Automatic Detection of Happy/Sad Face Images with DL-AN Analysis

The application of the deep-learning (DL) is common in image based detection and analysis. Due to its better accuracy, the DL-methods are commonly adopted in a variety of image examination approaches to find the best solution for the chosen problem. In this work, the DL-model based framework is considered to classify the digital face images into happy and sad, which is an essential process during the psychological examination. In this work, EffcientNet-V2 (ENV2) model based scheme is developed to get a better detection result. The different phases in this work include; (i) face image collection and resizing, (ii) feature extraction with ENV2-model and SoftMax based classification, (iii) implementing 3-fold cross validation and performance confirmation. This work considered the ENV2-models for the examination and the outcome of the developed system is verified using 1200 images per class and this tool helped to get an accuracy >92%.