What is Face Match technology?
Face match technology, also known as face recognition technology, compares an image that contains a face to one or more other facial images and establishes whether the faces likely belong to the same person; i.e. whether they are considered a match.
What is face match technology used for?
Face match technology is used for a wide range of purposes, including the following:
Know Your Customer (KYC)
Financial service providers are legally required to enforce anti-money laundering (AML) measures, and one of these measures is KYC. This means that they need to know who they are doing business with and how much of a risk those people are likely to pose. In this context, face matching is used to verify the identity of an existing or potential customer, answering the question: is this person who they say they are?
Verification is usually carried out by using an AI algorithm to compare a selfie of the customer with the photo contained in the customer’s identity document. The algorithm then determines whether or not there is a match.
With mobile payments becoming more popular, and with the Payment Services Directive 2 (PSD2) coming into effect in the EU towards the end of 2019, biometric payment verification is becoming increasingly important. PSD2, for example, makes two-factor authentication mandatory for a wide range of customer-initiated payments (excluding those made in cash). In practice, this means that there will be an increased need for biometric identity verification.
Face matching is convenient for payment verification, as customers can easily use their smartphones to take selfies that can be compared to their enrolled image that the payment provider has in their database.
Face recognition can also be used for payments in frictionless shopping scenarios, where shoppers’ faces are scanned by facial recognition cameras as they leave a retail store. Goods can be tagged and tracked using RFID chips, and shoppers’ banking details can be stored along with their facial images in a database. Taken together, this means that shoppers’ bank accounts can be debited as they walk out of a store, without needing to go through a traditional checkout process.
When it comes to access control, face matching technology is connected to a camera, which takes a snapshot of the person trying to gain access to a room, building, event, application, or device. The face matching technology then compares the snapshot to a database of faces of individuals who have been given clearance for access. If there is a match, access is granted. One of the advantages of using face recognition matching in this context is that many people are already familiar with the technology, since face recognition is now used to unlock many newer smartphones as well as smart home security systems such as Ring and Nest.
A study has also shown that “three out of four frequent flyers in the U.S. favor the use of biometric facial recognition to identify both domestic and foreign travelers” – which shows that there is huge potential for using face recognition in the context of access control.
This refers to the process where face matching technology is used to group the same or similar-looking faces together. In a digital photo album, for example you can use face grouping to cluster all your selfies, all your photos of your best friend, and so on.
Retail stores, airports, etc. can use face matching technology to recognize when individuals who are on a list of known shoplifters, wanted criminals, or terror suspects, for example, enter the premises. These use cases would require video-compatible face matching technology (which Sightcorp currently does not provide).
Sightcorp’s FaceMatch solution
At Sightcorp, we’ve developed a deep learning-based facial recognition solution called FaceMatch. Currently, FaceMatch is available as a Software Development Kit (SDK), for Windows and Linux desktops and can run offline after a one-time authentication connection. FaceMatch delivers industry-leading performance, while being customizable in terms of confidence thresholds and giving you the option to retrain the models on your customer data.