Apparatus and method with liveness consideration
A liveness test method includes detecting a face region in an input image; generating, based on the detected face region, weight map data related to a face location in the input image; generating concatenated data by concatenating the weight map data with feature data generated from an intermediate layer of a liveness test model or image data of the input image; and generating a liveness test result based on a liveness score generated by the liveness test model provided with the concatenated data.
1 . A processor-implemented method, comprising:
detecting a face region in an input image;
generating, based on the detected face region, weight map data related to a face location in the input image;
generating concatenated data by concatenating the weight map data with feature data generated from an intermediate layer of a liveness test model or with image data of the input image; and
generating a liveness test result based on a liveness score generated by the liveness test model provided with the concatenated data.
2 . The method of claim 1 , wherein
the weight map data comprises a first region corresponding to the face region in the input image and a second region corresponding to a non-face region in the input image, and
a weight of the first region and a weight of the second region are different from each other.
3 . The method of claim 1 , wherein a weight of the weight map data varies based on a distance from a center of a corresponding region of the weight map data corresponding to the face region.
4 . The method of claim 3 , wherein
the weight map data comprises a reduced region of the corresponding region, the corresponding region, and an extended region of the corresponding region, and
the reduced region, the corresponding region, and the extended region overlap to be disposed based on the center of the corresponding region.
5 . The method of claim 4 , wherein
a first weight of the reduced region is greater than a second weight of a region between the corresponding region and the reduced region, and
the second weight is greater than a third weight of a region between the extended region and the corresponding region.
6 . The method of claim 1 , wherein
the concatenated data is generated by concatenating the weight map data and the feature data,
the provision of the concatenated data includes a provision of the concatenated data to another intermediate layer of the liveness test model, and
the other intermediate layer is subsequent to the intermediate layer.
7 . The method of claim 1 , wherein
the concatenated data is generated by concatenating the weight map data and the image data of the input image, and
the provision of the concatenated data includes a provision of the concatenated data to an input layer of the liveness test model.
8 . The method of claim 1 , wherein the generating of the weight map data comprises generating the weight map data using a neural network-based weight map generation model.
9 . The method of claim 8 , wherein a weight of a first region of the weight map data corresponding to the face region is different from a weight of a second region of the weight map data corresponding to an occlusion region in the face region.
10 . The method of claim 1 , wherein the generating of the concatenated data comprises:
adjusting a size of the weight map data to correspond to a size of the feature data; and
generating the concatenated data by concatenating the feature data and the weight map data of which the size is adjusted.
11 . A non-transitory computer-readable storage medium storing instructions that, when executed by one or more processors, cause the one or more processors to perform the method of claim 1 .
12 . An apparatus comprising:
one or more processors; and
a memory comprising one or more non-transitory storage media that store instructions that, when executed by the one or more processors, causes the apparatus to:
detect a face region in an input image;
generate, based on the detected face region, weight map data related to a face location in the input image;
generate concatenated data by concatenating the weight map data with feature data generated from an intermediate layer of a liveness test model or with image data of the input image; and
determine a liveness test result based on a liveness score determined by the liveness test model provided with the concatenated data.
13 . The apparatus of claim 12 , wherein
the weight map data comprises a first region corresponding to the face region in the input image and a second region corresponding to a non-face region in the input image, and
a weight of the first region and a weight of the second region are different from each other.
14 . The apparatus of claim 12 , wherein a weight of the weight map data varies based on a distance from a center of a corresponding region of the weight map data corresponding to the face region.
15 . The apparatus of claim 12 , wherein
the concatenated data is generated by concatenating the weight map data and the feature data,
the provision of the concatenated data includes a provision of the concatenated data to another intermediate layer of the liveness test model, and
the other intermediate layer is subsequent to the intermediate layer.
16 . The apparatus of claim 12 , wherein
the concatenated data is generated by concatenating the weight map data and the image data of the input image, and
the provision of the concatenated data includes a provision of the concatenated data to an input layer of the liveness test model.
17 . The apparatus of claim 12 , wherein
the execution of the instructions further causes the apparatus to generate the weight map data using a neural network-based weight map generation model, and
a weight of a first region of the weight map data corresponding to the face region is different from a weight of a second region of the weight map data corresponding to an occlusion region in the face region.
18 . An electronic device, comprising:
a camera configured to obtain an input image;
one or more processors; and
a memory comprising one or more non-transitory storage media that store instructions that, when executed by the one or more processors, causes the electronic device to:
detect a face region and a non-face area in an input image;
generate, based on the face region and the non-face area, weight map data comprising a plurality of regions each having different weights;
generate concatenated data by concatenating one of the regions of the weight map data with feature data generated from an intermediate layer of a liveness test model; and
generate a liveness test result based on a liveness score generated by the liveness test model provided with the concatenated data,
wherein the liveness test model is a machine learning model or neural network model.
19 . The electronic device of claim 18 , wherein
the weight map data comprises a first region corresponding to the face region in the input image and a second region corresponding to a non-face region in the input image, and
a weight of the first region and a weight of the second region are different from each other.
20 . The electronic device of claim 18 , wherein
the provision of the concatenated data includes a provision of the concatenated data to another intermediate layer of the liveness test model, and
the other intermediate layer is subsequent to the intermediate layer.