<p>Skin cancer, particularly melanoma is a global threat to human health due to its high aggressive nature. Detection of melanoma at an early stage is important so as to boost the patient’s chances of survival, however, diagnosis at this stage is difficult. According to International Agency for Research on Cancer -specialized agency of World Health Organization, it was estimated around 353,947 cases of melanoma as well as 62,551 deaths would occur by the year 2025, indicating that early detection should be facilitated. In India, 243 new cases and 77 deaths has recorded in 2022, while the United States expect about 100,640 average new cases and 8290 average deaths in 2024. The paper focuses on developing convolutional neural networks (CNN) Model to predict the presence of melanoma skin cancer from skin lesion images of the patient. It also addresses the issues of class imbalance and differences in image quality using CNN and data augmentation. An easy-to-navigate web application is developed with Streamlit which allows users to upload pictures for diagnostic categorization; results are sent directly via a patient’s email or to the respective healthcare professional, ensuring timely follow-up action. The proposed work performed well which resulted in accuracy of around 92% compared to the existing works. The proposed work has increased the effectiveness of early melanoma diagnosis, in turn, positively affect patient’s outcomes.</p>