Distribution of Group Photos using face recognition (IE)

Abstract

Face recognition has become popular over the years. One example is Facebook’s facial recognition photo tagging function. However, there is no current available software doing distribution of group photos. If a person wants to distribute a large number of group photos to each person manually, he/she may take a few days to do it. This project aims at building a face recognition software which can do a quick and accurate distribution of group photos for many social occasions such as graduation trips and wedding parties. OpenCV is used to do the face recognition. The Deep Metric Learning is employed in the software. Histogram of Oriented Gradient (HOG) is applied to do the face detection. The software is written in Python. The user interface allows the user to upload his/her selfie and enter his/her email address. A zip file containing all group photos that the user appears in will be sent to the user's email. The result illustrates that the software can successfully distribute group photos via email with an accuracy of about 90%. The result may be affected by the quality of the dataset and the input image. Further enhancement of the distribution could be done by other social media applications like WhatsApp and WeChat.