Sightcorp logo

Facial Recognition Search

Everything about Facial Recognition Search

     

What is facial recognition search?

Facial recognition search is AI-based technology that makes it possible to search a database of images for facial images that match a given face. In other words, if you provide it with an image of a face, the technology will search through an associated database and return all images that contain that face.

How does facial recognition search work?

This technology uses various AI algorithms to perform the following tasks:

  • Detect the face or faces within an image
  • Convert each facial image into a faceprint (unique code)
  • Compare a given faceprint to a database of faceprints to determine whether there is a match

 

How you can use facial recognition search

You can use facial recognition search technology in various applications, including access control systems, search engines, entertainment apps, and KYC (Know Your Customer) procedures. Here are some examples:

Access Control:

In a biometric access control system that uses face recognition, a camera is placed at the access point. This camera takes a picture of each person as they attempt to gain access, and sends the picture to a computer running face recognition software.

The software then generates a faceprint for the image, and compares this faceprint to the database of faceprints associated with people who are authorized for access.

If there is enough similarity between the newly generated faceprint and one of the faceprints in the database of authorized persons, access is granted. If no match can be established, access is (temporarily) denied, and an alert may be triggered. Depending on how the system is set up, an alternative authentication method, such as a manual check, may be initiated at this point.

KYC:

In KYC systems as used by banks and other financial institutions, face recognition can be used to help eliminate fraud during the customer onboarding process.

How this works is that when a prospective customer wants to open an account, they provide a facial image to the bank during the registration process. This facial image is then processed by facial recognition software, using the same steps as in the access control example above. Instead of comparing the faceprint to a database of authorized persons, however, the faceprint may instead be compared to a database of persons who have been blacklisted or identified as known fraudsters. If there is a match, an alert will be triggered and further investigation will be required.

Entertainment Apps:

Fun apps such as celebrity lookalike finders Looky and StarByFace allow users to submit their facial images and find out which celebrities they most resemble. These apps also use facial recognition algorithms, but often with a lower confidence threshold to increase the possibility of finding a match, since finding an exact match is not the purpose of the app.

Search Engines:

Search engines use facial recognition search to enable users to search the internet for faces that are similar to the face in an image submitted by a user. So, for example, if a user has only an image of a person, and they want to find out who it is, they can upload the image to the search engine. The search engine will then use facial recognition algorithms to find all images on the web containing faces that are sufficiently similar to the one submitted by the user.

The same concept is used in digital photo albums, such as Google Photos, to help users easily find photos of a specific person, or to group all photos containing certain individuals.

Learn More About What You Can Do With Face Search

Where to find facial recognition search technology

Facial recognition software companies specialize in providing this technology to clients who want to integrate it into their own systems or develop their own apps using this technology.

If you’re looking for state-of-the-art, 100% deep learning-based face recognition algorithms that deliver best-in-class accuracy, you can contact Sightcorp to learn more about our facial recognition product, FaceMatch.

Request Trial

Here are other articles you might find interesting:

 

   

Technical Specifications

The table below shows how FaceMatch SDK performs on the Labelled Faces in the Wild (LFW) dataset:

FPRTPRThreshold (Inverse of distance)
0.10.99900 ±0.002130.55448
0.010.99667 ±0.005370.59791
0.0010.99367 ±0.006050.62989

FPR = False Positive Rate
TPR = True Positive Rate

These results are an indication only and are based on the specific dataset Labelled Faces in the Wild. Customers can expect similar performance, with possible variations due to hardware and the availability of annotated data.