WO2023065989A1

WO2023065989A1 - Plant disease and insect pest diagnosis method and system, and readable storage medium

Info

Publication number: WO2023065989A1
Application number: PCT/CN2022/121668
Authority: WO
Inventors: 徐青松; 李青
Original assignee: 杭州睿胜软件有限公司
Priority date: 2021-10-21
Filing date: 2022-09-27
Publication date: 2023-04-27
Also published as: CN113947697A

Abstract

A plant disease and insect pest diagnosis method and system, and a readable storage medium. The plant disease and insect pest diagnosis method comprises: acquiring a plant image to be diagnosed, and pre-identifying the plant image by using a disease diagnosis engine to obtain a disease pre-identification result (S1); if the confidence of the disease pre-recognition result is less than a first preset value, outputting an interaction question associated with the disease pre-recognition result (S2); and acquiring an answer for the interaction question, obtaining a disease result for the plant image according to the answer, and outputting diagnosis information corresponding to the disease result (S3). In this way, by outputting the interaction problem associated with the disease pre-identification result, a user selects the answer to perform processing, so as to determine the disease result for the plant image according to the answer, which is equivalent to further performing auxiliary confirmation on the disease pre-identification result by means of additional extra input information, thereby effectively improving the diagnosis accuracy of a plant disease.

Description

Plant disease and insect pest diagnosis method, diagnosis system and readable storage medium

technical field

The invention relates to the technical field of object recognition, in particular to a method for diagnosing plant diseases and insect pests, a diagnosing system and a readable storage medium.

Background technique

During the growth of plants, they often encounter problems such as diseases and insect pests. Currently, plant diseases can be identified by a plant disease diagnosis engine. However, the existing plant disease diagnosis engine has a limited accuracy in identifying plant diseases, causing problems for users.

Contents of the invention

The object of the present invention is to provide a method for diagnosing plant diseases and insect pests, a system for diagnosing plant diseases and insect pests, and a readable storage medium, so as to solve the problem of low identification accuracy of existing plant diseases.

In order to solve the above technical problems, according to the first aspect of the present invention, a method for diagnosing plant diseases and insect pests is provided, which includes:

Obtaining a plant image to be diagnosed, and using a disease diagnosis engine to pre-identify the plant image to obtain a disease pre-identification result;

If the confidence of the disease pre-identification result is less than a first preset value, output an interactive question associated with the disease pre-identification result; and

The answer to the interactive question is obtained, and the disease result of the plant image is obtained according to the answer, and the diagnosis information corresponding to the disease result is output.

Optionally, the interactive question includes at least two options, and the answer to the interactive question is selected from at least two options.

Optionally, the interactive question includes at least two levels, and the different selection branches at the upper level correspond to different branch questions at the lower level.

Optionally, the pre-identification results of the disease are at least two.

Optionally, each of the pre-identification results of the disease is associated with at least one interactive question, or at least two of the pre-identification results of the disease are associated with at least one interactive question.

Optionally, if the confidence of the disease result is not less than a second preset value, the diagnosis information includes a reference image and/or a disease block diagram, the reference image is a preset image corresponding to the disease result, The disease block diagram shows an image of the plant image at least including a part of the disease area, and marks the disease area.

Optionally, if the confidence of the disease result is not less than a second preset value, the diagnosis information includes summary diagnosis data and detailed diagnosis data.

Optionally, if the confidence of the disease result is less than the second preset value, and/or the image of the plant to be diagnosed does not meet the first preset condition, the diagnostic information includes diagnostic summary data, and prompts Reacquire the plant image.

Optionally, before using the disease diagnosis engine to pre-identify the plant image, the method for diagnosing plant diseases and insect pests further includes: acquiring a species identification result corresponding to the plant image;

Furthermore, after the step of pre-identifying the plant image by using the disease diagnosis engine, the disease pre-identification result is screened out according to the second preset condition.

Optionally, if the identification confidence of the species identification result is not less than a third preset value, the method for diagnosing plant diseases and insect pests further includes: displaying the species identification result and common diseases of the corresponding plant species.

Optionally, the species recognition result is obtained by using a species recognition engine to recognize the plant image, or the species recognition result is extracted from a recognized species recognition result page.

