Face Identity Detection and Recognition using Novel Convolutional Neural Network in Comparison with Haar Cascade to Improve Accuracy

Abstract

The main aim of the project is to recognize faces using the Novel Convolutional Neural Network algorithm in comparison with the Haar Cascade algorithm for the Google AI images dataset. Materials and Methods: Recognition of face is performed using CNN Algorithm (N=10) and Haar Cascade algorithm (N=10). CNN algorithm is a supervised machine learning algorithm. The Haar Cascade algorithm is a simple approach mainly used for classifying. Google AI Image dataset is used for Recognition of face. These samples are divided into two types: training samples (n=52,000(75%)) and test samples (n=17500(25 <sup xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">%</sup> )). By the help of CNN. Accuracy is calculated for face recognition. Results: The accuracy of face recognition using CNN algorithm is 91.01 % and Haar Cascade algorithm is 85.02 %. There is a significant difference between Adaboost algorithm and Support Vector algorithm with <tex xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">$0.04(\mathrm{P}\unicode{x00A1}{0.05})$</tex> ). Conclusion: CNN Algorithm appears to have better accuracy than the Haar Cascade algorithm in recognition face.