Here are the Codes, Datasets and the Android App required to create the Skin Cancer Detection App.

The dataset is orignally available at https://dataverse.harvard.edu/dataset.xhtml?persistentId=doi:10.7910/DVN/DBW86T . follow the steps given on https://github.com/syed-hamza/Skin-Cancer-Android-App/blob/master/Data.md for dataset prepartion.

The data should be arranged in the following struture.
https://github.com/syed-hamza/Skin-Cancer-Android-App/blob/master/input.md

1. Inception-V3 - https://github.com/syed-hamza/Skin-Cancer-Android-App/blob/master/src/SkinCancer_inceptionModel.ipynb
2. nasnet - https://github.com/syed-hamza/Skin-Cancer-Android-App/blob/master/src/SkinCancer_nasnetmobile.ipynb

The given code uses inception_v3 at https://keras.io/applications/#inceptionv3.Inception model gave me a slightly better result than nasnetmobile and training the model was comparitively quick.

validation accuracy:85.76%

validation accuracy:84.72%

From this we can conclude that eventhough nasnet fluctuates less than inception inception provides a better accuracy,hence is better suited for the app.

I am a student in highschool and this is my first project using Deep Learning. The following project took me several weekends along 2 months. I hope it is helpful to everyone reading this. All suggetions are welcome.