2results about How to "Improve face recognition performance" patented technology

The invention discloses a face recognition device based on building construction, and particularly relates to the field of face recognition devices.The face recognition device comprises a box body, an inner cavity of the box body is divided into a water storage cavity and an equipment cavity through a partition plate, the equipment cavity is formed in the top of the water storage cavity, and a flow stabilizing plate and a filtering plate are arranged in the middle of the water storage cavity of the box body. According to the face recognition device, the rotating box can be controlled to rotate up and down and rotate horizontally by starting the first stepping motor and the second stepping motor, the shooting angle and direction of the face recognition lens can be adjusted, and the face recognition device can be conveniently adapted to workers at different positions and with different heights for face recognition; when the lens needs to be cleaned, the supporting rod can be automatically controlled to rotate, the sponge wiper wipes and cleans the lens, a worker can conveniently wipe and clean the face through the paper roll, the water pump is started to spray out liquid through the water pipe, cleaning water is conveniently provided for cleaning the face of the worker, and the face recognition effect is improved.

This application provides a method, apparatus, device, and storage medium for training a deep learning model for face recognition. The method includes: inputting a face sample image into a dual-branch basic unit; extracting a first feature map and a second feature map using a first branch and a second branch respectively; concatenating the first and second feature maps to obtain a transformed feature map; processing the feature map output from the target stage in the backbone network to obtain a feature vector for each target stage; aggregating the feature vectors to obtain a target feature vector; determining the angle between the target feature vector and the class center of each face category; determining an additive margin parameter based on the number of samples corresponding to the face category in the face sample image and the maximum number of samples in the largest category; inputting the angle and the additive margin parameter into a loss function to calculate the loss value; and training the face recognition deep learning model using the loss value. This application improves the accuracy and performance of the face recognition model.