Palm print identification method based on YOLOv9 and Real-ESRGAN

Biometrics are commonly used in e-commerce, payment, and access control applications. Yet, after the global COVID-19 pandemic, people have become accustomed to wearing masks and regularly disinfecting contact identification devices. But, because the mask covers the face or the contact device, it can cause problems with recognition performance and hygiene. To this end, we design palmprint recognition systems that are difficult to forge and do not require contact. It overcomes the limitations of previous contact-based palm scanning and incorporates Real-ESRGAN for data preprocessing. The next step is to combine YOLOv9 for training and identification. This allows for good performance with small amounts of data. The experimental results show that the accuracy can reach $\mathbf{9 3. 5 7 \%}$.