CN110134814B - An information system for ginger planting based on image recognition - Google Patents

Abstract

An information system based on image recognition, including a central service system and a terminal system; specifically, the central service system is composed of a pest identification system, a pest control measure system, an information management system, and a pest monitoring system; first, through the terminal The way of accessing the system and the central service system can easily and quickly serve many ginger growers, which is convenient for system maintenance and information update, and is more conducive to rapid promotion, high coverage, and timely and accurate information; secondly, the pest identification system combines pests and diseases The prevention and control measures system can provide reasonable and effective prevention and control measures at the first time when farmers inquire, with high efficiency and accuracy. Ginger planting and yield.

Description

An information system for ginger planting based on image recognition

technical field

The invention relates to a comprehensive information system, in particular to an image recognition-based ginger planting information system.

Background technique

In daily life, young ginger is in great demand. my country's young ginger has a wide planting area and high output and output value; in the process of young ginger planting, the pests and diseases of young ginger are extremely harmful. However, in the current environment of green and healthy life, how to rationalize The problem of effectively using drugs to ensure the yield of young ginger is that it is impossible to quickly and timely understand the types of diseases and insect pests during the planting process of young ginger, so as to prescribe the right medicine, reduce drug abuse, and how to monitor, statistics and comprehensively analyze the information on diseases and insect pests during the planting process of young ginger.

SUMMARY OF THE INVENTION

The object of the present invention is to provide a kind of ginger planting information system based on image recognition, through the method of image recognition, complete the long-range accurate classification and identification of diseases and insect pests in the young ginger planting process, information recording and prevention and control methods are provided.

The object of the present invention is realized according to the following technical solutions:

An information system for ginger planting based on image recognition, including a central service system and a terminal system;

The central service system is under the same management and provides services; including a pest identification system, a pest control measure system, an information management system, and a pest monitoring system;

The disease and insect pest identification system uses image recognition technology for the input image, extracts the image feature vector according to the ginger disease and insect pest identification model, searches the disease and insect pest image retrieval database by the feature vector, and returns the feature vector that best matches the input image and has a confidence level greater than a threshold. The value of the pest category of ginger;

Described juvenile ginger disease and insect pest identification model is a deep neural network multi-classification model completed by offline training; the input is a juvenile ginger picture, and the output is a juvenile ginger disease and insect pest feature vector;

The image retrieval database of diseases and insect pests is the disease and insect damage feature vectors of images of diseases and insect pests in the process of planting ginger; there is not less than one image of diseases and insect pests for a class of diseases and insect pests in the process of planting ginger, that is, there are not less than one feature vector of diseases and insect pests;

The retrieval is to calculate the feature vector and the confidence level of the disease and insect pest feature vector in the disease and insect pest image retrieval database, and the maximum confidence level is the retrieval result;

The confidence calculation formula:

In the formula, x is the feature vector, x _i is the feature vector of pests and diseases in the image retrieval database of pests and diseases, ||xx _i || is the two-norm calculation of the vector xx _i , and s _i is the confidence level of x with x _i ;

When the maximum confidence level is greater than the threshold value, output the pest category corresponding to the maximum confidence level, otherwise, the output result is empty;

The system of pest control measures mainly provides different control measures according to the categories of pests and diseases in the planting process of young ginger;

The control measures include the name of the drug used, the dose used, the frequency of use, precautions and known control effects;

the system of pest control measures, using the category of pests and diseases for retrieval;

Described information management system, mainly to the record of data in the process of planting ginger; each ginger grower has an independent information record, which is indexed by the grower ID; the information record contains information items that are planting number, planting site coordinates , Aberdeen ginger varieties, disease and insect pest incidence records, pesticide application records, historical production records;

The grower ID is an independent and non-repeating character string;

The disease and insect pest incidence records are the disease and insect pest incidence time, and the disease and insect pest category;

The pesticide application records are pesticide application time, pesticide name, and application dose;

The described disease and insect pest monitoring system, according to the records in the information management system, analyzes and predicts the disease and insect pest damage of all ginger planting areas according to the disease and insect pest incidence position, and takes timely and accurate measures; the calculation formula adopts distance calculation, and calculates the disease and insect pest damage value in each region;

为表征仔姜种植j地是否发生病虫害；v_i为计算出的仔姜种植i地病虫害危害因子，值越大危害越高；where d _ij is the distance between site i and site j where ginger is planted;

In order to characterize whether pests and diseases occur in the j site of the ginger planting; v _i is the calculated pest and disease hazard factor of the ginger planting site i, the larger the value, the higher the damage;

The terminal system has the functions of taking pictures, displaying and accessing the central service system, and is used for the collection of the information of the ginger in the process of planting the ginger, reporting to the central service system and obtaining professional pest control measures by the ginger growers;

The accessing the central service system includes using the grower account to inquire about information records, uploading images of young ginger to obtain identification results of pests and diseases, and obtaining control measures for pests and diseases.