In order to solve the above technical problem, according to the second aspect of the present invention, there is also provided a readable storage medium on which a program is stored, and when the program is executed, the method for diagnosing plant diseases and insect pests as described above is realized.

In order to solve the above technical problems, according to the third aspect of the present invention, a plant disease and pest diagnosis system is also provided, which includes a processor and a memory, and a program is stored in the memory, and when the program is executed by the processor , realizing the method for diagnosing plant diseases and insect pests as described above.

In summary, in the method for diagnosing plant diseases and insect pests, the system for diagnosing plant diseases and insect pests, and the readable storage medium provided by the present invention, the method for diagnosing plant diseases and insect pests includes: acquiring an image of a plant to be diagnosed, and using a disease diagnosis engine to analyze the image of the plant Perform pre-identification to obtain a disease pre-identification result; if the confidence of the disease pre-identification result is less than a first preset value, output an interactive question associated with the disease pre-identification result; and obtain an answer to the interactive question , and obtain the disease result of the plant image according to the answer, and output the diagnosis information corresponding to the disease result.

With such a configuration, by outputting the interactive question associated with the pre-identification result of the disease, the user selects the answer for processing, so that the disease result of the plant image is determined according to the answer, which is equivalent to the pre-identification result of the disease through additional input information. Further auxiliary confirmation effectively improves the diagnostic accuracy of plant diseases.

Description of drawings

Those of ordinary skill in the art will understand that the provided drawings are for better understanding of the present invention, but do not constitute any limitation to the scope of the present invention. in:

Fig. 1 is the flowchart of the plant disease and insect pest diagnosis method of the embodiment of the present invention;

Fig. 2 is a schematic diagram of a low-confidence diagnosis details page according to an embodiment of the present invention;

Fig. 3 is a schematic diagram of a high-confidence diagnosis details page according to an embodiment of the present invention;

Fig. 4 is a schematic diagram of an interactive problem in an embodiment of the present invention;

Fig. 5 is a schematic diagram showing common diseases of a certain plant according to an embodiment of the present invention.

Detailed ways

In order to make the purpose, advantages and features of the present invention clearer, the present invention will be further described in detail below in conjunction with the accompanying drawings and specific embodiments. It should be noted that the drawings are all in very simplified form and not drawn to scale, and are only used to facilitate and clearly assist the purpose of illustrating the embodiments of the present invention. In addition, the structures shown in the drawings are often a part of the actual structures. In particular, each drawing needs to display different emphases, and sometimes uses different scales.

As used in this specification, the singular forms "a", "an" and "the" include plural objects, the term "or" is usually used in the sense of including "and/or", and the term "several" Usually, the term "at least one" is used in the meaning of "at least one", and the term "at least two" is usually used in the meaning of "two or more". In addition, the terms "first", "second "Two" and "third" are used for descriptive purposes only, and should not be understood as indicating or implying relative importance or implicitly indicating the quantity of the indicated technical features. Thus, a feature defined as "first", "second" and "third" may explicitly or implicitly include one or at least two of these features. Those of ordinary skill in the art can understand the specific meanings of the above terms in this specification according to specific situations.

Description is made below with reference to the accompanying drawings.

Please refer to Fig. 1, an embodiment of the present invention provides a method for diagnosing plant diseases and insect pests, which includes:

Step S1: Acquire the plant image to be diagnosed, and use the disease diagnosis engine to pre-identify the plant image to obtain the disease pre-identification result;

Step S2: if the confidence of the disease pre-identification result is less than a first preset value, output an interactive question associated with the disease pre-identification result; and

Step S3: Obtain the answer to the interactive question, and obtain the disease result of the plant image according to the answer, and output diagnosis information.

Each step of the method for diagnosing plant diseases and insect pests provided in this embodiment will be described in detail below with reference to the accompanying drawings and several examples.

Step S1: Regarding the plant image to be diagnosed, in some examples, the plant image uploaded by the user can be directly obtained. In other examples, after receiving the user instruction, corresponding prompt information may be generated and output to prompt the user to upload the plant image. Furthermore, the prompt information may also include specific requirements for plant images, such as prompting the user to upload images of the whole plant, partial images of plant stems, leaves, etc., or partial images of parts with obvious lesions, etc. In this case, preprocessing such as marking can also be performed on multiple plant images, such as marking the whole plant image, partial plant images (including marking the parts of the plant reflected in the plant image), etc., in order to better accurately identify species or diagnose pests and diseases.

