SmartSkinAI: Innovations in Dermatological Preliminary Diagnosis using AI

SmartSkinAI: Innovations in Dermatological Preliminary Diagnosis using AI

Prof.Shubhangi Deshpande, Mansi Mokashi, Tanaya Chavan, Pritam Patil, Samiksha Pathare

Department of Information Technology

PVG’s College of  Engineering and Technology & G.K.Pate(Wani) Institute of Management

Pune, India

Abstract - Dermatological conditions are becoming predominant across the globe.People in different geographical locations suffer from various skin diseases such as acne, melonama, eczema and many more.However, they are unaware about the severity of these skin diseases as they may worsen with time.Preliminary diagnosis of skin conditions play a vital role in prediction of the disease. Skin disease diagnosis at present includes a series of pathological laboratory tests for the identification of the correct disease. At the moment, AI-powered diagnostic technologies can help physicians in making quicker and more precise diagnoses.This research proposed an efficacious solution by implementing Convolutional Neural Network architecture MobileNet(v2).The dataset is gathered from various sources to collect images of ten different skin disorders.Then classification techniques are implemented for predicting the type of skin disease.The trained model is deployed on the web using Django framework and the recognition of skin diseases can be done remotely using this system. Out of different performance evaluation measures,accuracy and loss is calculated to verify the working of model. Our model comprising of MobileNet(v2) achieved accuracy upto 99.00%. Moreover, we proposed and implemented a web-based model for the real-time recognition of skin diseases. This approach can aid health professionals by recognizing different skin diseases more efficiently and making the diagnosis process more user-friendly for the patients.

Key Words:  Convolutional Neural Network (CNN), skin disease, MobileNet V2, classification, preprocessing,train-test split