Literature review on classification model for human skin disease

---

Résumé

The rising global burden of skin diseases requires precise and efficient diagnostic tools to improve patient care and treatment outcomes. Recent advances in artificial intelligence, particularly deep learning, have significantly contributed to the automation of the classification of skin diseases. This review presents a detailed analysis of various classification models, including ResNet, VGG, DenseNet, EfficientNet, and Transformers, evaluating their performance based on accuracy, recall, and the F1-score. Although traditional CNN architectures remain effective, the rapid advancement of AI models underscores their growing limitations and the risk of technological obsolescence. Novel approaches, such as Transformers and hybrid deep learning frameworks, show promise in achieving superior diagnostic performance while optimizing computational efficiency. By critically addressing the strengths and limitations of existing classification models, this review offers valuable insights for researchers and dermatology professionals, facilitating the adoption of state-of-the-art AI-driven diagnostic solutions.

Mots-clés

Skin diseases; Classification model; Artificial intelligence; Deep learning; Convolutional neural network