Regarding the disease diagnosis engine, it may be a pre-trained disease diagnosis engine, and the disease diagnosis engine may include a neural network model, specifically a convolutional neural network model or a residual network model.

The convolutional neural network model is a deep feedforward neural network, which uses a convolution kernel to scan the plant image, extracts the features to be identified in the plant image, and then identifies the features to be identified in the plant. In addition, in the process of recognizing plant images, the original plant images can be directly input into the convolutional neural network model without preprocessing the plant images. Compared with other recognition models, the convolutional neural network model has higher recognition accuracy and recognition efficiency.

Compared with the convolutional neural network model, the residual network model has more identity mapping layers, which can avoid the phenomenon of saturation or even decline in the accuracy rate as the network depth (the number of stacked layers in the network) increases. The identity mapping function of the identity mapping layer in the residual network model needs to satisfy: the sum of the identity mapping function and the input of the residual network model is equal to the output of the residual network model. After the introduction of identity mapping, the residual network model has more obvious changes in the output, so the recognition accuracy and recognition efficiency of plant diseases can be greatly improved.

Step S2: Regarding the confidence level, since the disease diagnosis engine is not 100% reliable in identifying the disease, it has a certain possibility of error, so the probability that the disease pre-identification result identified by the disease diagnosis engine is consistent with the corresponding real disease ( That is, the degree of confidence that the pre-identification result of the disease is close to the real disease) is called confidence. It is easy to understand that if the confidence level is closer to 1, it means that the disease pre-identification result identified by the disease diagnosis engine is closer to the corresponding real disease, and the recognition result is more credible, and if the confidence level is closer to 0, it represents the disease The less credible the pre-identification result of the disease identified by the diagnosis engine is.

Optionally, considering the recognition accuracy and operation convenience of the plant pest diagnosis method, the first preset value can be set and adjusted according to the actual situation of different application scenarios. In an example, the first The preset value can be set to 0.9, for example. If the confidence of the disease pre-identification result is not less than the first preset value (such as ≥0.9), it can be considered that the authenticity of the recognition result is high, and it is not necessary to use interactive questions to assist confirmation, but directly obtain the disease based on the disease pre-identification result. result. If the confidence level of the disease pre-identification result is less than the first preset value, that is, the confidence level is lower than the first preset value (eg, <0.9), the confirmation is assisted by interactive questions, so as to improve the diagnostic accuracy of plant diseases.

Step S3: With regard to obtaining answers to interactive questions, in some examples, information obtained by various means such as user touch, click, input, or voice recording may be obtained as feedback answers to interactive questions. After obtaining the user's answer to the interactive question, such as after selecting the option, the disease result of the plant image can be obtained, and the diagnosis information can be output, such as jumping to the diagnosis result page.

In some embodiments, after the disease diagnosis engine pre-identifies a certain plant image to be diagnosed, only one disease pre-identification result can be obtained, that is, for a certain plant image to be diagnosed, after the disease diagnosis engine performs pre-identification, Only one disease pre-identification result can be obtained, or after getting more than two disease pre-identification results, only the only one remains after being screened out by other preset conditions (such as the third preset condition, see below for details) If there is a pre-identification result of the disease, the interactive question can be raised for the unique pre-identification result of the disease. After getting the user's answer to the interactive question, the disease pre-identification result can be further determined as the disease result, or if the answer to the interactive question does not meet the disease, the disease pre-identification result can not be output, and the disease pre-identification result can be further Abandoned, and the plant was determined to be free from disease and pests.

In some embodiments, for a certain plant image to be diagnosed, after the disease diagnosis engine performs pre-identification, at least two disease pre-identification results can be obtained, or after obtaining multiple disease pre-identification results, other preset After the conditions are screened out (such as the third preset condition, see below for details), at least two disease pre-identification results remain.

