A Hybrid CNN–YOLO Framework for Skin Disease Classification and Real-Time Lesion Detection

Abstract

Skin diseases are a significant global health issue. They can be as mild as common rashes or as serious as deadly cancers like melanoma. Getting the right diagnosis quickly is key it helps save lives and lowers healthcare costs. Traditionally, doctors examine the skin with their eyes and rely on their experience, but this approach can vary from doctor to doctor, take a lot of time, and isn’t always available everywhere. Now, thanks to breakthroughs in Artificial Intelligence (AI) and especially Deep Learning (DL), computers can help doctors analyze medical images and make more accurate diagnoses automatically.In this work, we introduce a smart system for detecting skin diseases. It uses two powerful AI tools: Convolutional Neural Networks (CNN) to sort images into different disease types, and You Only Look Once (YOLO) to quickly find and highlight problem spots on the skin. We trained and tested this system with a large set of skin images called the HAM10000 dataset, which includes many kinds of skin conditions. The CNN learns to recognize unique patterns in the images, while YOLO helps pinpoint exactly where the skin problem is— and it does this in real time. Our results show that this combined approach is very accurate at classifying diseases and spotting the affected areas, making it a useful tool for doctors and healthcare workers in real-world settings.