Let an Oosto expert show you how to protect your customers, guests, and employees with touchless access control, video monitoring, and real-time watchlist alerting.
Actions always speak louder than words. That’s why we’ve embedded ethical AI and data privacy into our facial recognition solutions and processes — pioneering new methods to reduce demographic bias, comply with privacy regulations, and protect the identities of bystanders.
Oosto leverages millions of representative images and state-of-the-art augmentations from a variety of data sources, composed of people from different races, skin tones, ethnicities, ages, and genders.
We have identified as many as 60 different metatags for a single photographic image. This process is performed algorithmically. Our data scientists audit the quality and accuracy of the metadata.
We do not use the Fitzpatrick Skin Type scale which is prone to error and misclassification. Our AI-based technology gives us much higher resolution in the skin color as we capture millions of skin tone categories from video streams.
After the metadata has been applied and audited, we leverage our neural network to cluster the images to help find areas (not just skin color) where bias may have crept into the model.
Oosto’s training data is based on images of people in real world conditions — from large crowds to low light environments, extreme angles, and obscured faces.
We take seriously our responsibility to comply with laws and regulations intended to protect our customers and the public at large.
Upon the detection of a face, we have enabled the option to blur the faces of bystanders during video playback that were not enrolled to the watchlist.
Oosto enables our customers to choose whether to record the faces of individuals not enrolled in the watchlist. Once this feature is enabled, the images of faces of non-enrolled individuals will not be displayed in the gallery and will not be saved on the server. This enables our clients to avoid unnecessary compliance risk and only record and process detections of individuals on an organization’s watchlist.
Generally, facial recognition involves identification based on the comparison of newly captured images with images stored in a database of reference images, typically photos of people, to find possible matches. Some solutions have amassed billions of facial images from Google, LinkedIn, Facebook, and other popular websites, usually without the consent of those people.
Conversely, Oosto starts each commercial engagement with a blank slate as our reference database — and we only identify persons of interest based on who has already been flagged as potentially dangerous, criminal, or unauthorized.
Oosto is not a SaaS company per se, so we don’t have access to our customers’ systems or their video data. Our software runs in each customer’s closed environment and does not require any outside connections.
Even when the data is at the customer site, we only store the vector images of POI — that is we maintain a minimum of data locally as required for processing watchlist alerting. Any image of a non-POI is only stored in RAM and not on the local server. The Oosto system only stores mathematical vectors of the POI persons. All facial signatures and facial images can be encrypted in transit with AES-256 bit (based on customer needs) from the camera to the local server.
We recognize that facial recognition has the potential to be misused if placed in the wrong hands, and that we have an inherent responsibility to ensure our technology and products are used properly. As stated in our end-user license agreement, all customers are prohibited from using the technology for inappropriate, improper, or unlawful purposes.
Commitment to AI Ethics
Oosto understands the great value and potential of its technology and systems, as well as the significant benefits they can provide to society. At the same time, we recognize that such powerful technology has the potential to be misused if placed in the wrong hands, and that we have an inherent responsibility to ensure that our technology and products are used properly.
Check out the six ethical principles that guide us here.