description for "test for eye images"

We used keras library to train a sequential neural network to classify the images of an eye based on the possibility of the presence 
of cataract. 

Evaluation result on Test Data : Loss = 0.6931448446379768, accuracy = 0.5015015015015015

The low accuracy score was due to the lack of a dataset. We had to manually download 17 cataract and normal eye images and use data
augmentations to generate total of 1000 images per class. In other words the original images, and the hencce generated images, were able 
to train a neural net for a 50.1% accuracy only. Provided a better dataset, the accuracy for the trained model would also increase
significantyl.