ass日本风韵熟妇pics男人扒开女人屁屁桶到爽|扒开胸露出奶头亲吻视频|邻居少妇的诱惑|人人妻在线播放|日日摸夜夜摸狠狠摸婷婷|制服 丝袜 人妻|激情熟妇中文字幕|看黄色欧美特一级|日本av人妻系列|高潮对白av,丰满岳妇乱熟妇之荡,日本丰满熟妇乱又伦,日韩欧美一区二区三区在线

In order to solve the problems of low accuracy, slow processing speed, easy to be disturbed by the background environment and difficult to detect target diseases of the existing wheat disease detection algorithms, a wheat disease detection system based on cloud architecture was designed by combining advanced smart phone hardware, convenient WeChat mini program application and efficient cloud service platform. The system mainly included cloud server module and WeChat mini program module. The cloud server side was mainly used for image receiving and model processing. Using CSS and Java Script language to develop WeChat mini program for data upload, information feedback and information display. In order to ensure the feasibility of the model deployment in cloud server, an improved wheat disease detection model based on YOLO v8n（C2f- Faster-Slim-Neck-YOLO v8n, CS-YOLO）was proposed. Combining with FasterNet ’s advantages of lightweight, this model proposed to replace C2f Bottleneck module with FasterNet Block, which reduced the model size and improved the model ’s feature fusion ability and detection accuracy. In the Neck network, GSConv and VoV-GSCSP module in Slim-Neck design paradigm were used to improve the neck of YOLO v8n, reducing the calculation amount of the model and improving the detection accuracy of the model. The test results showed that for the wheat disease data set collected in the field environment, the floating point computation and model memory occupation of the improved model were reduced by 24.4% and 17.5% respectively compared with the baseline model of YOLO v8n, and the average accuracy was increased by 1.2 percentage points compared with the original model. It was superior to YOLO v3-tiny, YOLO v5, YOLO v6, YOLO v7, and YOLO v7-tiny algorithms. Finally, the lightweight detection model CS-YOLO was deployed on the cloud server and the detection function was transformed into an API interface. The applet called the server connection by requesting its interface. After receiving the request, the server passed the data to the model deployed on the cloud server. By using the WeChat mini program to invoke the detection model for disease image type recognition and disease location detection, the mean average precision was 89.2%, which can provide technical support for wheat disease type recognition and disease location detection.