This project has been developed for the Multimodal Interaction course at Sapienza University of Rome (2021-22).
Project report | Project presentation
By Davide Quaranta and Federico Cernera.
LittleAntispoof is a multimodal face liveness detection module that can be used in the context of face anti-spoofing.
The system uses a challenge-response interaction style, composed of gaze verification, emotion verification, speech verification, and eye blinking check.
Specifically, recognition attempts that must be checked for liveness, will pass though the following phases, in a random order:
- Gaze challenge: the user is asked to look towards a certain direction, randomly chosen.
- Emotion challenge: the user is asked to provide an emotion-distorted face expression (e.g. happy face), randomly chosen.
- Speech challenge: the user is asked to pronounce a randomly chosen phrase.
During challenges, the user's eye blinking rate is also checked, in order to determine whether it is within a configured one.