An Innovative Approach on Face Emotion Recognition using Ant-Lion Optimization Model

Emotion analysis is an interesting area of research that contains various integrations of contextual data collected from real-life incidents. ‎Various existing implementations of emotion detection address facial expressions that signify changing moods. The proposed system is ‎suggested by considering the inconsistencies and rankings of various critical parameter concentrations in existing frameworks for emotion ‎detection, such as empathic concern (intention towards others) and fortune and personal distress, which relate to the situation of comfort in ‎response to others' emotions. Emotions need to be understood deeply to recognize the feelings of humans accurately. In the presented ‎system, the feature extraction used for facial expression handling is implemented using a neural network with respect to facial emotions, ‎such as happy, Fear, sad, angry, surprised, and neutral. In practical humans, they can generate various facial expressions during ‎communication that vary in very intense manners. The proposed system-wise hybrid feature extraction and facial expression identification ‎technique, utilizing Viola-Jones cascade object detectors and MSER feature extraction technique with speed-up robust feature extraction ‎‎(SURF) technique, is represented as multimodality network features (MMNF) for classification. The identified features are further classified ‎using an LSTM (Long Short-Term Memory) model, combined with a CNN (Convolutional Neural Network) hybrid architecture. Further, ‎the system parameters are optimized through the Hybrid Ant-Lion optimization technique. The presented system achieved 98% accuracy ‎compared with various state-of-the-art approaches.