Skin Disease Classification: A Comparison of ResNet50, MobileNet, and Efficient-B0 | ||
| Journal of Current Multidisciplinary Research | ||
| Volume 1, Issue 1, January 2025, Pages 1-7 PDF (410.86 K) | ||
| Document Type: Original research | ||
| DOI: 10.21608/jcmr.2025.327880.1002 | ||
| Authors | ||
| Mahmoud Y. Shams* 1; Esraa Hassan1; Sara Gamil1; Aya Ibrahim1; Esraa Gabr1; Sarah Gamal1; Esraa Ibrahim1; Fares Abbas1; Ahmed Mohammed1; Anwar Khamis1; Mohamed Hamed1; Mahmoud Mokhtar1; Roheet Bhatnagar2 | ||
| 1Faculty of Artificial Intelligence, Kafrelsheikh University, Kafrelsheikh 33516, Egypt | ||
| 2Department of Computer Science and Engineering Manipal University Jaipur,Rajasthan, India | ||
| Abstract | ||
| Background: Skin diseases are among the most common health issues worldwide, affecting mil-lions of individuals annually. Conditions such as Acne and Rosacea, Eczema, Exan-thems and Drug Eruptions, Scabies, Lyme Disease, Tinea and other fungal infections, and Vasculitis can significantly impact patients' quality of life. Early and accurate diagnosis is crucial for effective treatment and management. However, the diagnosis of skin diseases often requires specialized expertise, which may not be readily availa-ble in all healthcare settings. Misdiagnosis or delayed diagnosis can lead to inappro-priate treatments and worsening conditions. Methods: In this paper, we propose Deep Learning techniques for classifying six common skin diseases. Leveraging the DermNet da-taset, we utilized the EfficientNet-B0 model, known for its accuracy and efficiency, to categorize these dermatological conditions. Our methodology involved data aug-mentation, transfer learning, and fine-tuning the EfficientNet-B0 model. Results: The pro-posed approach achieved an impressive 99% accuracy on the validation set, demon-strating the potential of advanced convolutional neural networks for automated skin disease diagnosis. Furthermore, we compared EfficientNet-B0 with other popular models, including ResNet50 and MobileNet, revealing superior performance both in accuracy and computational efficiency. Specifically, the models achieved accuracy rates of 93%, 94%, and 99% for ResNet50, MobileNet, and EfficientNet-B0, respec-tively. Conclusion: These findings highlight the reliability and effectiveness of the proposed model compared to state-of-the-art approaches.Tumors have been rarely documented in the Arabian dromedary (Camelus dromedarius). | ||
| Keywords | ||
| Dermatological Conditions; Data Augmentation; EfficientNet-B0; Receiver Operating Characteristic | ||
|
Statistics Article View: 619 PDF Download: 487 |
||
