June Trimester 2025 Title: Deep Learning-Based Classification of Skin Diseases and Cancer
The project have developed a deep learning model to classify seven types of skin lesions using the HAM10000 dataset. Four architectures — EfficientNet-B0, DenseNet-121, InceptionV3, and Swin Transformer — were compared, with the Swin Transformer achieving the best accuracy of 88.9%.
The project demonstrates the potential of AI to assist dermatologists in early skin cancer detection.
Application-Based Project
Group 5:
- Loh Chia Heung (Leader) 2301684
- Tan Yi Xin 2101990
- Bester Loo Man Ting 2207066
- Cornelius Wong Qin Jun 2104603
Model Selection In This Project (4):
- Loh Chia Heung: Swin Transformer
- Tan Yi Xin: EfficientNet-B0
- Bester Loo Man Ting: DenseNet-121
- Cornelius Wong Qin Jun: Inception V3
Kaggle Dataset Link:
https://www.kaggle.com/datasets/kmader/skin-cancer-mnist-ham10000