Penerapan Convolutional Neural Network dengan ResNet-50 untuk Klasifikasi Penyakit Kulit Wajah Efektif

Skin diseases have a high prevalence in Indonesia, reaching 12.95% of the population, so early detection is an important step in handling them. This research aims to implement deep learning based on Convolutional Neural Network (CNN) with ResNet-50 architecture to improve the accuracy of facial skin disease classification through medical images. The data used comes from the Augmented Skin Conditions (Kaggle) dataset with a total of 2,394 images, which are processed through preprocessing, augmentation, and division of training and testing data with a ratio of 80%: 20%. The augmentation process resulted in image variations, but vertical distortions were found due to zooming settings and possible shearing effects. The model achieved an accuracy of 94%, higher than the previous study on pneumonia classification using ResNet-50, which obtained an accuracy of 86% and was affected by data imbalance and similarity of visual features between classes. These results show that ResNet-50 can overcome the vanishing gradient problem and extract complex features from medical images optimally. With this performance, this model can be applied in artificial intelligence systems to assist medical personnel in the early detection of skin diseases quickly, accurately, and efficiently.