Access control with face recognition and heterogeneous information
The use of multimodal face attributes in facial recognition systems is described. In addition, use of one or more auxiliary attributes, such as a temporal attribute, can be used in combination with visual information to improve the face identification performance of a facial recognition system. In some examples, the use of multimodal face attributes in facial recognition systems can be combined with the use of one or more auxiliary attributes, such as a temporal attribute. Each of these techniques can improve the verification performance of the facial recognition system.
1 . A computer-implemented method of using a facial recognition system to identify a person, the method comprising:
extracting an attribute of the person by applying a first representation of a first image of the person to a previously trained attribute classifier machine learning model to generate an attribute classifier output;
applying the attribute classifier output and a distance measurement output generated using a second representation of a second image of the person to a previously trained fusion verification machine learning model;
generating a facial recognition system output using the previously trained fusion verification machine learning model;
applying the facial recognition system output to a joint classification model;
applying an auxiliary attribute to the joint classification model;
generating a joint classification output using the joint classification model; and
controlling access to a secure asset using the joint classification output.
2 . The method of claim 1 , wherein the attribute is a first attribute, wherein the previously trained attribute classifier machine learning model is a previously trained first attribute classifier machine learning model, and wherein the attribute classifier output is a first attribute classifier output, the method comprising:
extracting a second attribute by applying the first representation of the first image to a previously trained second attribute classifier machine learning model to generate a second attribute classifier output; and
applying the second attribute classifier output to the previously trained fusion verification machine learning model.
3 . The method of claim 1 , wherein the attribute includes at least one of age, gender, ethnicity, or head angle.
4 . The method of claim 1 , wherein the first representation of the image includes at least one vector.
5 . The method of claim 1 , wherein the distance measurement output is generated by applying a first distance measurement to a distance measurement classifier, the method comprising:
performing a face embedding;
applying the face embedding to a classification pipeline to identify a similar image;
generating a second distance measurement from the identified image; and
applying the second distance measurement to the distance measurement classifier to generate the distance measurement output.
6 . The method of claim 1 , wherein the auxiliary attribute includes a temporal attribute.
7 . The method of claim 1 , wherein the auxiliary attribute includes social pooling of the person.
8 . The method of claim 1 , wherein the auxiliary attribute includes a height of the person.
9 . The method of claim 1 , wherein the first representation of the image is the same as the second representation of the image.
10 . The method of claim 1 , wherein the first image is the same as the second image.
11 . The method of claim 1 , wherein controlling access to a secure asset using the joint classification output includes controlling access to a secured entrance to a building.
12 . A computer-implemented method of using a facial recognition system to identify a person from an image of the person, the method comprising:
performing a face embedding;
applying the face embedding to a classification pipeline to identify a similar image;
generating a classification pipeline output based on the identified image;
applying the classification pipeline output to a joint classification model;
applying an auxiliary attribute to the joint classification model;
generating a joint classification output using the joint classification model; and
controlling access to a secure asset using the joint classification output.
13 . The method of claim 12 , wherein the classification pipeline output includes a distance measurement.
14 . The method of claim 12 , wherein the auxiliary attribute includes a temporal attribute.
15 . The method of claim 12 , wherein the auxiliary attribute includes social pooling of the person.
16 . The method of claim 12 , wherein the auxiliary attribute includes a height of the person.
17 . The method of claim 12 , wherein the joint classification model includes a previously trained machine learning model.
18 . The method of claim 12 , wherein generating the joint classification output includes:
generating a joint probability using the joint classification model.
19 . The method of claim 12 , wherein controlling access to a secure asset using the facial recognition system output includes controlling access to a secured entrance to a building.
20 . The method of claim 1 , wherein the fusion verification machine learning model fuses the attribute classifier output and the distance measurement output to obtain a fused feature representation.