Hybrid siamese heterogeneous convolutional neural networks with electric eel foraging optimization-based deep feature vectors concatenation for eye disease detection.

In order to prevent vision loss and blindness, eye problems must be identified and diagnosed. In order to avoid blindness and vision loss, it is crucial to identify the main cause of eye illness. In previous studies, several DL algorithms have been used for detecting eye disease but do not provide sufficient results. To overcome these challenges is suggested. In this publication, a Hybrid Siamese Heterogeneous Convolutional Neural Network with Electric eel foraging optimization-based Deep Feature Vectors Concatenation for Eye Disease Detection (DeepSiaHetCNN-EeFOA) is suggested. There are datasets like databases similar datasets. These data are subsequently preprocessed using the Kuwahara Filtering with Anisotropic Dissemination (KFAD) technique. The Unet with split attention-based MobileNetV2 network (U-SA-MNetV2) approach is then used to segment eye diseases and extract features. DeepSiaHetCNN-EeFOA is then used for classifications to identify normal and abnormal areas and to identify the kind of eye illness. The system demonstrated was run using Python. The proposed DeepSiaHetCNN-EeFOA is tested on four datasets. It outperforms existing approaches by achieving 99.9% accuracy & 99.8% recall. This demonstrates the method's exceptional effectiveness and promise for further development in the sector.