Ethnicity Recognition Under Difficult Scenarios Using HOG
Sadam Al-Azani, El-Sayed M. El-Alfy · 2019
With the rapid advance of globalization, analyzing nationality and race/ethnicity groups is becoming an emerging research topic that has multi-disciplinary real-world applications such as surveillance systems and targeted advertisements. This paper presents an approach to automatically predict the ethnicity groups of individuals based on their facial characteristics. Several ethnicity groups are considered in this study including: Asian, Indian, and others (like Hispanic, Latino and Middle Eastern). The proposed approach extracts features based on the Histogram of Oriented Gradients (HOG) texture descriptor. Then, it trains a support vector machine (SVM) to detect ethnicity with promising achievable results when evaluated on a publicly available dataset of labelled images.