How and why we pseudonymize people in CCTV video-feeds

After having written about why we do not use biometrics and how this improves retail customers’ shopping experience regarding their privacy, we can’t ignore the elephant in the room. We now turn to the proactive measures we take to ensure the data we need for our solutions is employed safely, making sure we are enabling our customers to reduce the amount of losses incurred through theft, without jeopardizing their customers’ personal data.

After model training for our solutions has been completed and they are able to recognise suspicious incidents reliably, further video data is only kept on location in an encrypted format, for as long as it is legal to do so. That is the very least that needs to be done, in order for our solutions to be GDPR compliant. As stated previously, the lack of precedence makes it hard to tell exactly what is required. Additionally the fact that retail stores are considered private property would  enable us to not take any additional measures. However, we would rather do too much than too little, both for the sake of ethics and to give shoppers the best customer experience possible.

We have trained our solutions to be able to recognise humans in a store’s video feed and blur out the faces of all shoppers. This does not require our solutions, to analyse facial data in any way, that might enable it to recognise individual customers’ faces. After all, it is easier to recognise something as a face without considering whose face it is. For our solution Checkout, this pseudonymization process is carried out on the devices in the store and all computation is done on premise.

With Signatrix solutions employed, store managers do not need their staff to monitor your security video feeds at all times in order to prevent theft at the PoS or open entrance area. Honest shoppers will be watched considerably less than they would be without an artificial eye on them.

These measures might sound great looking at the products we currently have available, but what about future products? Rest assured that we will keep working hard to safeguard the privacy of shoppers. Without giving away too much, we are currently working on a solution that is able to re-identify customers across feeds from multiple cameras, without having to resort to facial recognition or persistent, biometric features. We are confident that we can continue to shape the future of retail, without turning it into a dystopian one.

‍