Technology and medicine together can create very powerful tools to help patient and doctors. Moreover, computers can detect some disease that is hardly detectable with human eyes. Even so, doctors need years of studies and experience to be able to accurately detect what a good machine learning model can in a shorter amount of time.
What it does
Based on an image of a person's retina, it predict the person's severity of diabetic retinopathy.
How I built it
I used a pre-trained machine learning model, the VGG 16, and I did transfer learning on the dataset of retina images. Moreover, I did data augmentation on my dataset since the dataset did not contain a large amount of images.
Challenges I ran into
I was very new to machine learning and I had no idea how to implement a convolutional neural network. Hence, I had to do a lot of reading and research in order to achieve my goal.
Accomplishments that I'm proud of
Even if my project is relatively simple, I am proud that I have achieved this on my own.
What I learned
I learned a lot about convolutional neural networks and computer vision. Moreover, I learned how to do transfer learning and image data augmentation. I also learned how to use tools from libraries like pandas, numpy and keras.
What's next for Blindness Detector
Next, I would like to implement a web application so that my model can be used to predict one's retina's severity of diabetic retinopathy.
Log in or sign up for Devpost to join the conversation.