Comparative Study of Face Detection Using Cascaded Haar, Hog and MTCNN Algorithms

This paper compares a few effective and widely used face identification methods. In the area of image processing, automatic face detection is a highly difficult problem, and countless methods and algorithms have been proposed to solve it. We regularly observe face detection approaches being used to accurately identify faces in pictures, videos, and live video streams. There have been numerous face detection methods proposed till date. A brief overview of three face detection systems is provided in this comparison study. We have made an effort to describe a few key algorithms, such as the Haar Cascade Classifier, HOG, and MTCNN. A detailed comparative analysis shows that MTCNN is highly effective technique for face detection with efficiency of 99.7% and it detects faces from all views (i.e. frontal and side pose) as well as faces with accessories(glasses etc.). However Haar Cascade Classifier and HOG are simpler techniques and are easy to implement.