Research On Identification Method And System Design Of Maturity Of Pleurotus Ostreatus Stick Based On Deep Learning

Pleurotus ostreatus is one of the main varieties in the edible fungus industry and occupies a very important position.The quality of Pleurotus ostreatus sticks during the culture period determines the output and quality of Pleurotus ostreatus mushrooms.During the cultivation of Pleurotus ostreatus sticks,the requirements for environmental parameters are different in each period.It is necessary to identify the maturity of Pleurotus ostreatus mushroom sticks in order to judge the timing of environmental regulation,the timing of bagging and fruiting,and improve the yield and quality of Pleurotus ostreatus.At present,the identification of the maturity of Pleurotus ostreatus can only be determined by professional production personnel,which is timeconsuming,labor-intensive,error-prone and omission-prone.In response to the above problems,based on the key technology development and industrialization of edible fungus intelligent industrial production,a major scientific and technological innovation project in Shandong Province,the paper analyzes the regulation method of environmental parameters during the cultivation of Pleurotus ostreatus sticks,studies Pleurotus ostreatus stick identification mode based on the improved YOLOv5 s,and designs and implement the maturity identification system of Pleurotus ostreatus sticks based on B/S architecture.The main research content and results are as follows:(1)A data set of images of oyster mushroom sticks was constructedIn this paper,the mycelium stage of Pleurotus ostreatus mushroom sticks was used as the research object.During the field investigation,through the sampling method of fixed-point shooting,the image collection of Pleurotus ostreatus mushroom sticks in the germination period,colonization expansion period,extended growth period and maturity period was completed.Firstly,the collected images were analyzed,and then the data set was divided according to the ratio of 7:1:2 by random sampling,and the data set was marked with the Label Img tool.Secondly,in order to ensure the generalization ability of the model,the data set was enhanced by random rotation,flipping,cropping,contrast transformation,adding noise and other methods.The enhanced image data totaled 1459 pieces.(2)Analysis of four recognition models for maturity of oyster mushroom sticks based on convolutional neural networkBased on the Mobile Net V3 small,YOLOv5s,YOLOv7,and YOLOv7-tiny network models,model training was carried out on the image data set of oyster mushroom sticks,and then the experimental results were verified and compared.The experimental results show that the Precision and Accuracy of the Mobile Net V3 small model are the lowest,and the Precision,Accuracy,and m AP@0.5 of the YOLOv5 s model are the highest,reaching 98.5%,99.3%,and99.3%,respectively.(3)The maturity recognition model of Pleurotus ostreatus based on the improved YOLOv5 s was constructedIn view of the YOLOv5 s model with the best overall performance,there are still problems such as model parameters,large amount of calculation,and unsatisfactory reasoning speed.The method of replacing the extracted feature network is used to improve,and the original CSPDarknet53 structure is replaced by the Mobile Net V3 small network,and the improved YOLOv5 s The model is compared with the original model.The experimental results show that compared with the original model,the improved YOLOv5 s model has a parameter quantity of3.5M,a decrease of 51.4%,and a floating-point calculation amount of 6.3GFLOPs,a decrease of 60.1% when the accuracy and average precision are basically unchanged.%,the reasoning speed is 61.8 frames per second,an increase of 12.1%,and the expected effect is achieved.(4)The maturity identification system of Pleurotus ostreatus sticks based on B/S architecture was designed and implementedBased on the needs of users,using the Spring Boot-Vue framework and My SQL database technology,a maturity identification system for oyster mushroom sticks is designed.The system includes four functional modules: process information management,cultivation shed information management,Pleurotus ostreatus stick maturity identification and bag opening information management.Users can use the system through PC or mobile terminals,realizing the safe storage and scientific management of the maturity data of Pleurotus ostreatus sticks.