The problem Online Diabetic Retinopathy Detector solves

This disease is a diabetic complication that causes damage to the blood vessels of the light sensitive
tissues located at the back of retina It takes place in the form of swelling, leakage or the blockage of blood
through the blood vessels in the retinal region.This can result in irreversible vision loss.The detection of
diabetic retinopathy in early stages is crucial to prevent the risk of progression into a further stage that is
non-proliferative diabetic retinopathy. The presence of diabetic retinopathy is visible in the fundus
images in the form of microaneurysms and macular edema. In this study we propose the use of a deep
convolutional neural network to analyze the fundus images for detection microaneurysms and macular
edema.The detection of diabetic retinopathy in early stages is difficult as the lesions, exudates are subtle
in nature making it hard to detect. Moreover, the shortage of trained personnel and clinics equipped with
appropriate tools worsens the situation.

Challenges we ran into

The most challenging part was finding a suitable dataset. We also faced some issues in hosting it and training the model.We used Heroku App to overcome the problem of hosting.