Syntheic Human Aging Study

Robust authentication for low-power consumer devices such as doorbell cameras pose a valuable and unique challenge. This work explores the effect of age and aging on the performance of facial authentication methods. Due to collecting data on a person from birth to old age being difficult and the lack of large-scale age data from multiple identities in existing age datasets, synthetic aging data is adopted in our research. The figure below show an overview of our synthetic aging data.  

As a part of this study, we want to explore the quality of synthetic aging data, and the gap between the real age data and the synthetic age data. We are conducting a Mean Opinion Score (MOS) Study to enable human evaluators rate the generated synthetic aging data and we invite the general public to participate in this MOS study.

For more information on participating in this study, please click here.