Optionally, the disease result obtained in step S3 also has its own confidence level, and the output diagnostic information may be different according to the confidence level of the disease result. In one example, with the second preset value as the threshold, when the confidence of the disease result is not less than the second preset value, the disease result is considered to be of high confidence, and when the confidence of the disease result is less than the second predetermined value When setting the value, it is considered that the result of the disease is of low confidence, and the second preset value can be set to 0.7, for example. Based on this, the output diagnostic information can also be divided into a high-confidence diagnostic detail page and a low-confidence diagnostic detail page. In some embodiments, according to the answers to the interactive questions, more than two disease results can be obtained, and different disease results have different confidence levels. In an exemplary embodiment, for example, the diagnostic information displaying different disease results can be switched by sliding left and right.

In one example, diagnostic information can be extracted from a content management system. A content management system (CMS) can be a software system that sits between the front-end and back-end systems or processes of the WEB. The content management system can be used to submit, modify, publish, etc. content such as text files, pictures, data in databases, tables, and the like. The content management system can also provide content grabbing tools to automatically grab content from third parties such as text files, HTML web pages, Web services, databases, etc., and put them into the corresponding content library of the content management system after analysis and processing. middle. The content management system can also assist the WEB front-end to provide content to users in a personalized manner, that is, provide a personalized portal framework to better push content to users based on WEB technology. In the content management system in the embodiment of the present disclosure, descriptive content on plants and their diseases may be stored, and these descriptive content may be text or pictures, for example, may include various fields, articles, etc., so that It enables users to obtain introductions about plants and their diseases in the diagnostic information extracted and output from the content management system, such as interesting stories, uses of plants, maintenance methods and descriptions of diseases, etc.

One-to-one correspondence with each species information may include a species name (UID1) to distinguish different species. Similarly, the one-to-one correspondence with each disease result may include a disease name (UID2) to distinguish different diseases. When extracting relevant diagnostic information in the content management system, it can be retrieved according to UID1 and UID2. When a large amount of data is pre-stored in the content management system, it can cover most diagnostic situations and provide users with corresponding diagnostic information.

Based on the content management system, the relevant information of multiple disease results can be output to the user in the form of one disease result corresponding to one card. Users can switch and display the results of various diseases and related information by sliding cards on the interactive interface.

In some embodiments, for different plant images, at least part of the diagnostic information may change with different plant images if the determined disease result is the same. In this way, even if the obtained disease results are the same, the output diagnostic information can be adaptively changed according to the plant image input by the user, which realizes a more flexible output and helps to make the output diagnostic information consistent with the user's input. To improve the user experience and reduce the confusion caused by the mismatch of input and output to users.

In some embodiments, diagnostic data may include diagnostic summary data and/or diagnostic detailed data. In the diagnostic summary data and the diagnostic detailed data, different fields can be set to store the data extracted from the content management system in corresponding fields. The step of outputting diagnostic information may include: In the content management system, according to the determined disease result, corresponding data is extracted according to preset output fields to generate diagnostic information, and the diagnostic information is output. The preset output fields can be set by the user through an interactive interface according to their own needs, or the preset output fields can also be several relatively fixed fields. In the content management system, corresponding diagnostic data extracted according to the determined identification information may be filled in a corresponding template with a preset output format to form diagnostic information.

Preferably, if the complete data cannot be extracted according to the preset output fields in the content management system, the diagnostic information can be generated by searching corresponding documents.

In some embodiments, the diagnosis summary data may include at least one of a condition name corresponding to a condition name field in the preset output field and a diagnosis summary corresponding to a diagnosis summary field in the preset output field. The diagnosis detailed data may include symptom analysis corresponding to the symptom analysis field in the preset output field, solution corresponding to the solution field in the preset output field, and preventive action corresponding to the preventive action field in the preset output field at least one of . By storing symptom analysis, solutions, and preventive measures in different fields, diagnostic information can be easily generated based on content management systems, and it is also convenient for users to view.

