Automated Cataract in Images detection using Keras Neural Networks— Part 2

This is part two of the blog series, Automated Cataract Detection. In first part we how how to use OpenCV to train a cascade which can somewhat differentiate between an healthy and an cataract affected eye. For part1, please refer to below link.

Automated Cataract detection — Part 1 using OpenCV

In this blog we will try to achieve the same same thing but using Convolution Neural Networks. We will implement these networks via python’s keras 2 api. Lets get started !!

Pre-requisites :

• Image Augmenter
• Python 3.6
• CUDA 9.0 and cuDNN : Training a model like this, you would definitely wanna use that high end GPU of yours. Please refer to my previous blog for more details on installing CUDA and cuDNN on windows.
• Keras 2 API with Tensor Flow backend

Installing dependencies

For the sake of keeping the post small, I am not going into the details of keras setup. Please refer to Keras official site for that.

Pre Processing the data — Augment and Resize