| CPC G06V 40/172 (2022.01) [G06N 3/045 (2023.01); G06V 10/82 (2022.01); G06V 40/168 (2022.01)] | 10 Claims |
|
1. A system of mode selection face recognition with parallel CNNs, installed and operated in a calculator, comprising:
a parallel convolutional neural network module, comprising a plurality of convolutional neural network units, each convolutional neural network unit extracts a feature vector from at least one received facial image for a facial characteristic and binds the feature vector to a specific ID;
a mode selection module, receiving a plurality of facial images transmitted externally, determining the facial characteristic of each facial image, and transmitting each facial image to at least one convolutional neural network unit corresponding to the facial characteristic thereof;
a database module, storing extracted feature vectors and bound IDs according to different convolutional neural network unit; and
an ID judgment module, executing the following tasks in a testing stage:
a. receiving at least one test feature vector extracted by at least one convolutional neural network unit associated with a test facial image, and comparing the at least one test feature vector with the feature vectors extracted by the corresponding convolutional neural network unit stored in the database module;
b. selecting the bound ID of one feature vector closest to the at least one test feature vector as a voting ID; and
c. calculating a number of times being selected for each voting ID and finding the one with the largest number of times being selected,
wherein, in a learning stage, the mode selection module, for each specific person having a corresponding ID, inputs at least one facial image of the specific person to the associated at least one convolutional neural network unit according to the facial characteristic, respectively, to build the content of the database module.
|