Abstract:
The fundamental objective of this project is to design a prototype system which can be
used to detect skin diseases that are Acne detection, Scabies detection, Melanoma
detection, Nevus detection, Seborrheic keratosis. This paper focuses on improving the
image processing techniques to increase the quality of image and neural network
techniques to organize the skin diseases listed above. The blueprint is based on
TensorFlow which is mainly used for classification, perception, prediction and creation
since it is an open source artificial intelligence library and retraining image classifier
using complex neural network. The prototype has been experimented on four variety of
skin diseases. When a sample test image is given to prototype, the image will be tested
using old trained complex network model. As a result, by implementing this technique
skin diseases are recognized with up to 99.52 percent accuracy rates. This prototype
has the capability of further improvement in the future. This project has an API endpoint
which is the point of entry in a communication channel. When two systems are
interacting for Android application and IOS it can be used very methodically by sharing
skin diseases’ images.
This project has another objective which is to deliver as much information as possible
regarding skin diseases detection with this prototype to the general public.