Comparative Analysis of Accuracy and Loss during Cataract Detection using Deep Learning

Abstract

Cataracts are a type of eye condition that affects people of all ages, although they are most prevalent in people who
are between the ages of 40 and 50. Cataracts are one of the most common disorders that affect humans. If not
treated, this condition might result in blindness. Early detection is key to preventing this sort of eye illness from
progressing to a more serious stage. Therefore, we are going to present our model that makes use of Deep Learning
to analyse Fundus Images in order to diagnose this condition. In comparison to other models, this one has much
lower computational costs thanks to the inclusion of fully linked hidden layers that are equipped with cost functions
and activation functions. Proper training and optimisation of the model are also possible thanks to these features.
The model is given a total of 1130 photos that have been altered and 1130 images that have not been affected in
order to ensure that it is trained correctly and does not become overfit.