Perkembangan pesat teknologi deep learning, khususnya Convolutional Neural Network (CNN), menawarkan potensi besar dalam diagnosis penyakit kulit yang menjadi masalah kesehatan signifikan di Indonesia. Tingginya prevalensi penyakit kulit seperti Eszema dan Psoriasis, dipengaruhi oleh faktor lingkungan dan keterbatasan akses layanan dermatologi, menyoroti urgensi solusi deteksi dini berbasis teknologi. Penelitian ini bertujuan membandingkan kinerja arsitektur VGG16 dan ResNet101 dalam klasifikasi citra penyakit kulit Eszema dan Psoriasis pada wajah. Dataset citra diperoleh dari Kaggle, mencakup 1.677 gambar Eszema dan 2.056 gambar Psoriasis. Kedua model arsitektur CNN, VGG16 dan ResNet101, dilatih dan dievaluasi menggunakan metrik akurasi, presisi, dan recall. Perbandingan ini akan mengidentifikasi kelebihan dan kekurangan masing-masing arsitektur dalam konteks klasifikasi penyakit kulit spesifik ini. Hasil penelitian diharapkan dapat memberikan kontribusi signifikan terhadap pengembangan sistem klasifikasi penyakit kulit yang lebih efisien dan akurat. Model yang dihasilkan berpotensi menjadi alat bantu diagnosis awal bagi tenaga medis dan dermatologis, serta sarana deteksi dini bagi masyarakat umum, mendukung peningkatan kesehatan kulit di Indonesia.

The rapid advancement of deep learning technology, particularly Convolutional Neural Networks (CNN), offers significant potential for diagnosing skin diseases, which pose a substantial public health issue in Indonesia. The high prevalence of conditions like Eczema and Psoriasis, influenced by environmental factors and limited access to dermatological services, underscores the urgent need for technology-driven early detection solutions. This study aims to compare the performance of VGG16 and ResNet101 architectures in classifying facial images of Eczema and Psoriasis skin diseases. Image datasets were sourced from Kaggle, comprising 1,677 Eczema images and 2,056 Psoriasis images. Both CNN architectures, VGG16 and ResNet101, were trained and evaluated using accuracy, precision, and recall metrics. This comparison will identify the strengths and weaknesses of each architecture in the context of this specific skin disease classification task. The research findings are expected to contribute significantly to developing a more efficient and accurate skin disease classification system. The resulting models have the potential to serve as initial diagnostic aids for medical professionals and dermatologists, as well as an early detection tool for the general public, thereby supporting improved skin health in Indonesia.