Aidentifye (Project EDR)

An Eye Disease Recognition Web App that takes in an eye picture and outputs if the eye is normal or has a disease. Using CNN, we trained a model on publicly available eye images on the web which resulted in a limited but proof-of-concept application. It was intended to be a potential assistant to eye experts that can cross-check the results of the app. But if it was given even more quality data and training experimentation, it can serve non-experts too in doing initial checks, even if limited. Definitely not intended to replace eye experts, but it is a neat working proof-of-concept or prototype.

Features

  • - Takes an eye image through device camera

  • - Outputs a label based on if the eye photo suggested a normal eye or an eye with a disease, limited to Normal, Trachoma, Ectropion. Cataracts. Conjunctivitis, and Pterygium.

  • - Also provides brief descriptions of the diseases

Credits

  • - My groupmates

  • - Numerous resources on python libraries and model training