Automatic face comparison, taylored to your application
Videmo's FaceSDK Software helps you make sure that two pictures actually depict the same person. It analyzes the images and outputs its result in mere fractions of a second. The FaceSDK does this without requiring high quality, high resolution pictures. It can also process pictures extracted from a video. According to your security requirements, the FaceSDK can support a human operator doing the comparison, speed up the process drastically by taking over a large part of the work, or make the matching in a fully automatic way.
Verifying ID photos
Even for humans, it is often not easy to decide if a face on an ID photo belongs to the person currently at hand. The FaceSDK helps you by computing the similarity between the current picture and the ID picture in an automatic way.
Safely issuing identity cards
When issuing identity cards with portrait pictures (for example employee cards), it is not easy to guarantee the provided picture actually portrays the correct person. To ensure this, the applicant may be asked to provide, in addition to the new portrait picture, a photocopy of his or her personal ID card, driver's license, etc. With the help of the FaceSDK you can then make a plausibility check by automatically comparing the two pictures, thereby securing and speeding up the procedure.
Searching through image databases
Large image or video databases can only be sighted at a considerable expense in time or manpower. If an image of the sought person is available, the FaceSDK can help to automatically compare it with all the images in the database. The results are then sorted by degree of similarity to the query picture.
Manual or automatic?
Different application areas have different requirements concerning the reliability of the matching result. The FaceSDK can therefore be used in a number of ways:
The face comparison via the FaceSDK supports a human operator by providing an impartial "second opinion". The ultimate decision is however still always taken by the human.
The FaceSDK allows an automatic decision for all cases in which the algorithm is very confident. Only images with low confidence are provided to a human for verification. The operating point can hereby be chosen to make the system more or less strict, such that for example only 10% of all images still need to be submitted for verification by the human. A factor of 1:10 means substantial savings in time and effort.
With the hekp of the FaceSDK all images are compared independently and without human intervention. Compared to manual verification, the system can work around the clock at extremely high speeds, and without fatigue-related drops in accuracy.