Skin-Cancer-Detector-Android

Overview:

This Repo contains the project done by:

Nikhil Jain

Kushagra Srivastava

Nhan Ton

Rebecca Wang

For the 8th HackUmass hackathon.

The original idea was forked from ________________

and we have used that basic detector as a canvas to create an all-in-one health application, complete with features such as:

->Google Maps Integration

->References and Resources

->A complete UI/UX overhaul

->Emergency Contacting features

TODO:

1. Integrate Maps

2. Fix UI

3. Contact Features

4. Elaborate upon relevant documentation

ReadMe from the original rep (that covered only the skin detector part):

Porting my previous Skin Cancer Detector developed in Python using Tensorflow and Keras over to Android. The application lets the user to use an existing image from a gallery or take a photo in real time. The program will then display its prediction.

Skin Cancer Detector achieved a total accuracy of 87% after being trained on ISIC, PH2, and Complete MedNode online databases.

Note: Since the model was trained on high quality images taken in various hospitals, in order for the program to predict correctly, the inputted image needs to be of as high quality as possible for accurate predictions. It's difficult for the program to recognize the skin cancer features on low quality images.

WARNING: The program is to be used as a supplementary tool. You should not be solely relying on the predictions of the program. If you're concerned about skin cancer, do not hesitate to contact a professional dermatologist.