A Hybrid Predictive Architecture Formulation Using Deep Learning And Histogram Of Gradients For Compound Emotion Recognition

Abstract

Facial emotion recognition has gained at- tention of researchers all over the world in the past few decades. Initially, emotions were classified in the seven basic categories which included happy, sad, angry, etc. However, human emotions are rarely this simple. They are usually combinations of dominant and complimen- tary emotions and are known as Compound Emotions. Two different ways have been commonly adapted for the recognition of these emotions from facial images: firstly, by using handcrafted features, or by using deep learning networks. This research analyzes the perfor- mance of a much simpler designed deep learning model named as Sequential-Convolution Neural Network (S- CNN) and four predefined deep learning networks for the recognition of compound emotions from facial im- ages. The objective of this paper is to replace sophisti- cated state-of-the-art prediction models with a straight- forward but effective approach. Therefore, this research suggests a hybrid network that maintains the S-CNN model’s design simplicity while boosting performance. The features extracted by the S-CNN model and the handcrafted features are combined in the hybrid S-CNN model. This process keeps the hybrid model’s architec- ture simple while improving its metrics values and in- creasing its accuracy to 99.62% when compared to other state-of-the-art models. The source code for this research can be found in our GitHub repository: SCNN_Hybrid_model.