Machine Learning Approach for Skin Cancer Detection and Classification

Abstract

The Skin Cancer Detection System shows the whole process and web app features are made in this study. Security is kept strong with login and registration pages that include password rules and user authentication. Once logged in, where users can see the main screen and  upload a dermoscopic image or take a live photo with their camera for quick skin cancer prediction. Both uploaded and captured images are processed by a deep learning system that is built with Flask. This system identifies the skin lesion and classifies it as either melanoma or benign. It also creates visual explanations called Grad-CAMs, which highlight the parts of the image that helped the model make its decision, making the results easier to understand. The system has a real-time mode that analyzes video frames on the fly to detect any unusual changes. It also keeps user data private by storing everything locally. The results screen shows the predicted category, confidence level, and the area of the lesion, making it simple for both medical experts and regular users to understand the diagnosis. In short, this system is an accurate, easy-to-use, and ethical AI tool that helps detect skin cancer at an early stage.