US 12,271,457 B2
Method and apparatus for deep learning-based real-time on-device authentication
Myungsu Chae, Daejeon (KR)
Assigned to NOTA, INC., Daejeon (KR)
Filed by NOTA, INC., Daejeon (KR)
Filed on Jun. 17, 2022, as Appl. No. 17/843,768.
Application 17/843,768 is a continuation of application No. PCT/KR2020/001618, filed on Feb. 4, 2020.
Claims priority of application No. 10-2019-0170529 (KR), filed on Dec. 19, 2019; and application No. 10-2020-0009739 (KR), filed on Jan. 28, 2020.
Prior Publication US 2022/0318359 A1, Oct. 6, 2022
Int. Cl. G06F 21/32 (2013.01); G06N 3/04 (2023.01); G06V 10/77 (2022.01); G06V 10/82 (2022.01); G06V 40/16 (2022.01)
CPC G06F 21/32 (2013.01) [G06N 3/04 (2013.01); G06V 10/7715 (2022.01); G06V 10/82 (2022.01); G06V 40/168 (2022.01); G06V 40/172 (2022.01)] 8 Claims
OG exemplary drawing
 
1. A deep learning-based authentication method executed in a computer device,
wherein the computer device comprises at least one processor configured to execute computer-readable instructions included in a memory, and
wherein a registration step and verification step for face-based authentication are performed by using only a feature extractor of a detection model for face detection without a separate feature extractor for face classification,
wherein the deep learning-based authentication method comprises:
detecting, by the at least one processor, a location of a region of interest (ROI) occupied by a face portion in an input image by using the detection model;
extracting, by the at least one processor, a feature map from the input image by using the feature extractor of the detection model;
extracting, by the at least one processor, a fixed length feature for the face portion using the feature map and ROI pooling for the detected location of the ROI; and
classifying, by the at least one processor, a face included in the input image based on the fixed length feature,
wherein extracting the fixed length feature comprises:
performing a first ROI pooling for a first feature map output by a first layer of the feature extractor,
performing a first convolution operation on results of the first ROI pooling through a first convolution layer,
performing a second ROI pooling for a second feature map output by a second layer of the feature extractor subsequent to the first layer,
performing a second convolution operation on results of the second ROI pooling and results of the first convolution operation through a second convolution layer,
performing a third ROI pooling for a third feature map output by a third layer of the feature extractor subsequent to the second layer, and
performing a third convolution operation on results of the third ROI pooling and results of the second convolution operation through a third convolution layer.