Marwan Mohamed, Mohamed Yasser, Mohamed Gamal, Abdelrahman Ashraf, Dr. Fatma Helmy, Eng Mahmoud ELSahhar
27th October 2020
Face liveness detection is an essential task to detect whether the face in front of the camera belongs to a live person or just an input image. This process can be called face spoofing. The importance of detecting the liveness is to prevent that kind of attack (face spoofing). All companies that use a security system for entrance gates will use the proposed project to prevent illegal access to their systems. However, illegal access is not only gained by an image but also by videos and 3d masks. The challenge in the proposed project is to handle the various inputs. Traditional and nontraditional machine learning techniques will be used in face liveness detection. Capsule deep learning technique will be proposed to detect its validity in enhancing CNN deep learning technique.
Face recognition is a biometric system that is working on taking features from someone’s face then it compares these features with the data of people in a database of known faces. It spreads very rapidly in the last decade. It has been used in many fields such as payments, companies used face recognition in attendance systems, also used in forensics, access, and security fields. Face liveness detection is one of many methods that are used to prevent face spoofing attacks. It is relatively new to sense face liveness since the common security methods are fingerprints and passwords. However, many companies are in bad need to detect face spoofing in order to prevent any illegal access to their systems. Cheaters can gain illegal access by showing an image (or a video or 3D-mask) of an authorized person standing in front of the security camera. The role of the security system is to offer access to the faces that belong to living persons. Therefore, detecting face liveness will play an important role in preventing face spoofing attacks.
Face recognition systems is famous due to its accessibility and simplicity compared to the other biometrics like fingerprint and palm, and all the world depend on it as they use it in many of their personal applications such as personal computers, laptops and mobile phones, and commercial applications such as ATMs, airports, border control and online banking, etc. So, the proposed project will help in making the face recognition system more secure and dependable so that Companies that work in payment fields can benefit from our project by verifying and checking whether the person himself is doing the transaction.
1.3 Problem Statement
The problem that this project will focus on is liveness detection to prevent face spoofing attacks. The attacker may use one of the face spoofing attacks like (images, live videos, or 3D-masks) in front of the camera to gain unauthorized access. The proposed system aims to detect input liveness. Deep learning classifiers will be trained and tested to work on large datasets and distinguish between live faces and non-live faces.