Abstract: Systems and methods for detecting a presentation attack. Datasets of original images and attack samples are used to train a model that creates in a feature space of samples of original images a set of elements corresponding to each sample. Elements form a limited set normalized to a disjoint set so that the boundary of the disjoint set divides the space of all possible collections of features into “legitimate” and “all others.” “All others” include signs of known attacks, anomalies and signs of unknown attacks.

Abstract: Systems and methods for performing liveness detection. A method includes transmitting an image or a video by a threat detector to one or more filters for a first level check for detecting a threat based on analysis of certain characteristics, transmitting the image or the video to one or more special models for a second level check, wherein the one or more special models are configured for detecting multiple types of threats on the image or the video, transmitting the image or the video to a general model ensemble for a third level check to classify the image or the video according to individual features into original and fake, and in response to detecting the threat, registering the threat by the threat detector.

Abstract: Systems and methods for detecting liveness of imagery data. The imagery data, i.e., an image, is captured by an imagery application installed on a smartphone. The image is encoded and transferred to the liveness detector. At the liveness detector, the image is decoded and split into a set of patches by an image processor. Each of the sets of patches is analyzed. A liveness detector analyzes each patch of the set individually to determine if the patch is live or fake. By aggregating analysis results of all patches, liveness of the image is detected.

Abstract: Systems and methods for detecting a presentation attack. Datasets of original images and attack samples are used to train a model that creates in a feature space of samples of original images a set of elements corresponding to each sample. Elements form a limited set normalized to a disjoint set so that the boundary of the disjoint set divides the space of all possible collections of features into “legitimate” and “all others.” “All others” include signs of known attacks, anomalies and signs of unknown attacks.

Abstract: Systems and methods for patching a general machine learning model in a system for detecting presentation attacks. The machine learning model includes a false positive or a false negative misclassification. A patch model is trained using samples misclassified by the general machine learning model to correct the misclassification by the general machine learning model once deployed in the system pipeline.

Abstract: Systems and methods for performing liveness detection. A method includes transmitting an image or a video by a threat detector to one or more filters for a first level check for detecting a threat based on analysis of certain characteristics, transmitting the image or the video to one or more special models for a second level check, wherein the one or more special models are configured for detecting multiple types of threats on the image or the video, transmitting the image or the video to a general model ensemble for a third level check to classify the image or the video according to individual features into original and fake, and in response to detecting the threat, registering the threat by the threat detector.

Abstract: Systems and methods for detecting liveness of imagery data. The imagery data, i.e., an image, is captured by an imagery application installed on a smartphone. The image is encoded and transferred to the liveness detector. At the liveness detector, the image is decoded and split into a set of patches by an image processor. Each of the sets of patches is analyzed. A liveness detector analyzes each patch of the set individually to determine if the patch is live or fake. By aggregating analysis results of all patches, liveness of the image is detected.