HYBRID WILD GEESE LADYBUG BEETLE OPTIMIZATION-BASED DEEP MAXOUT NETWORK FOR HUMAN EMOTION RECOGNITION

Emotions play a vital role in everyday life as they influence human decision-making, planning, reasoning, and other mental processes. The various sources like physiological signals, behavior, speech, and facial expressions are highly used for identifying emotions. There are several key factors for identifying human interaction as the understanding and recognizing emotions of humans are equally important. Detecting human emotions from images is a particularly powerful and challenging task in social communication. Recently, Deep learning (DL) technologies are advancing rapidly in the field of emotion recognition as it is highly helpful in providing solutions to real-world challenges. Thus, this paper presents an emotion recognition approach using facial images. In this research, an automatic model named Hybrid Wild Geese Ladybug Beetle Optimization-Deep Maxout Network (WGLBO-DMN) for human emotion recognition. The WGLBO is the integration of the Wild Geese Algorithm (WGA) and the Ladybug Beetle Optimization (LBO). At first, the Kalman filter is employed to pre-process the input face image and is sent to the feature extraction phase, where features like Statistical, Shape Local Binary Texture (SLBT), Oriented Fast and Rotated BRIEF (ORB), Scale Invariant Feature Transform (SIFT), Speeded-Up Robust Features (SURF), and CNN features are extracted. The extracted features are highly helpful in recognizing human emotions, like surprise, sadness, neutrality, anger, happiness, fear, and disgust, which are done by using DMN and the devised WGLBO algorithm is used for tuning the DMN. The “face expression recognition dataset” is employed in this study to validate the performance of the model. Here, the presented WGLBO_DMN achieved the best performance with a high True Positive Rate (TPR) of 93.6%, True Negative Rate (TNR) of 92.8%, and accuracy of 93.4%, respectively.