Face Recognition and Fake Face Detection

We have an attendance application based on mobile apps with facial recognition mechanism. Due to server capacity reasons, We don't enable facial recognition function, but when attendance process occurs, We saved image for each timesheet process. Face verification process is carried out in the back process or pending verification. In a month we have more than 300,000 face data. But during the verification process after we checked manually, the library we used could not display data with high accuracy. Fake methods are widely used, such as using photos and videos for the attendance process. Then there are also the challenge that the photos from the attendance process are not all of good quality, because the mobile devices used for attendance are very diverse. Maybe FaceMe can help us with this problem? Thank you