Visual identification detection system based on rice seedlings

Abstract

The invention discloses a visual identification detection system based on rice seedlings, relates to the field of visual identification detection of rice seedlings, and aims to overcome the defects of a rice transplanter operation detection method, assist a driver in mastering rice seedling leakage information, reduce manual monitoring on a rice transplanter and the like by a rice seedling transplanting and seedling leakage detection product. An MCU control board is arranged on the surface of the computing mainboard, a power switch and a function switch are embedded into the left end of the computing mainboard, the MCU control board is connected with an RDK X3 through a UART asynchronous transceiving transmitter, the MCU control board is connected with a GPIO small lamp board, an RTK board is arranged on the surface of the computing mainboard, and the RDK X3 is connected with a 4G module through a USB. The seedling leakage monitoring system of the rice transplanter can improve the agricultural production efficiency, reduce the agricultural production cost, improve the agricultural product quality and promote the agricultural intelligent development, and is of great significance to agricultural modernization.

Description

Technical Field

[0001] The invention relates to the field of rice seedling visual recognition detection, in particular to a rice seedling visual recognition detection system. Background Art

[0002] Rice is one of the most important crops in the world. The number of rapid rice transplanters is increasing rapidly, but there has been no mature solution to the demand for operation monitoring of rapid rice transplanters.

[0003] In the actual use of the rice transplanter, there are often problems such as failure of working parts, insufficient water for seedlings, and negligence of the operator during the transplanting operation. This leads to stuck seedlings, stuck seedling trays, lack of seedlings, and continuous large-scale leakage of seedlings. As a result, the seedling density in some areas is low, the utilization rate of cultivated land decreases, and the yield per mu is affected. During the operation, there are special people on the transplanter responsible for monitoring the transplanting status. When seedlings are leaking, people need to be hired to manually fill the seedlings, resulting in optimizable labor costs and reduced labor intensity requirements. In many cases, filling the seedlings requires re-finding the specific location of the leaking seedlings in the paddy field, which reduces the efficiency of normal operations. Incomplete filling will reduce the average per-mu income of rice planting.

[0004] The rice transplanting missed seedling detection technology aims to fill the gaps in the rice transplanter operation detection methods. It can assist the driver in grasping the information on missed seedlings, reduce the need for manual monitoring on the rice transplanter, provide guidance on the seedling replenishment area, reduce labor costs, save the seedling replenishment time window, reduce production losses caused by low seedling coverage, and improve the economic benefits of agricultural production.

[0005] High-speed rice transplanters are developing into unmanned rice transplanters, and seedling leakage information is essential for the implementation of unmanned systems. There is a gap in the existing technology for rice seedling visual recognition and detection technology. Therefore, a rice seedling visual recognition and detection system is proposed to address the above problems. Summary of the invention

[0006] In order to make up for the deficiencies of the prior art, at least one technical problem raised in the background technology is solved.

[0007] The technical solution adopted by the present invention to solve its technical problem is: the rice seedling visual recognition and detection system described in the present invention includes a computing mainboard, a camera is connected to the left end of the computing mainboard, a flat plate is fixedly connected to the upper end of the computing mainboard, and an antenna and a three-color warning light are connected to the right end of the computing mainboard; it is characterized in that an MCU control board is provided on the surface of the computing mainboard, and a power switch and a function switch are embedded in the left end of the computing mainboard.

[0008] Preferably, the MCU control board is connected to the RDK X3 via a UART asynchronous receiver-transmitter.

[0009] Preferably, the MCU control board is connected to a GPIO small light board.

[0010] Preferably, an RTK board is provided on the surface of the computing main board.

[0011] Preferably, the RDK X3 is connected to the 4G module via USB.

[0012] The present invention mainly lies in the selection and connection relationship of hardware.

[0013] The present invention is beneficial in that:

[0014] The rice transplanting missed seedling detection product is designed to fill the gaps in the rice transplanter operation detection method. It can assist the driver in grasping the missed seedling information, reduce the need for manual monitoring on the rice transplanter, and provide guidance on the seedling replenishment area to reduce labor costs, save the seedling replenishment time window, reduce the production reduction caused by low seedling coverage, and improve the economic benefits of agricultural production.

[0015] The seedling leakage detection module uses visual sensors, image processing host and machine learning technology to identify the seedling claws that are leaking seedlings during the transplanting operation and output warning information according to the severity of the leakage. The use of high-speed transplanters has been fully popularized in large-scale rice planting areas. The seedling leakage detection module can assist the driver to grasp the information of seedling leakage and reduce the detection labor required on the transplanter. The background system generates a heat map of the seedling field coverage, and users can view the seedling shortage situation in real time and guide the needs of replenishing the missing seedlings.