Preferably, if the confidence of the disease result is less than the second preset value (i.e. low confidence), and/or, the image of the plant to be diagnosed does not meet the first preset condition, the diagnosis information includes Diagnose summary data and prompt to reacquire plant images. Please refer to FIG. 2 , which shows a diagnosis details page (Diagnose) showing disease results with low confidence (ie, the confidence is less than the second preset value) in an exemplary embodiment. Wherein the first column 101 shows the diagnosis summary data, specifically including the disease name (in the example shown in Figure 2 for brown spot disease Brown spot) and diagnosis summary (in the example shown in Figure 2 for The brown streaks on the leaves of your plant are signs of brown spot. The second column 102 prompts to reacquire the plant image. For example, the user can click the shooting button 104 next to it to take and upload the plant image again, so as to return and re-execute step S1. As for the first preset condition here, for example, the clarity of the plant image can meet the requirements, or if the plant image includes the entire range of the diseased area, etc., those skilled in the art can set it according to the actual situation. If the plant image does not meet the first preset condition, such as the range of the plant image taken by the user is too small to cover the entire range of the disease area, it will be difficult to obtain reliable disease results later. In these cases, it may prompt to re-acquire the plant image, so as to return and re-execute step S1.

Preferably, if the confidence of the disease result is not less than a second preset value (ie, high confidence), the diagnostic information includes diagnostic summary data and diagnostic detailed data. Please refer to FIG. 3 , which shows a diagnosis details page (Diagnose) showing disease results with high confidence (ie, the confidence is not less than the second preset value) in an exemplary embodiment. In the example shown in FIG. 3, the details page includes a first column 101 and a second column 102. The first column 101 shows diagnostic summary data, including disease name and diagnosis summary (in the example shown in FIG. The name is Brown spot, and the diagnostic summary is The brown streaks on the leaves of your plant are signs of brown spot. The brown streaks on the leaves of your plant are signs of brown spot). The second column 102 displays the detailed diagnosis data, which may include various corresponding contents related to the disease, such as symptom analysis, solutions, prevention and so on.

Further, if the confidence of the disease result is not less than the second preset value (ie, high confidence), the diagnostic information includes a reference map and/or a disease block diagram, and the reference map is corresponding to the disease result The preset image, the disease block diagram shows the image of the plant image at least including part of the disease area, and the disease area is marked with a frame.

Please continue to refer to FIG. 3 , a reference image may be included in part of the diagnostic information that may be adaptively changed following the plant image. In the example shown in FIG. 3 , the details page includes a third column 103 and a fourth column 100 . The third column 103 shows the reference graph. The reference image corresponds at least to the disease outcome, and the reference image is similar to the plant image. In this way, the output diagnostic information can no longer be fixed, but can replace the relevant pictures used for explanation in the output diagnostic information according to the plant image input by the user, so that these pictures used for explanation are consistent with the The plant images taken by the user are more similar, so that the user will not feel that the image in the output diagnostic information is too different from the plant image taken by the user, so as to avoid causing trouble to the user and improve user experience.

Optionally, the reference image can be preset in the candidate reference gallery of the content management system. In the content management system, the corresponding candidate reference gallery is determined according to the disease result; in the candidate reference gallery, based on the similarity with the plant image and/or Or the matching degree with the species information, determine the priority of one or more reference pictures extracted and each reference picture corresponding to one or more reference pictures; and then output one or more reference pictures, so that One or more reference images are arranged in descending order of priority.

Wherein, each reference image in the content management system can be marked with UID1 of corresponding species information (UID1 may include species, variety, variety, genus, family, etc.) and UID2 of disease result. Based on UID1 and UID2, the reference images can be classified, screened, and so on. For example, according to UID2, one or more reference images corresponding to each disease result can be combined into a candidate reference gallery corresponding to the corresponding disease result. When selecting the required reference picture in the reference gallery, the type of plant corresponding to the reference picture can be determined according to the UID1 marked on each reference picture. By displaying the reference picture, it can help the user to better identify the disease of the plant, especially when the picture of the plant taken by the user is not clear or the shooting part is not good.

In general, the higher the similarity with the plant image and the higher the matching degree with the species information, the reference image will have a higher priority. The reference images with higher priority may be displayed preferentially or arranged at the front position of the multiple displayed reference images, so as to facilitate viewing by the user.

