Skin diseases pose significant challenges in the field of dermatology. In recent years, Convolutional Neural Networks (CNNs) have emerged as a powerful tool for image recognition and analysis tasks. This research paper presents a comprehensive study on the application of CNNs for skin disease diagnosis. We propose a CNN-based framework for skin disease diagnosis, which utilizes a large dataset of dermatological images to accurately identify various skin diseases. The proposed model leverages the deep learning capabilities of CNNs to learn discriminative features from input images, enabling accurate and efficient diagnosis. We demonstrate improved accuracy and efficiency in skin disease diagnosis by employing pre-trained models. Our proposed model enables accurate classification of skin diseases into high, medium, and low severity categories by leveraging a large dataset of annotated images, assisting healthcare professionals in prioritizing treatment strategies. In conclusion, this research paper presents a comprehensive study on the application of CNNs for skin disease diagnosis, skin lesion classification, melanoma skin cancer classification, and skin disease severity classification. The proposed models showcase significant advancements in the field of dermatology, providing accurate and efficient tools for dermatologists and healthcare professionals. The findings of this research contribute to improving the diagnosis, classification, and severity assessment of skin diseases, ultimately enhancing patient care and treatment outcomes.