Mask wearing recognition method

Abstract

The invention discloses a mask wearing recognition method, and the method comprises the steps of carrying out the face detection model training and mask wearing recognition model training, and the face detection model training process comprises the steps: building a face detection network based on an MTCNN, obtaining a face detection model with higher robustness through training, predicting a facearea and a face key point position in the spectral image, and reducing the detection range of the mask wearing model. The mask recognition model training process comprises the steps that a sample ROIis marked on a face spectral image, sample coordinate information and category information are acquired through a sample ROI file, and a data set is constructed; model parameters are obtained by training an SVM classifier and used for recognizing the mask in the face image. According to the method, automatic non-contact mask wearing detection can be achieved, the face positioning speed is high, the detection speed is high, the situation that the mouth and the nose are shielded by clothes and the situation that the mask is correctly worn can be accurately distinguished, false detection causedby clothes shielding is greatly reduced, and the accuracy is high.

Description

technical field [0001] The invention relates to a mask wearing recognition method, which belongs to the field of target detection. Background technique [0002] In crowded places such as hospitals and public transportation, staff need to check whether pedestrians wear masks one by one, and the workload is quite heavy. Existing mask detection methods generally cannot distinguish whether the mask is worn correctly or is covered by hands or clothing. Therefore, a new method is needed to accurately distinguish between wearing a mask correctly and being blocked. Contents of the invention [0003] The purpose of the present invention is to provide a mask wearing identification method, improve the sorting speed and sorting accuracy of various and irregular complex parts on the production line, improve the sorting efficiency, and thereby increase productivity. [0004] A mask wearing identification method, the method comprises the following steps: [0005] S1. Carry out face det...

AI Technical Summary

Problems solved by technology

Existing mask detection methods are generally unable to distinguish whether the mask is worn correctly or is covered by hands or clothing, so a new method is needed to accurately distinguish between wearing a mask correctly and being blocked

Images