How is Anonymisation done?
Local facial keypoints computing
Anonymization is responsible for the handling of the camera, loading an anonymization model, and converting images into facial key points locally. No images are sent to the cloud. For transparency and auditability, the code is open source.
Open Source Code
👉 Checkout the GitHub repository GitHub
Facial key points are computed on-device. The anonymization model takes images as input and extracts hundreds of points estimated in 3D space. The anonymization computation duration depends on the CPU of the local machine. Recent laptops and smartphones compute at 30 frames per second.
Why is anonymity important?
Emotions are by essence, a very intimate information. We provide facial expression recognition software for use cases that are aligned with our values : contribute to human well-being, helping people connect with each other and respect each individual’s right to privacy.
We make no compromise on guaranteeing our customers' trust.
Trust, respect & ethics
Acceptability for such technology is key, and trust is as important as the use case it deserves.
That’s why we have designed a perfectly private-friendly architecture where no identifiable information is sent to the cloud.