I do not believe there is a way to fix facial recognition, because it does not respect the people we choose to be.
Facial recognition is built around two troubling objectives:
- A way of capturing our identities without our knowledge
- A way to identify specific incriminating evidence of our origins
We can see this clearly in one of facial recognition greatest triumphs – the ability to match a person’s image and identify with a similar image taken 10 years ago. That’s impressive, but it is misses the point.
At Lauretta.io, we believe that the person you are today, is not the same as the person that walked down this road 10 years ago. We believe that the lines around our eyes and the outfits we choose to wear are a true representation of the person we choose to be today.
That’s why we built Pollen, a drop-in alternative for facial recognition systems used for classification for human attributes with two major differences.
Pollen cannot be repurposed for surveillance
Pollen is not a biometric solution. It is designed to use every part of your outfit as part of how it recognises your character. This includes your hair, your shoes, your jacket and the silhouette of your bag against your body. This is important, because the identifier attached to you today, cannot be associated with the identifier attached to you on a different day. You cannot change your face and but you can easily change your outfit.
When leave a premise, your identity is forgotten, essentially eliminating the value of our commercial systems as police surveillance tools.
Pollen is built to identify you as you want to be identified
Pollen is built to classify who you are, based on the people and style that you emulate. If you dress like a woman, it will classify you as a woman. If Pollen finds itself deployed a place with people dress in a gender-fluid manner, it will designate them as such.
Pollen has no interest in who your parents are, what gender you were born. That is because it is only interested in who you want to be.
Pollen isn’t a step backwards, it is a rethinking of how we move forward
We built Pollen as a response to the ethical problems we see in facial recognition. However, we also built it to be technically superior. Using the same camera hardware, Pollen can be deployed at 5 times further away with comparable accuracy to facial recognition. Furthermore, in our lab tests, it shows absolutely zero racial bias. Finally, because every image is fundamentally anonymous, there are more than 1000 times more data available to train this system compared to facial recognition.