Skin Cancer Image Classification Based on MobileNet-V4 with Composite Data Augmentation

Skin cancer is a prevalent cancer worldwide, necessitating effective early detection methods. This research focuses on skin cancer image classification using MobileNet-V4, leveraging its suitability for deployment on mobile devices. Addressing the challenge of imbalanced datasets in skin cancer images, a composite data augmentation technique is proposed. This technique aims to enhance model robustness and classification accuracy. The deployed model offers potential applications in using smartphones and similar devices as auxiliary diagnostic tools, thereby facilitating early screening for skin cancer patients. The study contributes to expanding the accessibility and efficiency of skin cancer diagnosis through mobile-based solutions.