Fungal Skin Disease Classification Using Densenet-201

Fungal skin infections, such as ringworm, athlete's foot, nail fungus, are common diseases that frequently occurs and causes discomfort, it needs to diagnose as early as possible but it seems challenging to diagnose by using Traditional diagnostic methods as they get delayed and time consuming, Also the existing automated solutions use CNN, which has low accuracy. It is necessary to have better automated solutions for quicker and more accurate detection. This project develops a deep learning system with DenseNet-201 for classifying fungal skin diseases. It is chosen as it has strong feature extraction and accuracy in medical image analysis. It includes Deep Generative Adversarial Networks (DGANs) Data augmentation techniques that increase the dataset from 450 to 4,500 images, and generate synthetic images that addresses data scarcity and imbalance. Transfer learning and the Stoichastic Gradient Descent (SGD) optimizer is used to improve training efficiency and stability by adjusting weights more effectively during backpropogation. The model achieved a classification accuracy of 97 %, that offers a reliable tool for diagnosing fungal infections. The results shows that this model provides a high level of accuracy in diagnosing fungal skin diseases, that makes it a reliable tool for assisting dermatologists in making faster and more precise diagnoses. In conclusion, this deep learning approach is a big step forward in dermatology and has potential to provide broader medical image classification tasks in the future.