AI-Based Skin Disease Detection Using Convolutional Neural Network

DOI

10.2991/978-94-6239-789-7_19How to use a DOI?

Keywords

Skin disease detection; CNN; HAM10000; MobileNetV2; EfficientNet; transfer learning; deep learning

Abstract

Skin diseases are a problem for millions of people all around the world. It is really important to find out what is wrong with someone’s skin away so they can get the help they need. Usually doctors who are experts in skin problems are the ones who can tell what is wrong. Sometimes these doctors are not available to help people who live in places with limited resources. This paper is about a system that uses computers to help find skin diseases. This system was trained using a lot of pictures of skin problems called the HAM10000 dataset. There are 10,015 pictures, in this dataset. They show seven different types of skin diseases: Actinic Keratosis, Basal Cell Carcinoma, Benign Keratosis, Dermatofibroma, Melanoma, Melanocytic Nevus and Vascular Lesion. Three convolutional neural network (CNN) architectures were implemented and compared: a custom-built CNN, MobileNetV2 (transfer learning), and EfficientNet-B0 (transfer learning). The models were trained with class-weighted loss functions to address the significant class imbalance present in the dataset. The custom CNN achieved the highest validation accuracy of 54.02% and an F1-score of 0.3602 after 10 epochs. MobileNetV2 reached a best validation accuracy of 50.57%, while EfficientNet-B0 showed poor convergence under the given training configuration. Results highlight the challenges of multi-class skin lesion classification on imbalanced dermoscopy data and suggest that extended fine-tuning, data augmentation, and architecture improvements can significantly boost performance.

Copyright

© 2026 The Author(s)

Open Access

Open Access This chapter is licensed under the terms of the Creative Commons Attribution-NonCommercial 4.0 International License (http://creativecommons.org/licenses/by-nc/4.0/), which permits any noncommercial use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license and indicate if changes were made.

Download article (PDF)

TY  - CONF
AU  - Gungun Verma
AU  - Vishakha Sharma
AU  - Yashika Yashika
AU  - Md. Shahid
PY  - 2026
DA  - 2026/10/08
TI  - AI-Based Skin Disease Detection Using Convolutional Neural Network
BT  - Proceedings of theInternational Conference on Innovation and Practices in Artificial Intelligence (ICIPAI 2026)
PB  - Atlantis Press
SP  - 265
EP  - 280
SN  - 1951-6851
UR  - https://doi.org/10.2991/978-94-6239-789-7_19
DO  - 10.2991/978-94-6239-789-7_19
ID  - Verma2026
ER  -