Face Mask Detection Using Autoencoder Algorithm Comparing with Recurrent Concurrent Neural Network for Improving Accuracy

This study aims to evaluate the effectiveness of two classification algorithms-AutoEncoder and Recurrent Convolutional Neural Network (R-CNN)-in predicting face mask detection. To conduct this analysis, a dataset of 1,065 face images was used. The sample size for each group was determined to be 32 based on a G power analysis, with an alpha level of 0.05 and a beta value of 0.5, ensuring a 95 % confidence interval and 80 % power. The results show that while both algorithms were used to predict face mask detection, there were notable differences in their performance. The AutoEncoder classifier achieved an accuracy of 94.30% in detecting face masks, while the R-CNN classifier reached a higher accuracy of 97.44%. The significance value of 0.001 indicates a strong statistical significance in the results. Despite the higher accuracy of the R-CNN classifier, the study's findings suggest that the AutoEncoder performs comparably well and is a viable option for face mask detection. In summary, while the R-CNN classifier slightly outperforms the AutoEncoder in terms of accuracy, the AutoEncoder still proves to be a robust tool for face mask detection tasks.