Age-Invariant Face Recognition Based on Self-Supervised Learning

Face recognition across aging has grown into an extremely prominent and challenging job in the field of facial recognition in recent times. Many researchers have made improvements to this field, but there is still a massive gap to fill in. This study explores the effectiveness of Self-Supervised Learning (SSL), specifically the Bootstrap Your Own Latent (BYOL) technique, to improve age-invariant facial recognition models. The experimental results demonstrate that this method greatly enhances the model’s performance, achieving accuracy gains from up to more than 5%, even on challenging datasets such as FGNET. These findings highlight the potential of SSL methods such as BYOL in advancing the field of age-invariant facial recognition.