Automated Glaucoma Detection from Fundus Images using Deep Learning

One of the main causes of blindness, glaucoma, requires early identification in order to effectively treat it. Our suggested design has models like CNN, VGG 16, ResNetV3. And among them we have taken CNN which has performed with better accuracy. The present research offers an automated method for reliable glaucoma diagnosis using fundus pictures and deep learning algorithms. Convolutional Networks are used to provide fine segmentation and noise reduction by spatial convolution. The Cup-to-Disc Ratio (CDR) is used in the study to classify glaucoma and measure its severity. Model is used to differentiate between fundus pictures of glaucoma and non-glaucoma. The strategy suggested consists of feature extraction, image validation, optic disc and cup segmentation using CNN, and final classification that determines if an input picture is healthy or glaucomatous. The selected method exhibits great accuracy in structural feature-based classification for glaucoma detection, outperforming conventional machine learning techniques.