Heterogeneous face recognition model training method, face recognition method and related devices

Abstract

The invention provides a heterogeneous face recognition model training method, a face recognition method and related devices. The heterogeneous face recognition model training method comprises the steps: combining a visible light face image and a heterogeneous face image to generate a visible light-heterogeneous combined image; and training the heterogeneous face recognition model through the visible light-heterogeneous joint image, the visible light face image sample set and the heterogeneous face image sample set. According to the heterogeneous face recognition model trained through the training method, the visible light photo base library can be used for recognizing the heterogeneous face images, the heterogeneous photo base library does not need to be used for comparison, and the practicability of the heterogeneous face recognition method is improved.

Description

technical field [0001] The present application relates to the technical field of image recognition, in particular to a heterogeneous face recognition model training method, a face recognition method and related devices. Background technique [0002] At present, in the field of security, when encountering dark scenes such as night, in order to extract valuable information from the captured images, the usual practice is to set up a supplementary light next to the camera that can improve the image quality and clarity. Although this method can solve the problem of uneven illumination to a large extent, the supplementary light also brings an unfriendly experience to people facing the camera, especially for cameras on the road, and may even affect the The driver poses certain safety hazards. [0003] At present, devices with infrared functions can alleviate the influence of light and other external environments on face recognition algorithms to a certain extent, but the current m...

AI Technical Summary

Problems solved by technology

[0002] At present, in the field of security, when encountering dark scenes such as night, in order to extract valuable information from the captured images, the usual practice is to set up a supplementary light next to the camera that can improve the image quality and clarity. Although this method can solve the problem of uneven illumination to a large extent, the supplementary light also brings an unfriendly experience to people facing the camera, especially for cameras on the road, which may even affect the Driver poses certain safety hazards

[0003] At present, devices with infrared functions can alleviate the impact of light and other external environments on face recognition algorithms to a certain extent, but the current mainstream solution in the industry is to collect the target personnel's infrared ID photo database for later comparison, but The cost and practicability of this method are not as good as those based on ordinary ID photos as the base library. In addition, it is unrealistic to obtain a large-scale infrared base library in order to be applied in large-scale scenarios.

Similar to identifying infrared images, the method of identifying heterogeneous images through the constructed sketch image base library and other heterogeneous image base libraries is not very practical

Images