Analogizing the Potency of Deep Learning Models for Glaucoma Detection

The ocular disorder inhibiting vision, has been abruptly proliferating due to evolution in mankind and its tack of existence. One such vision-threatening malady is Glaucoma, which mutilates the optic nerve of the eye which can cause sudden blindness. Ophthalmologists utilize copious methodologies to detect the disease but the accuracy is constrained due to humane limitations. There has been indagation for recognizing Glaucoma incorporating Machine Learning and Deep Learning Models with varied ideations but the efficacy isn’t apt for such cardinal affliction. In this paper, we have collated various Deep Learning Architectures based on diverse parameters, such as MobileNetV2, DenseNet121, InceptionV3, InceptionResNetV2, ResNet50, and VGG16, appealed over 02 Classes categorized into ‘G’ and ‘NG’. The dataset has been aggregated from various sources and preprocessed for utmost proficiency. Profuse metrics have been integrated for validating the potency of our contrivances namely Precision, Recall, F1-Score, Cohen Kappa Score, and AUC. The modus operandi gave us phenomenal results with high veracity.