Skin Docnet: Real-Time Melanoma Vs Benign Classification Using Efficientnet

The proposed Skin DocNet system demonstrates the application of deep learning for automated melanoma and benign skin lesion classification using the EfficientNet architecture. The developed approach integrates image preprocessing, input validation, deep learning-based classification, and web-based result visualization into a unified framework. The system processes an uploaded skin lesion image and generates a classification result with associated confidence information. Grad-CAM is incorporated to provide a visual interpretation of the model's decision, thereby improving the transparency of the prediction process. The overall framework demonstrates the potential of combining efficient deep learning architectures with accessible web technologies for preliminary skin lesion assessment. Although the proposed system provides a useful foundation for computer-assisted skin lesion screening, further research is required before it can be considered for clinical use. Future work can focus on expanding the dataset with more diverse images and increasing the number of lesion categories. Improving model generalization across different imaging conditions and patient populations is also important. The system can additionally be extended to mobile platforms for image-based assessment using smartphone cameras. Advanced explainable-AI methods and ensemble learning techniques may be investigated to improve prediction reliability and interpretability. Integration with hospital information systems and cloud-based deployment could further improve accessibility, scalability, and information exchange. Real-time image or video analysis can also be explored as an extension of the current static-image classification framework.

[1] M. Alruwaili and M. Mohamed, “An Integrated Deep Learning Model with EfficientNet and ResNet for Accurate Multi-Class Skin Disease Classification,” 2025. [2] Y. Gulzar, S. Agarwal, S. Soomro, M. Kandpal, C. W. Onn, S. Saini, and A. Bounsiar, “Next-generation Approach to Skin Disorder Prediction Employing Hybrid Deep Transfer Learning,” 2025. [3] A. Aboulmira, H. Hrimech, M. Lachgar, M. Hanine, C. O. Garcia, G. Mezquita, and I. Ashraf, “Hybrid Model with Wavelet Decomposition and EfficientNet for Accurate Skin Cancer Classification,” 2025. [4] M. Wei, Q. Wu, H. Ji, J. Wang, T. Lyu, J. Liu, and L. Zhao, “A Skin Disease Classification Model Based on DenseNet and ConvNeXt Fusion,” 2023. [5] R. Sadik, A. Majumder, A. A. Biswas, B. Ahammad, and M. M. Rahman, “An In-Depth Analysis of Convolutional Neural Network Architectures With Transfer Learning for Skin Disease Diagnosis,” 2023. [6] L. Hoang et al., “Classification of Multiclass Skin Lesions using Lightweight Deep Learning,” 2022. [7] K. A. Muhaba et al., “Deep Learning-Based Automatic Skin Disease Diagnosis from Clinical Images and Patient Data,” 2022. [8] S. N. Almuayqil, S. Abd El-Ghany, and M. Elmogy, “A Hybrid Deep Learning Model for Feature Fusion in Multi-Type Computer-Aided Diagnosis for Early Skin Disease Signs,” 2022. [9] P. R. Kshirsagar, H. Manoharan, S. Shitharth, and A. M. Abdulrhman, “Deep Learning Methods for Automated Skin Disease Prognosis,” 2022. [10] V. Anand, S. Gupta, D. Koundal, and S. Mahajan, “Automated Skin Disease Diagnosis Using Dermoscopy and Deep Learning,” 2022.