Penerapan Upscaling GFPGAN Terhadap Klasifikasi Citra Ekspresi Wajah Manusia Menggunakan Arsitektur VGG19

Facial expression recognition is a rapidly evolving field in artificial intelligence and digital image processing. However, a major challenge in facial expression classification is low image quality, which can reduce model accuracy. This study proposes the application of an upscaling method using Generative Facial Prior
Generative Adversarial Network (GFPGAN) for facial expression classification with a Convolutional Neural Network (CNN) based on the VGG19 architecture. Experiments were conducted in two scenarios: classification without upscaling and classification with GFPGAN upscaling. The results showed that the model without
upscaling achieved an accuracy of 76%, while the model with GFPGAN upscaling improved to 86%. This increase demonstrates that image quality plays a crucial role in enhancing facial expression classification accuracy. Future research can be
developed by increasing the dataset size and implementing a more optimized upscaling method to further improve model accuracy.