Design And Implementation Of Sensitive Image Monitoring System Based On Deep Learning

With the rapid development of the Internet,wireless communication technology has moved from the 4G era to the 5G era.Compared with the past,the information on the Internet has increased exponentially in both the transmission speed and the number of transmissions,including a large number of pornographic bad pictures,which seriously affects people's physical and mental health and the healthy operation of the Internet,making the task of purifying the network environment facing huge challenges.In recent years,with the rapid development of computer hardware levels,the parallel computing capabilities of large-scale GPUs have been rapidly improved,which has made it possible to apply deep learning technology to solve computer vision problems.Using computer vision technology based on deep learning to identify whether the pictures in the network are pornographic pictures with sensitive content can greatly reduce the harmful effects of these pictures on the Internet environment and the physical and mental health of netizens.The shortcomings of the image recognition method are not high in accuracy,and the detection can be completed at a lower cost,and it has high robustness.The recognition model can be continuously updated and maintained through data.Therefore,the use of sensitive image recognition and detection methods based on deep learning is becoming more and more important for the green operation of the Internet.By studying the application of deep learning in sensitive image recognition and related algorithms,this paper proposes a coarse classification and multi-network image recognition algorithm,and applies the algorithm to the system designed and implemented.The algorithm fully absorbs the Res Net residual idea,so that the network level is deep enough without degradation problems,and because of the rough classification idea,images of different target sizes are input into different neural network models in advance,so that different networks focus on extracting and learning image feature of the same size target and improves the accuracy of recognition.The proposed new algorithm is verified on a self-built data set.Compared with a single model such as Res Net,the AP value of different target sizes is greatly improved,and the overall m AP value is also greatly improved.Compared with other sensitive image recognition algorithms,it has certain real-time performance while ensuring accuracy.Finally,using the proposed algorithm,combined with software system design and realization of relevant theoretical knowledge,a sensitive picture monitoring system based on deep learning is designed and implemented.The system recognizes the content of the image and determines whether the image contains sensitive content such as pornography.The test results show that the system has a high recognition rate for sensitive images,and the error for sensitive image recognition is also within an acceptable range,which has high practical value.