Plant protection image non-dense pest detection method based on reinforcement learning technology
A technology of reinforcement learning and detection methods, applied in the field of plant protection image processing, can solve the problems of high resource consumption and large amount of calculation, and achieve the effects of enhancing action space, low memory, and improving accuracy
Active Publication Date: 2020-07-31
HEFEI INSTITUTES OF PHYSICAL SCIENCE - CHINESE ACAD OF SCI
View PDF3 Cites 6 Cited by
- Summary
- Abstract
- Description
- Claims
- Application Information
AI Technical Summary
Problems solved by technology
[0008] The purpose of the present invention is to solve the defects of large amount of calculation and high resource consumption in non-intensive pest detection of plant protection images in the prior art, and to provide a method based on reinforcement learning technology Non-dense pest detection method in plant protection images to solve the above problems
Method used
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View moreImage
Smart Image Click on the blue labels to locate them in the text.
Smart ImageViewing Examples
Examples
Experimental program
Comparison scheme
Effect test
Embodiment Construction
[0060] In order to have a further understanding and understanding of the structural features of the present invention and the achieved effects, the preferred embodiments and accompanying drawings are used for a detailed description, as follows:
[0061] Such as figure 1 As shown, a method for detecting non-dense pests in plant protection images based on reinforcement learning technology of the present invention comprises the following steps:
[0062] The first step is to obtain the plant protection image database: obtain the plant protection image database, and perform traditional preprocessing on the plant protection image database as required, such as scaling to a uniform size, data enhancement (rotation, translation), etc.
[0063] The second step is to extract the feature map of the plant protection image: input the plant protection image database into the convolutional neural network pre-trained by the traditional method for processing, or directly use the traditional met...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More PUM
Login to View More Abstract
The invention relates to a plant protection image non-dense pest detection method based on a reinforcement learning technology. Compared with the prior art, the defects of large calculation amount andhigh resource consumption of plant protection image non-dense pest detection are overcome. The method comprises the following steps: acquiring a plant protection image database; extracting a plant protection image feature map; constructing a pest target detection network; training the pest target detection network; acquiring a to-be-detected plant protection image and extracting a feature map; and obtaining a pest detection result. According to the method, target detection is carried out from coarse to fine by utilizing an action space of reinforcement learning and a feature space of deep learning, an optimal candidate region is searched by combining an improved reinforcement learning algorithm and a region selection network, and then the coordinates of the target candidate region are further refined by utilizing deep learning, so that the detection precision is further improved.
Description
technical field [0001] The invention relates to the technical field of plant protection image processing, in particular to a method for detecting non-intensive pests in plant protection images based on reinforcement learning technology. Background technique [0002] Target detection refers to using a rectangular frame in the image to be detected to mark out the target object in the image and classify and recognize it. It is mainly divided into two technical schools: deep learning and reinforcement learning. Among them, the popular target detection algorithms based on anchor points (deep learning) include Faster RCNN, FPN, etc., which need to pre-calculate many redundant target candidate regions, resulting in high consumption of computer resources. [0003] Due to the large amount of calculation of deep learning and the problem that current drones cannot load the weight of high-power chips, in the practical application of agricultural plant protection, drones or plant protect...
Claims
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to View More Application Information
Patent Timeline
Login to View More IPC IPC(8): G06K9/62G06K9/46G06K9/32G06N3/04G06N3/08G06T7/187
CPCG06T7/187G06N3/08G06T2207/10004G06T2207/20081G06T2207/20084G06V10/25G06V10/44G06N3/045G06F18/214
Inventor 谢成军周满张洁李瑞陈天娇陈红波胡海瀛刘海云
Owner HEFEI INSTITUTES OF PHYSICAL SCIENCE - CHINESE ACAD OF SCI