<![CDATA[This study raised a topic related to the classification by using eye diseases in humans. This study uses two optimizing options, namely SGD and Adagrad. The data used are 601 images consisting of 430 training images, 50 validation images, and 121 test images with a total of 4 classes. The method used in this study is the Convolutional Neural Network (CNN) method with the VGG-19 architecture, with input in the form of images that have gone through a preprocessing process, namely resizing and the CLAHE (Contrast Limited Adaptive Histogram Equalization) method of eye disease images. The test scenario consisted of 8 scenarios with different Optimizer and ClipLimit. The highest test results were obtained in the first scenario using the Adagrad optimizer and clipLimit of 1.0 with an accuracy value of 65.29%, precision of 66.53%, recall of 65.29%, and f1-score of 65. 40%.]]>
Copyrights © 2022
