REFINING OPHTHALMIC DISEASE CLASSIFICATION USING DEEP LEARNING TECHNIQUES

Abstract

EYE sickness has emerged as one of the famous sicknesses in the world. One 10th of the population within the world suffers from eye disorder of different shape. Major cause of this ailment is inadequate information, preservation of hygiene, eating regimen, age and lack of awareness. In this paper early detection has been attempted through use of deep studying methods for the eye detection. A number of the situations like Glaucoma, cataract, everyday eye health and diabetic retinopathy are being identified early by using deep learning methods. Labelled retinal photos are taken from public repository as dataset. MobileNetV2 is used as a characteristic extractor because of its performance in restrained material gear which will acquire the temporal and spatial features of eye pix, a hybridized version with CNN+BiLSTM is developed in supplement with the switch mastering. The accuracy of the version is improved by using great tuning, augmentation, records pre-processing and early stopping strategies which is covered within the education pipeline. The accuracy exceeds 90% which makes it healthy for use in the actual-global health care surroundings. The results show its performance in combining CNN with LSTM in category of clinical imaging dataset and the significance of deep gaining knowledge of in ophthalmology.