Skin Disease Detection Using Transfer Learning with VGG19: a Convolutional Neural Network Approach
Year: 2026
Type: conference-paper
Abstract: Millions of people worldwide have skin conditions and these pose enormous problems regarding early and proper diagnosis. Early diagnosis and classification is the key to successful treatment and an improvement in patient outcomes. The traditional methods of diagnosis would have become over reliant on the expertise of the skin and may have been prone to cost, time and even subjective bias. To check accuracy of model, a data got from Kaggle, containing 10000 high definition skin disease snapshots categorised into six training, are used: Enfeksiyonel, Ekzama, Akne, Pigment, Benign, and Malign. The generalizability of the model was improved by a large variety of preprocessing and augmentation techniques. The model that was developed could attain training and validation accuracy of$\mathbf{9 0. 1 2} \boldsymbol{\%}$and$\mathbf{8 5. 4 7 \%}$respectively. As a way of implementing the system in a real-life scenario, it is suggested that the system be introduced in a web-based platform of real-time image uploading and diagnosis. The findings suggest that AI-powered diagnostic systems have the potential to offer scalable, accurate and efficient dermatological services.
Institution: Siddhartha Academy of Higher Education
Language: English
Subfield: Oncology
Field: Medicine
Domain: Health Sciences
Open Access status: hybrid
OpenAlex ID: W7204809328