More preferably, a kind of ginger planting information system based on image recognition is characterized in that: the ginger grower uploads the image of the young ginger to obtain the identification results of diseases and insect pests, and obtains the prevention and control measures of diseases and insect pests, and the steps are as follows:

S01. The terminal system uses the camera function to take pictures of the diseased and insect-infested parts of ginger, and sends the image, grower ID and password to form query application information to the central service system;

S02, the central service system receives the query application information, and verifies the grower ID and password; if the verification fails, directly returns the query application failure information; if the verification succeeds, perform the following steps;

S03. The central service system invokes the pest identification system, inputs pictures of the parts of ginger with pests and diseases, and outputs the identification results and confidence of pest categories;

S04. The central service system invokes the pest control measure system, and uses the identification result of the pest category to search to obtain the corresponding control measures;

S05, the central service system calls the information management system to record the query information and identification results;

S06. The central service system sends information on successful query application, where the information on successful query application includes: identification results of pest categories, confidence levels, and corresponding control measures;

S07. The terminal system receives the successful information of the query application, and displays the identification result, confidence level and corresponding control measures of the pest category;

The grower ID and password are set for the grower's first registration;

The first registration includes setting the grower's ID and password, assigning the planting number, and improving the grower's information;

The grower information includes: the coordinates of the planting site and the variety of ginger.

More preferably, an information system for planting ginger based on image recognition is characterized in that: the output result of the pest monitoring system is a map of pest hazard levels;

The disease and insect pest damage level map uses the depth of color to represent the damage level.

The present invention has the following beneficial effects:

The invention provides an information system for ginger planting based on image recognition, which can conveniently and quickly serve numerous ginger growers by accessing the terminal system and the central service system, facilitates system maintenance, information update, and is more conducive to popularization Fast, high coverage, timely and accurate information; secondly, the central service system uses the pest identification system for the identification of pests and diseases, which is real-time and accurate, and automatic identification saves manpower and material resources. Combined with the pest control measure system, it can be given at the first time of the grower's inquiry. Reasonable and effective prevention and control measures have high efficiency and accuracy; finally, the central service system has the functions of information collection, recording and sorting, which can centrally analyze the planting situation and yield of young ginger in a large range.

Description of drawings

Figure 1: A schematic diagram of the structure of a ginger planting information system based on image recognition.

Figure 2: The flow chart of the one-time inquiry application system.

Detailed ways

The present invention will be specifically described below through the examples. It is necessary to point out that the following examples are only used to further illustrate the present invention, and should not be construed as limiting the protection scope of the present invention. The content makes some non-essential improvements and adjustments to the present invention.

Example 1

An image recognition-based ginger planting information system, as shown in Figure 1, includes a central service system and several terminal systems, wherein each terminal system independently accesses the central service system;

The central server system consists of a pest identification system, a pest control measure system, an information management system, and a pest monitoring system;

The disease and insect pest identification system adopts image recognition technology to extract the image feature vector from the input image according to the ginger disease and insect pest identification model. The feature vector is retrieved in the disease and insect pest image retrieval database, and the image feature vector that best matches the input image feature vector and whose confidence is greater than the threshold value is returned. Category of ginger diseases and insect pests;

The identification model of ginger diseases and insect pests is a deep neural network multi-classification model completed by offline training; the input is a picture of ginger; the output is a feature vector of diseases and insect pests of ginger;

The ResNeXt basic network is used in the offline training process of the ginger pest identification model, and the training loss function adopts the Triplet loss function;

