For small farmers’ plant protection operation scenarios, this paper constructed an intelligent monitoring and warning system that integrated data collection, pest and disease identification, risk warning and instruction push. The system jointly processes multi-source data such as field images, insect trapping, meteorological parameters and crop growth status. The lightweight visual recognition network and time series risk discrimination model are used to complete the disease and insect pest representation learning and early warning level generation, and the end-to-end collaborative mechanism is used to realize end-side collection and cache, edge screening and identification, and cloud strategy update. The experiment is carried out based on 4260 groups of samples. In the recognition tasks of eight types of common pests and diseases, the accuracy of the system reaches 94.6%, the F1 value is 93.8%, the early warning consensus rate is 91.7%, and the response delay is controlled at 1.9 s. The application results show that the system improves the inspection coverage of small farmers by 28.4%, reduces the response time of treatment by 31.6%, and improves the control efficiency per unit area by 24.9%. Further combined with the calculation of pesticide input, labor cost, production loss and average income per mu, the average net income per mu in a single season increased by 186.3 yuan, and the input-output ratio increased by 19.7%, indicating that the system has deployment value and economic benefits under the condition of decentralized operation.
Povzetek: Ta članek za male kmete vzpostavlja inteligentni sistem za spremljanje in zgodnje opozarjanje na rastlinske bolezni in škodljivce, ki združuje poljske slike, podatke iz vab za insekte in meteorološke podatke za prepoznavanje ter opozarjanje. Poskus je temeljil na 4260 vzorcih. Natančnost prepoznavanja je dosegla 94,6 %, stopnja skladnosti opozoril 91,7 %, odzivni čas pa 1,9 s. Povprečni neto dobiček na mu v eni sezoni se je povečal za 186,3 juana. Rezultati kažejo, da ima sistem praktično uporabno vrednost pri izboljšanju učinkovitosti zatiranja ter povečanju pridelovalnih prihodkov.
