Application of Deep CNN Networks in Ocular Disease Detection

Currently millions of individuals worldwide are suffering from ocular diseases. Diagnosis of ocular diseases by conventional methods is challenging, labor-intensive and prone to mistakes. Unfortunately, delayed diagnosis and treatment frequently results in blindness. Therefore, an automatic ocular illness detection method is the need of the hour. Fundus images are widely used for identifying ocular diseases. However, there is a chance that the patient may be suffering from multiple ocular diseases. In such cases the ophthalmologist cannot effectively identify the disease from the fundus images. To aid the ophthalmologist, this work aims to develop a revolutionary multi-class classification model for diagnosing ocular diseases from fundus images. The model's performance is assessed with DenseNet, Inception ResNet, EfficientNetB4, and EfficientNetB6, in terms of losses, accuracy, and precision.