The operator loads the image and gets a compliant photograph. When something is wrong, the program says so in plain language — about the input image and about the output — instead of simply rejecting it.
At the counter the variable is not the technology, it is the person using it. Enroll combines a face detection engine and an ICAO verification SDK to grant a constant output quality regardless of who captures the photograph, and to give the operator simple, clear error messages.
Images are rotated, cut and cropped based on face size and eye position. No technical training is required: load the photograph, get the result.
The same capture can produce several outputs using different algorithms, because a photograph destined for a database is not the same one a laser engraver needs:
Colour and black-and-white algorithms are meant to produce different results, so both are worth testing.
For each subject you see the triptych — original, colour primary image, black-and-white secondary — plus the table of ICAO checks before and after processing. You can read exactly which requirements were out of specification and how they came back in.
Original, colour primary and black-and-white secondary side by side: the difference is visible at a glance, no explanation needed.
Every requirement with its value before and after. It is the most direct way to understand what normalisation actually does.
One of the subjects wears glasses and a beard: precisely the conditions where geometric and lighting checks are most delicate.
The full list of supported capture devices, system requirements and the details of the client/server configuration are under review.
Finds the face and produces the compliant crop, even from a full-length shot.
Pure normalisation for already framed images, batch capable.
SDK to embed normalisation into an existing application.
ICAO compliance verification, requirement by requirement. Real online demo.
For an assessment against your own workflow, or a demonstration licence.
