Major Vegetable Pest Counting Algorithm Based On Machine Learning

With the rapid development of the national economy and the increase in population,the demand for vegetables in the country is increasing day by day.It is very important to improve the yield of vegetables when it is difficult to increase the vegetable cultivation area in large quantities.The impact of pests is a major factor in the decline in vegetable production.Obtaining the amount of pests in farmland is an important task for the management of vegetable pests.The traditional methods of pest number detection are manual counting methods.This method has the disadvantages of low efficiency,small range,and high strength.In order to meet the requirements of modern agriculture,the use of information technology to monitor the occurrence of pests in vegetable fields has become an important research hotspot.In order to obtain the dynamic quantity information of the pests in the real vegetable fields,this paper explored a kind of vegetable pest monitoring and counting algorithm based on machine learning with the research object of Thrips,Beetle,Bemisia tabaci,and Plutella xylostella.In this paper,a counting algorithm for major vegetable pest of southern based on machine learning is proposed on the needs of vegetable pests monitoring.The algorithm includes: interest region detection algorithm based on HSV color space;pest target extraction algorithm based on edge detection;pest feature representation method based on BOF model and vegetable pest classification method based on support vector machine.Furthermore,this paper designed a vegetable pest image acquisition equipment based on traps,which can achieve self-powered and large-scale monitoring of vegetable pests.In addition,this paper has designed a vegetable pest monitoring system.It can quickly obtain the number of various vegetable pests trapped by trapping plates in the farmland.And it can provide information of the occurrence of pests in the farmland to the farmers.It has a wide range of application prospects and practical value in vegetable pest control.200 photos had collected in a vegetable field in Dongshen Farm,Guangzhou for experimentation and analysis.The experimental results show that the accuracy of the algorithm proposed in this paper is 87% for the flea beetle count,90.94% for the Thrips,90.40% for the whitefly,and 89.91% for the diamondback moth.In the personal computer environment,the average time spent on processing each image is 12.45 s,and the longest time is 27.87 seconds.The computing performance and counting accuracy of this algorithm all meet the requirements of practical applications,and it has a wide range of application prospects.