StyleID: A Perception-Aware Dataset and Metric for Stylization-Agnostic Facial Identity Recognition
Researchers have introduced StyleID, a new dataset and evaluation framework designed to improve facial identity recognition in stylized images. Current identity encoders struggle with artistic transformations like cartoons or paintings, often misinterpreting stylistic changes as identity shifts. StyleID aims to address this by using human perception data to fine-tune these encoders, making them more robust to out-of-domain and artist-drawn portraits. AI
IMPACT Improves robustness of facial recognition models to artistic stylization, potentially impacting applications in digital art and media.