Development of Class Topper Optimized Transfer Learning-Mobile Net Classifier Model for Effective Skin Cancer Classification

Skin cancer is a rapidly growing health concern, necessitating early detection for better treatment outcomes. Traditional diagnostic methods, while effective, often rely on specialist expertise, which is limited in availability. To address this, automated systems for skin lesion classification are increasingly important. This project proposes a detection of Skin Cancer through the implementation of CTO-based Mobile Net Classification. The approach begins with preprocessing the image dataset using the Wiener filter (WF), which reduces noise and enhances image contrast, providing clearer data for subsequent steps. Following this, Fuzzy C-Means (FCM) clustering is employed for segmentation, enabling efficient grouping of image regions based on similarity. The segmentation process ensures precise identification of key regions, which are crucial for extracting texture-based features, capturing critical patterns in the image data. To optimize the classification process, the Class Topper Algorithm is employed for feature selection, refining the feature set by retaining the most significant attributes. Finally, Mobile Net, a lightweight deep learning model (DLM), is used for classification, efficiently handling image data while maintaining high accuracy. The combination of these techniques results in enhanced classification performance, providing accurate output images. This project is implemented using MATLAB software.