Face Recognition in an Unconstrained Environment using ConvNet
Kunal Chaturvedi, Dinesh Kumar Vishwakarma · 2020
With the recent advancements in the discipline of Facial Recognition, it has made it easier to detect and identify multiple faces at a time in a situation where the subjects could have varying face posture, expressions, appearances in a dark or lighted background. In this paper, the detection of faces from the captured images having single or multiple faces in an unconstrained environment is done by Histogram of Oriented Gradients (HOG) feature descriptor and SVM classifier which is not only fast and accurate but also improves the accuracy of the proposed deep convolutional architecture which performs by learning feature representations to identify whether the subject is present in the image or not. The experiments are conducted on the DTU Biometric Research Lab, AT&T, and the Essex Face databases. In real-time, the proposed methodology performs excellent with high recognition accuracy.