Innovative Facial Expression Identification for Criminal Identification using Unsupervised Machine Learning and Compare the Accuracy with CNN Classifiers

Abstract

Aim: To recognize human face expressions accurately using machine learning algorithms and compare the image features against the criminal records. Methods and Materials: The study contains 2 groups i.e., Unsupervised Machine Learning model is the first group and CNN model is the second group. The accuracies of each model are compared with different sample sizes from 500 to 2000 and the study parameters include alpha value 0.05, beta value 0.2, and the pretest power value 0.8. Results: This paper is an attempt to improve the accuracy of criminal identification using Unsupervised Machine Learning Algorithm. The AI based Application avoids the overfitting of the data. The proposed model has improved accuracy of 92.46 % with p<0.05 in face detection than the CNN Classifiers having accuracy of 86 %. Conclusion: The outcomes of the proposed model Unsupervised Machine Learning algorithm was compared with the CNN Classifiers algorithm and the proposed model seems to have higher accuracy than the Machine learning algorithm.