GMD Model Based on Multi-Label Classification for Detection and Diagnosis of Eye Diseases

  • Serhii Yaroshchak
  • Mahmoud Smaida
Keywords: multi-label classification, GMD model, eye diseases, confusing matrix, deep learning, diabetic retinopathy, glaucoma, myopia


The diagnosis of eye diseases is still considered a challenge in the health care sector. This is mainly due to the fact that eye diseases are difficult to recognize even for the optometrist as they come in a variety of forms. Therefore, using of AI should be made available to help the ophthalmologist to detect and diagnose of eye diseases. In this work, we will introduce GMD model based on multi-label classification with additional techniques to detect and diagnosis eye diseases. Confusion matrix will be used as evaluation performance model. In this study, we will compare three models of neural network, AlexNet, VGG16 and Inception-v3 with the GMD model in order to evaluate performance of the models. The dataset consists of four classes of eye diseases, Glaucoma, Myopia, Diabetic retinopathy and Normal. All these networks based on label classification deployed using GPU in Google Colab, according to all the experiments, we obtained results for each combination and observed that for multi-label classification, GMD model gives the best accuracy (95%) compared with the other models.