BEES: Bio-Inspired Enhanced Emotional Segmentation System for Depression Detection using Deep Learning CNN with Bee Colony Optimization

Growing number of cases of depression globally has encouraged AI researchers to investigate automated ways for timely diagnosis. In this paper we introduce the BEES (Bio-inspired Enhanced Emotional Segmentation System), a novel approach for depression detection from image data. This includes BCO-based emotional region partitioning and swarm inspired optimization algorithms to fine-tune CNN hyperparameters. Together with swarm-based voting for ensemble learning that increases prediction accuracies. Training and Evaluation. The FER-2013 dataset contains 30,524 facial emotion images. The optimized CNN model had an accuracy of 90.45%, and with ensemble enhanced to a final predictive performance of (92.15%) together with AUC;0.96%. The results show the excellent detect indicator of depression which is suitable by using our proposed method, and it proved that bio-inspired algorithms should be used in convolutional-based DL with clinical supervised data for mental health screening. Challenges like feature extraction, convergence and overfitting are solved by the presented framework which makes it a strong solution for scalable real-time depression detection.