Synthetic Anonymization
Synthetic Anonymization is our innovative approach to anonymizing personal information in images and videos. By generating “synthetic faces”, we utilize Generative AI to produce new facial images that do not correspond to real individuals, yet still bear statistical similarity to the original input.
Cameras are everywhere. As facial recognition technology continues to advance, our privacy faces new threats. Synthetic Anonymization is the technique that can help reclaim our privacy while still allowing researchers and analysts to make use of visual data sets.
What is Synthetic Anonymization?
Synthetic Anonymization is a method of creating synthetic facial images to losslessly anonymize visual data. Essentially, Synthetic Anonymization replaces original faces with hyper realistic, digitally generated ones, to safeguard privacy while still maintaining utility in visual data.
Instead of relying on wide-scale blurring in video footage, Synthetic Anonymization removes sensitive biometric data while preserving valuable analytical metrics to enable model training and unlock video analytics potential with real-time, lossless anonymization.
How does Synthetic Anonymization Work?
Synthetic Anonymization achieves lossless face anonymization by removing identifying characteristics (biometrics) and preserving head pose, facial expressions and eye movements.
Synthetic Anonymization leverages the generative power of AI to create unique and anonymous faces in real time. These faces do not correspond to real individuals and prevent the original face from being identified by facial recognition algorithms. Synthetic Anonymization also achieves perceptual change, preventing it from being identified by the human eye.
Who should use Synthetic Anonymization?
Synthetic Anonymization is optimal for any and all businesses that handle visual data. Whether using camera systems to collect and analyze data, developing computer vision systems, recording, storing or sending data to third parties, Synthetic Anonymization helps maximize privacy protection and demonstrate compliance with GDPR and other regulations.
Conclusion
As AI continues to evolve, it becomes increasingly reliant on visual data for machine learning. To prevent a dystopian future in which we have already lost our privacy, it is essential that we find ways to apply “privacy-by-design” principles and create alternative features where technology does not come at the expense of our fundamental right to privacy.
