Automated compound facial emotion recognition using hybrid deep learning model and DCHBO

Purpose: To create a system that effectively recognizes and categorizes human emotions from facial expressions using deep learning-based automated compound facial emotion detection strategy. Methods: A novel hybrid model based on deep learning combines the optimized bi-directional long short-term memory (O-Bi-LSTM) and deep neural decision forests (DNDF). The O-Bi-LSTM and DNDF are trained using the extracted features and the weight function of the BI-LSTM is fine-tuned using the new Dingo customized Honey Badger optimization algorithm (DCHBO) to further enhance the detection accuracy of the proposed model for automatically differentiating compound face emotions. Results: O-Bi-LSTM is concatenated, and the final detected outcome of the proposed model has been evaluated over the existing models. Conclusions: The superiority of the proposed model is demonstrated in detecting and classifying compound facial emotions accurately. © The Author(s), under exclusive licence to The Brazilian Society of Biomedical Engineering 2024.