The image retrieval database of diseases and insect pests is a collection of disease and insect pest feature vectors of images of diseases and insect pests in the process of planting ginger. Among them, there is no less than one image of diseases and insect pests for one type of disease and insect pests in the process of planting ginger, that is, a disease and insect pest may have no less than one feature vector of pests and diseases. ; In the retrieval process, the formula for calculating the confidence level of the input feature vector and the disease and insect pest feature vector in the disease and insect pest image retrieval database is as follows:

In the formula, x is the feature vector, x _i is the feature vector of pests and diseases in the image retrieval database of pests and diseases, ||xx _i || is the two-norm calculation of the vector xx _i , and s _i is the confidence level of x with x _i ;

When the retrieval result requires that the maximum confidence level is greater than the threshold, output the pest category corresponding to the maximum confidence level, otherwise, the output result is empty;

The system of pest control measures mainly provides different control measures according to different types of pests and diseases in the planting process of young ginger, including the name of the drug used, the dosage, the number of times of use, precautions and the corresponding known control effects; the system of pest control measures uses pests and diseases Search by category; among them, the prevention and control measures for non-hearing diseases and insect pests are obtained by professional staff through analysis and experiments, and have high accuracy;

The information management system records the data in the process of planting ginger; corresponding to each ginger grower, there is an independent information record; it is indexed by the grower ID; the record contains information items such as planting number, planting site coordinates, Ginger varieties, disease and pest incidence records, pesticide application records, and historical yield records; the grower ID is an independent and non-repetitive string, the disease and insect pest incidence records include the disease and insect pest incidence time, and the pest and disease types, and the pesticide application records include the pesticide name, pesticide application time, dosage;

The pest monitoring system is based on the records in the information system, organizes and analyzes regularly, and judges the pest and disease hazards of all ginger planting sites according to the predetermined amount of pests and diseases, so as to take timely and accurate measures; The distance is introduced into the calculation formula to obtain the pest and disease hazards of various regions. value:

为表征仔姜种植j地是否发生病虫害；v_i为计算出的仔姜种植i地病虫害危害因子，值越大危害越高；病虫害监控系统输出结果绘制为病虫害危害等级地图；病虫害危害等级地图使用颜色深浅表征危害等级，病虫害危害值越大，病虫害危害等级越高，颜色越深。where d _ij is the distance between site i and site j where ginger is planted;

In order to characterize whether pests and diseases occur in the j site of the ginger planting; v _i is the calculated pest and disease hazard factor of the site i of the ginger planting, the larger the value, the higher the damage; the output result of the pest monitoring system is drawn as the pest and disease hazard level map; the pest and disease hazard level map uses The color depth represents the hazard level. The larger the pest hazard value, the higher the pest hazard level and the darker the color.

The terminal system has the functions of taking pictures, displaying and accessing the central service system, which is used by the ginger growers to collect information on the ginger during the planting process, report it to the central service system, and obtain professional pest control measures; accessing the central service system includes using The grower account can query the information record, upload the ginger image to obtain the identification results of pests and diseases, and obtain the control measures of pests and diseases.

Example 2:

A kind of ginger planting information system based on image recognition, the ginger grower uploads the image of the young ginger to obtain the identification results of pests and diseases, and obtain the control measures of pests and diseases, as shown in Figure 2, the specific steps are as follows:

S01. The terminal system uses the camera function to take pictures of the diseased and insect-infested parts of ginger, and sends the image, grower ID and password to form query application information to the central service system;

S02, the central service system receives the query application information, and verifies the grower ID and password; if the verification fails, directly returns the query application failure information; if the verification succeeds, perform the following steps;

S03. The central service system invokes the pest identification system, inputs pictures of the parts of ginger with pests and diseases, and outputs the identification results and confidence of pest categories;

S04. The central service system invokes the pest control measure system, and uses the identification result of the pest category to search to obtain the corresponding control measures;

S05, the central service system calls the information management system to record the query information and identification results;

S06. The central service system sends information on successful query application, where the information on successful query application includes: identification results of pest categories, confidence levels, and corresponding control measures;

S07. The terminal system receives the successful information of the query application, and displays the identification result, confidence level and corresponding control measures of the pest category;

Among them, the grower ID and password are set for the first registration of the grower; the first registration includes setting the grower ID and password, assigning the planting number, and improving the information of the grower; the information of the grower includes the coordinates of the planting place and the variety of ginger.

