Automatic Analysis of Happy/Sad Face Image with Deep Learning Model
G R Abijith, Mohammad Musa Al-Momani, Md. Tabil Ahammed, Sorabh Lakhanpal, J. Bino, S. Prabha
Abstract
Deep learning (DL) is often used for detecting and analysing images. Because they are more accurate, the DL methods are often used in different ways to look at images to identify the best way to solve the problem at hand. This study uses a DL-model-based framework to sort digitised facial photos into happy and sad groups. This is an important step in the psychological test. This work creates a strategy based on the EfficientNet model to improve the detection results. This study contains several steps: (i) collecting and resizing face images, (ii) extracting features and classifying them using a DL model and SoftMax, and (iii) running 3-fold cross validation and confirming performance. This study looked at the DL-models for the test, and the results of the system that was made were checked using 1200 photographs per class. This tool helped to get an accuracy of more than 91%.