Face Recognition: Large-Scale Datasets For Face Anti-spoofing
As face recognition technology matures and its commercial application becomes more and more extensive, however, faces can be easily copied by means of photos and videos. Therefore, the counterfeiting of legitimate user faces is an important threat to the security of face recognition and authentication systems. At present, some progress has been made in liveness detection methods based on dynamic video face detection, face blinking, thermal infrared and visible light face correlation, etc.
● 2D Face Anti-spoofing
The commonly used living detection technology of face recognition technology generally adopts the method of command action coordination, such as face turn left, right turn, mouth opening, blinking, etc., and the wrong command matching is considered to be forged and deceived. Whether it is a real person or a photo taken by a camera, the final result is a two-dimensional picture, so it is difficult to judge whether the current face recognition technology is a real person or a photo in front of the camera. In addition, the recognition of groups such as twins and plastic surgery by face recognition also needs to be further studied.
● 3D Face Anti-spoofing
The 3D camera is used to shoot the face, and the 3D data of the corresponding face area is obtained, and further analysis is performed based on these data, and finally it is judged whether the face is from a living body or a non-living body. The sources of non-living bodies here are relatively wide, including photos and videos of various media, photos of various printed materials of different materials, 3D head models, etc.
Based on the 3D face data of living and non-living bodies, the most discriminative features are selected to train the classifier, and the trained classifier is used to distinguish between living and non-living bodies. The selection of features is crucial. The features we choose here contain both global information and local information. Such features are beneficial to the stability and robustness of the algorithm.
● Infrared Face Anti-spoofing
Near-infrared face liveness detection requires no command cooperation, and the detection success rate is high. According to the optical flow method, the temporal change and correlation of the pixel intensity data in the image sequence are used to determine the “motion” of the respective pixel positions, and the running information of each pixel point is obtained from the image sequence. and support vector machine for statistical analysis of data. At the same time, the optical flow field is more sensitive to the movement of the object, and the eye movement and blinking can be detected uniformly by using the optical flow field. This liveness detection method can realize blind testing without the user’s cooperation.
As the world’s leading AI data service provider, Datatang has accumulated around 350K images of 2D/3D face anti-spoofing and infrared face data. All data is collected with proper authorization with the person being collected, and customers can use it with confidence.
The collection scenes include indoor and outdoor scenes. The data includes male and female. The age distribution ranges from juvenile to the elderly, the young people and the middle aged are the majorities. The data includes multiple postures, multiple expressions, and multiple anti-spoofing samples.
The collection scenes include indoor and outdoor scenes. The dataset includes males and females. The age distribution ranges from juvenile to the elderly, the young people and the middle aged are the majorities. The device includes iPhone X, iPhone XR. The data diversity includes various expressions, facial postures, anti-spoofing samples, multiple light conditions, multiple scenes.
The 3D living detection data uses the front 3D structured light lens module of the iPhone to realize the collection of 3D face and fake face samples. It covers most of the data forms required in the live detection algorithm, in addition to the original real face action, mobile phone, face action adversarial sample-pad remake, photo adversarial sample-face photo and mask deception, it also includes 3D masks or mannequin head template. In terms of material for 3D masks or mannequin head, we selected materials such as sandstone and resin, which greatly improves the sample distribution richness of masks or mannequin head.
The collecting scenes of this dataset include indoor scenes and outdoor scenes. The data includes male and female. The age distribution ranges from child to the elderly, the young people and the middle aged are the majorities. The collecting device is realsense D453i. The data diversity includes multiple age periods, multiple facial postures, multiple scenes.
If you want to know more details about the datasets or how to acquire, please feel free to contact us: firstname.lastname@example.org.