Kvíz spojený s průvodcem · Střední úroveň
Face Liveness Detection and Anti-Spoofing Quiz
Check how face presentation attack detection differs from identity matching, what test results mean, and why capture integrity and fallback matter.
Otázka 1 z 8
Nástroj pro vyhledávání obličejů najde blízkou shodu se zaregistrovaným obrázkem. Jakou další otázku PAD řeší?
Náhled každé otázky8 otázky
Odpovědi a vysvětlení se zobrazují, jak budete v kvízu.
- Nástroj pro vyhledávání obličejů najde blízkou shodu se zaregistrovaným obrázkem. Jakou další otázku PAD řeší?
- Whether face quality is high enough for matching
- Whether the capture appears to be a bona fide presentation rather than an attack
- Whether the face matches an enrolled template
- Whether the identity document belongs to the presenter
- Proč samotná kontrola živosti nemůže prokázat nárokovanou totožnost?
- PAD has already verified the enrollment record
- Passing PAD confirms ownership of the account
- A liveness score verifies a government document
- A live-looking presenter could still be a different person
- Detektor zachytí vytištěné fotografie v laboratoři. Které tvrzení je oprávněné?
- It performed on the tested photo attacks under those test conditions
- It will also detect every realistic mask
- It proves the same error rate on every camera
- It establishes resistance to virtual-camera injection
- How does a stricter PAD threshold typically affect the user journey?
- It reduces both attack accepts and genuine rejects without new data
- It changes identity-match accuracy but leaves PAD errors fixed
- It may reject more attacks while also rejecting more genuine users
- It removes the need to monitor bona fide rejection
- Proč by měl tým vzdáleného onboardingu testovat vstřikování virtuální kamery samostatně?
- It is merely a higher-resolution physical photo
- It can substitute media after the lens and bypass checks aimed at physical presentations
- It changes only the stored enrollment portrait
- A printed-photo test necessarily covers injected video
- Který test rozdělil nejlepší sondy zobecnění PAD detektoru?
- Hold out subjects, sessions and attack instruments where feasible
- Split adjacent frames from each video between train and test
- Reuse the same subjects and sessions but change the file names
- Set the PAD threshold on the final held-out attack set
- A genuine user repeatedly fails because of lighting. Which design response follows the guide?
- Lower the threshold until every attempt passes
- Repeat the same capture indefinitely without escalation
- Treat every failure as confirmed malicious behavior
- Offer an accessible fallback or step-up review
- Which statement stays within ISO/IEC 30107-1’s scope?
- It requires one particular camera or infrared sensor
- It certifies a complete identity-proofing system
- It defines PAD concepts without specifying one universal detection method
- It guarantees detection of attacks not represented in a test