Snail pest control method based on visual target detection in dendrobium officinale cultivation

Abstract

The invention relates to a snail pest control method based on visual target detection in dendrobium officinale cultivation. The snail pest control method comprises the following steps of 1) collecting the pictures and the video data with snails; 2) preprocessing the data; 3) training a snail detection network; 4) monitoring the snails real-timely. The beneficial effects of the invention are as follows by processing the image and the video processed according to the characteristics of the agricultural environment by using a convolutional neural network and a sliding window technology, selecting a target, and labeling, the detection accuracy and efficiency of the method are obviously better than those of the existing method; meanwhile, due to the fact that the end-to-end convolutional neural network is used, the real-time processing speed is high, the requirement for real-time monitoring in the agricultural environment can be met, the advantages of being convenient to achieve and low in hardware equipment requirement are achieved, and the insect pest detection effect, especially the snail detection effect, in the intelligent agriculture can be effectively improved.

Description

technical field [0001] The invention relates to the field of agricultural pest control, in particular to a method for detecting and preventing agricultural pests based on convolutional neural network and visual target detection technology in the cultivation of Dendrobium officinale. Background technique [0002] At present, artificial intelligence technology has been widely used in agricultural production, especially in smart agriculture, such as seed detection, crop monitoring, automatic soil irrigation, and crop environment control according to weather. Pest detection is an important branch of crop monitoring. At present, pest control in the agricultural field is mostly done manually by growers, usually using drugs for control or manual monitoring for a long time. In Dendrobium officinale, snails are a very harmful pest, which will continue to harm young stems, leaves, flower buds and young fruits. It has the characteristics of small targets, various postures, and slow mo...

AI Technical Summary

Problems solved by technology

The above method analyzes that chemical-based pest control will lead to different pesticide residues. At the same time, computer technology is also used to control pests, but special equipment such as insect traps and hyperspectral imaging instruments are needed, while ordinary video is used to control pests. When detecting, when using the classification-based target recognition method to detect specific features, the extraction of candidate frames and the classification of candidate frames need to be trained separately, and the detection also needs to go through the candidate frame extraction network and the candidate frame classification network in order , so the timeliness is poor in actual use, and due to the characteristics of the method, there is a certain false detection rate for the background

Images