Facial recognition or Face Recognition compares or identifies people with face characteristics. Facial Liveness detection or Face Liveness Detection evaluates genuine presence signs during capture. Both are related to biometrics but perform different functions as an identity check.
In JAAK, facial comparison and liveness detection are part of an identity check solution. His combination helps to address two questions: if his face matches his reference and if his capture shows signs of genuine interaction.
These differences matter while designing a digital process. A company can get a photo, find a facial coincidence and still require evidence about the capture or document submitted. It can also get a favorable liveness detection without determining who that face belongs to.
For a company that receives credit requests, opens accounts or incorporates remote customers, understanding that difference allows us to define what evidence you require before continuing an operation.
Biometrics: a category that includes different forms of recognition
The biometrics includes automated recognition based on biological or behavioral characteristics. It can use face, prints, iris or certain behavioral patterns. NIST SP 800-63B-4 guide.
That's why biometrics and facial recognition aren't interchangeable terms: facial recognition is an application of biometrics.
When a supplier offers "biometric monitoring," the first question should be what feature it uses and for what task. It should also be requested to distinguish comparison, enrollment and monitoring about capture.
The general label does not describe, by itself, the whole process and its cover with respect to suplanting attempts.
What's Face Recognition for JAAK?
In a digital high process, face comparison allows to evaluate the link between the captured face and a reference. documentation JAAK describe the comparison of two facial images. In this article we use Face Recognition as a general term and facial comparison 1:1 for concrete operation.
Fold face with the information submitted
A hypothetical example is a credit request: the company needs to evaluate if the face caught corresponds to the photo of the identification. The comparison provides evidence about that link.
For operations such as this, the result should be interpreted with document review and capture. Facial similarity alone does not prove that all information submitted is true.
Review the quality and result of comparison
The documentation of JAAK It also describes information about similarity and quality of images. That allows to distinguish the result of comparison from problems affecting capture.
We recommend defining how each case is dealt with: continuing, applying for another capture or leading to a review under a policy. A insufficient result should not be transformed automatically into an allegation of fraud.
Differentiate 1:1 comparison from a 1: N search
Comparison 1:1 contrasts with two samples. A search 1: N looks for candidates among several references. In the case of selfie versus document, the relevant explanation is the first one. No need to assign a search at bases to that flow.
To place a face in an image is also different from comparing it. A box on a face alone does not prove that an identity has been established.
What's Face Liveness Detection at JAAK?
The liveness detection, also called liveness detection, looks for signs that a sample comes from a living person at the capture point. NIST glossary about detection of presentation attacks puts them as a subset of these methods.
Detection of presentation attacks, or PAD, addresses attempts to mislead the sensor, such as presenting photographs, reproductions or masks. PAD they are related but should not be treated as absolute equivalents without revising the method and scope.
The documentation Liveness of JAAK describes the analysis of face videos to evaluate presence and suplanting signs. It also includes selection of a video box for later analysis, including facial comparison.
In a remote request, this feature provides evidence about the capture and its result should be related to facial recognition and document as necessary. A person may be present and still have an identification that doesn't belong to them.
For the business, our recommendation is to design care for inconclusive catches and unfavourable results, along with the criteria of comparison. Thus, the liveness detection has a defined function within the process.
What does passive life prove?
The information and information and information JAAK with a passive modality. In terms of interaction, this means that an analysis does not require the challenges of movement proper to an active modality. The user does have to follow the direction of positioning and capture.
When designing the experience, we recommend that we explain clearly what a person should do and what will occur if his or her capture isn't complete. The passive modality should not be interpreted as an absence of quality requirements or as universal cover of attacks.
Key differences in a table
| Concepto | What it represents or evaluates | What doesn't come to an end by itself |
|---|---|---|
| Biometrics | Category of recognition by conduct or physical characteristics | What controls includes a specific implementation |
| Face Recognition / reconocimiento facial | Comparison or identification by face characteristics | Documentary authenticity or absence of handling throughout the session |
| Facial Comparison 1:1 | Relationship between a sample and a given reference | That all his stated identity was established |
| Facial identification 1: N | Search for candidates between several references | That a candidate is a confirmed identity without applying the appropriate criteria |
| Face Livenness Detection / Facial Liveness detection | Signs of genuine presence during capture | Who's the person or that all possible attacks are covered |
That distinction helps avoid a common expectation: that a single beneficial result will answer all identity questions.
