Research On Key Technologies Of Cervical Cell Images Based On Machine Learning

With the rapid development of the Internet and people’s increasing demand for medical resources and services,the development of smart medical care that combines artificial intelligence and big data technology is in full swing.Considering the huge harm of cervical cancer to people’s health and the effectiveness of early prevention,manual processing of cervical cell sample images is time-consuming and laborious.The accuracy rate cannot be guaranteed.Therefore,the automatic processing technology of cervical cell images combined with machine learning is of great significance.In view of the fact that cervical cell images have many overlaps and adhesions in the real environment,the image scale is extremely large,and the detection accuracy can meet the diagnostic requirements when the detection accuracy reaches the regional level,combined with the needs and conditions in practical applications,this article will no longer use the more popular segmentation and The idea of combining classification is changed to a scheme of target detection model for diseased cells and its nearby area.Compared with the traditional thinking of cervical cell image processing technology,the complexity of the model and the required computing power and time consumption are greatly reduced.Greatly improve the speed and efficiency of cervical cell imageassisted diagnostic procedures.The main research work of this paper is as follows:The type and function of the convolution kernel in the basic structure of the convolutional neural network are studied,and part of the traditional convolution in the neck module path aggregation network PANet in the YOLOv4 model is replaced with the over-parameterized convolution DO-CONV,adding The parameters that can be learned by the network are further enhanced,and the feature extraction ability of the model pair and the image is further enhanced.At the same time,the calculation amount and processing time of the model are not increased in the inference stage,and the detection accuracy of the model can be effectively improved on images of different scales.The role of attention mechanism in cervical cell image processing technology was studied,and an improved model of the YOLOv4 network based on the attention mechanism was proposed,and the SE module and the CBAM module in the attention mechanism were verified by experiments for feature extraction of the backbone network.The stage enhancement m AP、effect、accuracy rate 、recall rate 、F1-score and other indicators have been improved,and the model’s ability to detect lesions in cervical cell images is enhanced.