Claims (2)

1. An information system based on image recognition, comprising a central service system and a terminal system; The central service system is composed of a pest identification system, a pest control measure system, an information management system, and a pest monitoring system; The disease and insect pest identification system uses image recognition technology for the input image, extracts the image feature vector according to the ginger disease and insect pest identification model, searches the disease and insect pest image retrieval database by the feature vector, and returns the feature vector that best matches the input image and has a confidence level greater than a threshold. The value of the pest category of ginger; Described juvenile ginger disease and insect pest identification model is a deep neural network multi-classification model completed by offline training; the input is a juvenile ginger picture, and the output is a juvenile ginger disease and insect pest feature vector; The image retrieval database of diseases and insect pests is the disease and insect damage feature vectors of images of diseases and insect pests in the process of planting ginger; there is not less than one image of diseases and insect pests for a class of diseases and insect pests in the process of planting ginger, that is, there are not less than one feature vector of diseases and insect pests; The retrieval is to calculate the feature vector and the confidence level of the disease and insect pest feature vector in the disease and insect pest image retrieval database, and the maximum confidence level is the retrieval result; Confidence calculation formula:

In the formula, x is the feature vector, x _i is the feature vector of pests and diseases in the image retrieval database of pests and diseases, ||xx _i || is the two-norm calculation of the vector xx _i , and s _i is the confidence level of x with x _i ; When the maximum confidence is greater than the threshold value, output the category of ginger diseases and insect pests corresponding to the maximum confidence, otherwise, the output result is empty; The system of pest control measures provides different control measures according to the categories of pests and diseases in the planting process of juvenile ginger; The control measures include the name of the drug used, the dose used, the frequency of use, precautions and known control effects; the system of pest control measures, using the category of pests and diseases for retrieval; Described information management system, to the record of data in the process of planting ginger; each ginger grower has an independent information record, which is indexed by the grower ID; The information records include information items such as planting number, planting site coordinates, young ginger variety, disease and insect pest incidence records, pesticide application records, and historical yield records; The grower ID is an independent and non-repeating character string; The disease and insect pest incidence records are the disease and insect pest incidence time, and the disease and insect pest category; The pesticide application records are pesticide application time, pesticide name, and application dose; The pest monitoring system, according to the records in the information management system, analyzes and predicts the pest damage of all ginger planting areas according to the location of the disease and pest incidence, and takes timely and accurate measures; the calculation formula adopts distance calculation to calculate the pest damage value of each region;

where d _ij is the distance between site i and site j where ginger is planted;

In order to characterize whether pests and diseases occur in the j site of the ginger planting; v _i is the calculated pest and disease hazard factor of the ginger planting site i, the larger the value, the higher the damage; The terminal system has the functions of taking pictures, displaying and accessing the central service system, and is used for growers to access the central service system and display the results; The accessing the central service system includes using the grower account to inquire about information records, uploading images of young ginger to obtain identification results of pests and diseases, and obtaining control measures for pests and diseases; The said ginger grower uploads the image of the young ginger to obtain the identification result of pests and diseases, and obtains the control measures of pests and diseases, and the steps are as follows: S01. The terminal system uses the camera function to take pictures of the diseased and insect-infested parts of ginger, and sends the image, grower ID and password to form query application information to the central service system; S02, the central service system receives the query application information, and verifies the grower ID and password; if the verification fails, directly returns the query application failure information; if the verification succeeds, perform the following steps; S03. The central service system invokes the pest identification system, inputs the pictures of the pest and disease parts of ginger, and outputs the identification result and confidence of the pest category; S04. The central service system invokes the pest control measure system, and uses the identification result of the pest category to retrieve the corresponding control measures; S05, the central service system calls the information management system to record the query information and identification results; S06. The central service system sends information on successful query application, where the information on successful query application includes: identification results of pest categories, confidence levels, and corresponding control measures; S07. The terminal system receives the successful information of the query application, and displays the identification result, confidence level and corresponding control measures of the pest category; The grower ID and password are set for the grower's first registration; The first registration includes setting the grower's ID and password, assigning the planting number, and improving the grower's information; The grower information includes: the coordinates of the planting site and the variety of ginger. 2. a kind of young ginger information system based on image recognition as claimed in claim 1, is characterized in that: the output result of pest monitoring system is pest hazard level map; The disease and insect pest damage level map uses the depth of color to represent the damage level.

Images