[0016] The implementation of the rice transplanter seedling leakage monitoring system can improve agricultural production efficiency, reduce agricultural production costs, improve the quality of agricultural products, and promote the development of intelligent agriculture, which is of great significance for realizing agricultural modernization.

[0017] The specific advantages are as follows:

[0018] 1. In planting operations, rice transplanters are mainly used to replace manual labor to improve transplanting efficiency. With the construction of more and more unmanned farm experiments, the seedling leakage detection instrument can be well adapted to the unmanned rice transplanter to detect the transplanting results and generate the transplanting seedling coverage rate. The viewing end can check the seedling leakage area according to the seedling density to help personnel to carry out subsequent seedling replenishment.

[0019] 2. By timely discovering and replanting missed seedlings, the transplanting efficiency of farmland can be improved. Timely replanting of missed seedlings can ensure the density and uniformity of farmland and improve the quality and yield of agricultural products.

[0020] 3. Promote sustainable agricultural development: Through the application of automation technology, it reduces dependence on manual labor and reduces agricultural non-point source pollution: By timely replanting missed seedlings, the exposed area of ​​farmland is reduced, reducing the risk of soil erosion and agricultural non-point source pollution. BRIEF DESCRIPTION OF THE DRAWINGS

[0021] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative labor.

[0022] Figure 1 It is a schematic diagram of the structure of a rice seedling visual recognition and detection system;

[0023] In the figure: 1. Computing mainboard; 11. Power switch; 12. Function switch; 13. MCU control board; 14. RDK X3; 15. GPIO light board; 16. RTK board; 17. 4G module; 2. Tablet; 3. Camera; 4. Antenna; 5. Three-color warning light; DETAILED DESCRIPTION

[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0025] See also Figure 1 As shown, a rice seedling visual recognition detection system includes a computing mainboard 1, a camera 3 is connected to the left end of the computing mainboard 1, a flat panel 2 is fixedly connected to the upper end of the computing mainboard 1, an antenna 4 and a three-color warning light 5 are connected to the right end of the computing mainboard 1; an MCU control board 13 is provided on the surface of the computing mainboard 1, and a power switch 11 and a function switch 12 are embedded on the left end of the computing mainboard 1. The MCU control board is connected to the RDK X3 through a UART asynchronous transceiver transmitter, the MCU control board is connected to a GPIO small light board, an RTK board is provided on the surface of the computing mainboard, and the RDK X3 is connected to a 4G module through a USB. The implementation of the transplanter seedling leakage monitoring system can improve agricultural production efficiency, reduce agricultural production costs, improve the quality of agricultural products, promote the development of agricultural intelligence, and is of great significance for realizing agricultural modernization.

[0026] The MCU control board 13 is connected to the RDK X3 component 14 via a UART asynchronous receiver-transmitter.

[0027] The MCU control board 13 is connected to a GPIO small light board 15 .

[0028] An RTK board 16 is disposed on the surface of the computing mainboard 1 .

[0029] The RDK X3 component 14 is connected to the 4G module 17 via USB.

[0030] Working principle: when working, the image to be identified collected by the camera 3 is input into the trained neural network, and the network will output the category label of the image, thereby realizing automatic image identification. Whether the seedling leakage occurs is deduced based on the image recognition information.

[0031] After the host completes the image information detection in a single target area, it is necessary to mark the GPS information of multiple seedlings and upload it to the cloud platform. The GPS of seedlings at different positions in the marking detection relies on the standard GPS value of the host to calculate the GPS value of each seedling at different positions.

[0032] When the seedling leakage occurs, the control panel controls the warning light 5 to give a prompt (normally the green light is on, and the seedling leakage light is red).

[0033] The data detected during the operation is uploaded to the cloud computing platform through the communication module to collect data for the entire rice field.

[0034] A BP neural network model was established, and the rice seedling contour and morphological characteristic parameters extracted by the characteristic parameter extraction module were used as the input layer of the neural network, and the rice seedling morphological category was used as the output layer;