How to complement facial recognition and liveness detection JAAK
The guide KYC track API of JAAK describes biometric document and data capture and service co-ordination within a session. On that basis, let's imagine a remote credit request. The following example is illustrative and the order and components depend on the configuration.
- Documentary evidence. The company receives an identification and applying the established checks for its case.
- Facial capture. He gets an applicant sample under his condition.
- Presence. Expand genuine capture signalling with select monitoring.
- Comparison. It contrasts the sample with the corresponding reference.
- Decision. Reacts the set of results and addresses inconsistencies or exceptions.
For product and operations equipment, our recommendation is to document what question each step answers, what evidence leaves and what happens when the result does not allow to continue. For technology, it's important to maintain related results from the same session.
A good liveness detection with an unfavourable comparison, for instance, should not be automatically interpreted as a confirmed identity.
What's with deefakes and injection
A deepfake is a content generated or modified with AI to represent an appearance or performance. An injection attack introduces material into the digital flow, rather than simply presenting them to the camera. They can be combined but describe different aspects.
The Section 3.14 of NIST SP 800-63A-4 distinguishes these risks and warns that a biometric comparison does not prevent injection. It also indicates that presentation mechanisms are not sufficient to cover all of its cases.
That's why "includes liveness detection" should not be interpreted as an assertion of universal protection from deepfakes or flux manipulation.
To evaluate that protection, ask about concrete threats, tests and capture controls. NIST is used here as a technical reference, without giving them general obligation in Mexico.
How to interpret mistakes, scores and results
Facial Comparison can associate images from different people or do not associate images from the same person. NIST document these differences, together with the influence of image quality, in its demographic impact assessment.
In an implementation, she requested that the supplier explain what each score means and how to determine the decision. A score should not be presented as a percentage of certainty about the whole identity, unless there was a definition and evidence that supported that specific interpretation.
Also, a failure to capture or an unfinished result does not necessarily prove an attempt at fraud. We recommend to distinguish technical problems, insufficient evidence and sign of attack to decide appropriate care.
What a company should ask before applying
| Pregunta | What's his use? |
|---|---|
| What biometric modality and what operation do you use? | Distinguish comparison 1:1 search 1: N |
| What does the liveness detection evaluate? | Understanding its scope and its constraints |
| What attacks and conditions have been tested? | Interpreting reports and evaluations with context |
| How are capture flow protected and verified? | Assess threats that are not limited to the sensor |
| How are reattempts and exceptions met? | To avoid an end to an unjustified decision |
| What data are conserved and who can be accessed? | Review privacy, traceability and responsibilities |
In privacy, we recommend defining aims, accessibility, maintenance and disposal of data. Legal review should care for concrete treatment and industry. That article does not set specific Mexican law requirements.
Identification with JAAK: from result to decision
The aim of combining facial comparison, life and documentary evidence is to gather useful information to decide about an application. JAAK connect these components with a digital identity process and have to be defined for each implementation.
For an enterprise, we recommend evaluating three points:
- What you need to check: the link between the applicant and the reference, capture and appropriate evidence.
- How you will use the results: criteria for continuing, retrying or revising an application.
- How you will care for users: clear instructions, assistance and treatment of exceptions.
That approach allows to discuss concrete needs: from integrating customers to revising an increased risk interplay. The decision should be responsive to the process, available evidence and applicable framework.
Preguntas frecuentes
Are biometrics and facial recognition the same?
The biometrics are the general category. Facial recognition is one of its face-based applications.
A selfie's a liveness detection?
Not by capturing a photo. It's necessary to know what analysis and what signs are applied.
A good liveness detection confirms who I am?
He's evaluating his presence as his method and his link with an identity requires additional evidence.
What's the difference between Face Recognition and Face Liveness Detection?
Face Recognition compares or identifies with face. Face Liveness Detection analyzes signs of presence in capture. In an identity process they complement them: none of them substitutes for another's function.
What do you use JAAK To compare face with an ID?
Your documentation describes One-to-One, a facial comparison between two images. The use within a session depends on the set flow.
an assessment of PAD Proves protection against any beefake?
It should not be assumed. Content, attack mechanism, version and conditions effectively evaluated have to be reviewed.
Facial recognition and liveness detection for your identity process
Before applying, it distinguishes the task of comparison, the assessment of presence and the testing of evidence. It defines how you will interpret its results and how you will care for exceptions.
Do your company have to link an applicant's face with his identification and evaluate his capture? an identity check solution of JAAK and check what combination of facial comparison, life and documentary validation corresponds to your process.