensive tests were performed order determine features which sensitive to. These recognition rate 96.64% demonstrate effectiveness proposed solution.