| 總結: | Glaucoma and diabetic retinopathy are all ocular diseases that can cause significant vision loss and blindness. Early detection and treatment of these diseases is critical for vision preservation and preventing permanent vision damage. In recent years, deep learning techniques have been applied to ocular disease recognition with promising results, however the current system that available predict lower precision and it can lead to the poor decision making by ophthalmologist. Convolutional Neural Networks (CNNs) particularly have been shown to be effective in recognising various ocular diseases from eye images. In this study, CNN was trained with 600 eye images to identify glaucoma and diabetic retinopathy. A supervised learning approach was used to train the model from beginning to end. To increase the model's precision, various image augmentation methods were applied. The outcome demonstrates that, after classification's evaluation, the model managed to obtain 97% accuracy value. In the future, this system can train and validate CNNs using larger datasets, as well as collect and use more diverse datasets of eye fundus images to improve the accuracy of the system.
|
|---|