MobiDerma- Skin Condition Detection

Home Page
Skin detection Example
Skin Disease Classification: Improved Deep Learning model with better accuracy
Skin Severity Classification: Improved Deep Learning model with better accuracy

Inspiration

Current skin and health condition in Australia especially in rural areas, where it is difficult to find GP and have to travel remotely just for a medical visit.

What it does

Firstly it classifies different skin diseases like Acne, Psoriasis etc. and secondly, once skin disease is predicted it further classifies the severity of that detected disease and recommends the nearest dermatologist and health care centre for further medication.

How I built it

We built it using Xcode for mobile application and used TensorFlow to convert deep learning model and once app approximates that user needs medical attention than it lists the doctors and fixes an appointment with him/her.

Challenges I ran into

Integration into iOS app

Accomplishments that I'm proud of

Successfully integrated into iOS app

What I learned

How to train the Deep Learning model using TensorFlow

What's next for MobiDerma- Skin Condition Detection

To scale our Deep Learning model for all sorts of skin diseases which can be visually approximated. We wish to expand it to all dental problems as well, using the core concept of deep learning.

Built With

ios
tensorflow
xcode

Submitted to

UNIHACK 2019

Created by

I developed the iOS application to create the UI and integrate ML model into the application. I assisted my teammates in training ML model on TensorFlow.

Ojas Chimane
I worked on the training of ML model which was challenging in order to accumulate and train the dataset. Got to learn a lot on the ML front. Also helped in developing the idea of the business model.
The inspiration came from the fact that this application can improve the rural and remote sector health conditions.

karthikkalyanarama
I worked on the backend. I used Tensorflow to Train the data used for ML and AI Learning. After that converted the trained image data with .pb extension to IOS .mlmodel extension using tf-coreml.

Satabdi Dash
I implemented deep learning algorithm named Mobilenet which Convolution Neural Network Model to classify 2 things:
Classify Skin Diseases like Acne, Psoriasis and normal skin condition,
Secondly,
Once Skin disease is classified, trained and predicted second model to classify different degree of severity like Psoriasis Mild condition and Psoriasis Severe condition.

Both the models predicts with great accuracy.

I also collected different quality image data from internet.

Akshay Ijantkar
Data Scientist with 1 year relevant experience in data science,ML and DL and 2 years of experience in engineering domain like IoT.

Updates

Ojas Chimane started this project — Mar 16, 2019 08:59 PM EDT

Leave feedback in the comments!

Log in or sign up for Devpost to join the conversation.