
Research Article
Efficient Combination of Deep Learning Models for Skin Disease Detection
@INPROCEEDINGS{10.1007/978-3-031-86493-3_23, author={Mohamed Massamba Sene and Ndeye Fatou Ngom and Michel Seck}, title={Efficient Combination of Deep Learning Models for Skin Disease Detection}, proceedings={Innovations and Interdisciplinary Solutions for Underserved Areas. 7th International Conference, InterSol 2024, Dakar, Senegal, July 3--4, 2024, Proceedings}, proceedings_a={INTERSOL}, year={2025}, month={4}, keywords={Skin diseases identification Image segmentation Deep learning Image classification Computer vision}, doi={10.1007/978-3-031-86493-3_23} }
- Mohamed Massamba Sene
Ndeye Fatou Ngom
Michel Seck
Year: 2025
Efficient Combination of Deep Learning Models for Skin Disease Detection
INTERSOL
Springer
DOI: 10.1007/978-3-031-86493-3_23
Abstract
Identifying skin diseases is challenging due to their similar visual appearance, making it difficult to select features. Despite significant progress in the effective identification of skin diseases, the problem remains unsolved. This paper presents a new method for accurate skin disease detection, which combines the Attention U-Net architecture for image segmentation and a customized Convolutional Neural Network (CNN) for image classification. The first step of the proposed approach is training the segmentation model on a dataset of segmented skin disease images. The resulting model is then used to segment images from a skin disease images classification dataset before training the customized CNN using the segmented images to classify skin diseases. With the proposed approach, we were able to achieve an accuracy of 99% as well as high precision, recall and F1 score of 99% on the HAM10000 dataset. Comparative analysis with other similar studies demonstrates its effectiveness in accurately identifying these diseases.