[0035] Rice seedling morphology recognition: Use the optimal BP neural network model obtained by the classification model building module to train the rice seedling images processed by the preprocessing module to complete the recognition of rice seedling morphology, and finally display the recognition results; the specific steps of establishing the classification model are: 1) Crop the original image A of the rice field seedlings collected in the early stage, separate the qualified seedling single-hole image B, floating seedling single-hole image C, and injured seedling single-hole image D, and place the qualified seedling single-hole image B, floating seedling single-hole image C, and injured seedling single-hole image D in the qualified seedling subfolder, floating seedling subfolder, and injured seedling subfolder in the Data folder respectively; 2) Create a BP neural network in the MATLAB environment, including an input layer, a hidden layer, and an output layer, and each layer is interconnected. , the number of input layer nodes selects the input effective feature vector dimension, the number of hidden layer nodes selects the result with the highest accuracy and the smallest error through trial and error, the number of output layer nodes takes the output result vector dimension, and the output result for a single-hole seedling picture is one of the three seedling categories; 3) the normalized feature parameter data set in the training set in step 1) is loaded into the input layer; in the three-layer BP neural network, the activation function of the hidden layer neurons is the logsig function, and the output layer activation function is the purelin function, and the activation function is used to perform linear and nonlinear transformations on the neuron output; 4) the training function of the BP neural network is selected as the traingdx function, and the learning function is selected as the learngdm function The learning function uses the input and error of the neuron, the learning rate of the weight or threshold to modify the increment of the weight and threshold, so that the output error of a single neuron is minimized. When the partial derivative of the error with respect to the weight is greater than zero, the weight adjustment amount is negative, the actual output is greater than the expected output, and the weight is adjusted in the direction of reduction, so that the difference between the actual output and the expected output is reduced; when the partial derivative of the error with respect to the weight is less than zero, the weight adjustment amount is positive, the actual output is less than the expected output, and the weight is adjusted in the direction of increase, so that the difference between the actual output and the expected output is reduced; 5) Set the network weight coefficient, which includes the number of hidden nodes, the maximum number of training times, the required training accuracy, the display training iteration process and the training learning rate; during the training process, the training function continuously calls the learning function , by detecting whether the set number of training steps is reached or whether the training result meets the accuracy requirement, it is determined whether to end the training, and the final qualified seedlings, floating seedlings and injured seedlings BP neural network model is obtained; 6) reading the single-hole seedling image of the verification set cut out in step 1), calculating the relevant feature parameter data, sending the feature parameter data to the input layer of the BP neural network, matching it with the BP neural network model obtained in step 5), and outputting the judgment result of the seedling type, which is one of the three types of qualified seedlings, floating seedlings and injured seedlings; comparing the judgment result with the subfolder name in step 1) where the single-hole seedling image is located, and evaluating the training effect; adjusting the hyperparameters again according to the accuracy of the verification set to obtain the optimal BP neural network model;7) Read the single-hole seedling images that are not involved in the training and verification process in the test data set in turn, identify a number of them, and calculate the accuracy rate. ;

[0036] The above shows and describes the basic principles, main features and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments, and the above embodiments and descriptions are only for explaining the principles of the present invention. Without departing from the spirit and scope of the present invention, the present invention may have various changes and improvements, and these changes and improvements all fall within the scope of the present invention to be protected.

Claims

1. A rice seedling visual recognition detection system, comprising a computing mainboard (1), a camera (3) connected to the left end of the computing mainboard (1), a flat plate (2) fixedly connected to the upper end of the computing mainboard (1), and an antenna (4) and a three-color warning light (5) connected to the right end of the computing mainboard (1); characterized in that: An MCU control board (13) is provided on the surface of the computing mainboard (1), and a power switch (11) and a function switch (12) are embedded at the left end of the computing mainboard (1).

2. The rice seedling visual recognition detection system according to claim 1, characterized in that: The MCU control board (13) is connected to the RDK X3 (14) via a UART asynchronous receiver-transmitter.

3. A rice seedling visual recognition detection system according to claim 1 or 2, characterized in that: The MCU control board (13) is connected to a GPIO small light board (15).

4. The rice seedling visual recognition detection system according to claim 1, characterized in that: An RTK board (16) is provided on the surface of the computing mainboard (1).

5. The rice seedling visual recognition detection system according to claim 2, characterized in that: The RDKX3 (14) is connected to the 4G module (17) via USB.

6. The rice seedling visual recognition detection system according to claim 1, characterized in that: When working, the image to be identified collected by the camera 3 is input into the trained neural network, and the network will output the category label of the image to realize automatic image identification, and deduce whether the seedling leakage occurs based on the image identification information; After the host completes the image information in a single target area, it needs to mark the GPS information of multiple seedlings and upload it to the cloud platform. The GPS of seedlings at different positions in the marking detection relies on the standard GPS value of the host to calculate the GPS value of each seedling at different positions; When the seedling leakage occurs, the control panel controls the warning light 5 to give a prompt, the normal light is green, and the seedling leakage light is red; The data detected during the operation is uploaded to the cloud computing platform through the communication module to collect data for the entire rice field.

Images