Face Mask Wearing Detection In Agricultural Parks Based On Deep Learning

At present,my country has entered a new stage of normalization of epidemic prevention and control,and it is also a period when epidemic prevention and control is facing great challenges.During this period,my country's agricultural development was also greatly affected.As an important living guarantee base,the agricultural park is self-evident for its development and epidemic prevention work.With the development and progress of deep learning in target detection,it has become possible for deep learning algorithms to be applied to mask wearing detection.Therefore,the research on the detection of face mask wearing in agricultural parks based on deep learning has very important practical significance and is an important topic that needs urgent research.The main research work of this paper is as follows:(1)This paper proposes a face mask wearing detection model based on the Faster R-CNN algorithm based on the advantages of deep learning using deep neural networks to automatically extract target features.Data are collected from the public data sets WIDER FACE,MAFA,RMFD and the network and on-site.After data cleaning,the data set is labeled with the labeling software Label Img.The face is labeled as the face without a mask.face?mask is labeled as the face with a mask.By preprocessing the self-made data set,the main tasks are image size normalization,image denoising and color histogram equalization.In the image denoising process,four methods: median filter,mean filter,non-local mean(NL-means)and three-dimensional block matched filter(BM3D)are selected and compared,and structural similarity(SSIM)and peak signal-to-noise ratio are used.(PSNR)as an evaluation index,the results show that BM3 D has the best effect in denoising color images.(2)Comparing the face mask wearing detection model based on the Faster R-CNN algorithm with the YOLOv4 and SSD models,the results show that the Faster R-CNN effect is relatively excellent under the self-made data set.The average accuracy of face is96.62%,which is better than the other two models,which verifies the superiority and accuracy of the proposed scheme.The m AP detected by wearing the experimental mask can reach 97.06%.(3)This paper designs and implements a face mask wearing detection system in agricultural parks.Through the analysis of actual application scenarios,image collection and image collection equipment,the system mainly includes two major functions: real-time image collection and mask wearing detection modules,around these two The function completes the design and realization of the system.The target can be detected in real time through the USB camera.At the same time,the system can be deployed on an ordinary PC,and the detection function can be completed without a GPU.The test results show that the various functions and performance parameters of the system can meet the application standards.In summary,the deep learning-based detection method for face mask wearing in agricultural parks proposed in this paper establishes a scientific and effective supervision mode for mask wearing behavior,which greatly promotes the effect of urging people to wear masks,and is beneficial to Let my country's epidemic control work be implemented to ensure people's health.