In some embodiments, in the candidate reference gallery, based on the similarity with the plant image and/or the matching degree with the species information, determine the extracted one or more reference images and the one or more extracted reference images. Or the priority corresponding to each reference picture in multiple reference pictures may include:

Using a preset number of candidate reference images in the candidate reference gallery with the highest similarity with the plant image as the first reference atlas, and using all other candidate reference images in the candidate reference gallery as the second reference atlas;

In the first set of reference images, determining a first reference image matching the species information at a first species classification level, wherein the determined first reference image has a first priority;

In the second reference atlas set, determine a second reference atlas matching the species information at the first species classification level, wherein the determined second reference atlas has a second priority, and the second priority is lower than the first a priority;

In the first set of reference images, determining a third reference image matching the species information at a second species classification level higher than the first species classification level, wherein the determined third reference image has a third priority, and the third priority is lower than the second priority; and

In the second set of reference images, determine a fourth reference image that matches the species information at the second species classification level, wherein the determined fourth reference image has a fourth priority, and the fourth priority is lower than the fourth priority Three priorities.

In a specific example, six candidate reference images that are closest to the image features of the plant image uploaded by the user can be determined from the candidate reference gallery as the first atlas, and other candidate reference images in the candidate reference gallery are used as the second. Atlas. First, in the first atlas, find the first reference image that matches the species of the plant in the plant image, and the first reference image has the highest first priority; The second reference image that matches the species of the plant, the second reference image has a second priority lower than the first priority; then, in the first atlas, find the third reference image that matches the genus of the plant in the plant image A reference map, the third reference map has a third priority lower than the second priority; then, in the second atlas, find a fourth reference map matching the genus of the plant in the plant image, the fourth reference map Has a fourth priority lower than the third priority; then, in the first atlas, find a fifth reference image that matches the family of the plant in the plant image, and the fifth reference image has a lower priority than the fourth Fifth priority; finally, in the second atlas, a sixth reference image matching the family of the plants in the plant image is found, which has the lowest sixth priority. When displaying reference images, they can be displayed in the order of the first reference image, the second reference image, the third reference image, the fourth reference image, the fifth reference image, and the sixth reference image, wherein the first reference image is displayed in the most conspicuous position.

In some embodiments, in the candidate reference gallery, based on the similarity with the plant image and/or the matching degree with the species information, it is determined that one or more extracted reference images and one or more The step of prioritizing each reference image in the reference images may also include:

When a reference map matching the species information at a species classification level lower than or equal to a preset species classification level cannot be determined in both the first reference atlas and the second reference atlas, the predicted results corresponding to the disease result will be Let the default image be determined as the reference image.

For example, if you still cannot find a matching result reference image in the first atlas or the second atlas when searching for the classification level of family, you can use the preset default image in the reference gallery as the reference image without further searching .

Usually, the display ratio of the pictures in the displayed diagnostic information is between 3:2 and 1:1, so as to have a better display effect. However, the scale of the reference images screened from the candidate reference gallery may not be suitable for the above display scale. Typically, such plots can be stretched or cropped to fit the display scale. However, considering that the stretching of the reference image may lead to deformation of the characteristics of certain diseases, which is not conducive to the user's good identification of the disease, so in an exemplary embodiment, the reference image can be processed by cropping . in particular,

Methods for diagnosis of plant diseases may also include:

Crop the edge area of the original image that forms the reference image, so that the scale of the reference image obtained after cropping conforms to the preset display ratio, and the image features corresponding to the disease results in the reference image are located in the middle area of the reference image Inside.

Specifically, based on the area recognition model, when selecting reference image materials for the content management system, the pictures corresponding to the disease results whose image features are located in the marginal area are removed or ignored, and these pictures will not be included in the content management system . Alternatively, processing such as cropping may be performed on the picture when it is stored in the content management system. Alternatively, after the reference image to be output is determined according to the plant image, the reference image selected by the content management system may be processed, for example, cropped, before being output. Of course, in some other embodiments, it is also possible to pre-determine the positions of the features of the disease in the original image forming the reference image, and avoid these positions during the cropping process.

In the example shown in Fig. 3, the fourth column 100 shows the disease block diagram. The disease block diagram shows the image of the plant image taken by the user at least including part of the diseased area, preferably the whole plant image. The disease area is marked with a frame 100a to prompt the user where the disease occurs for comparison.

Regarding the way of outputting interactive questions, in some examples, prompts can pop up in the diagnostic page, such as display or voice prompts. Those skilled in the art can set and adjust according to actual conditions.

The method for diagnosing plant diseases and insect pests provided by this embodiment is described below through an example:

First, the user uploads an image of a plant to be diagnosed, which can be captured by the user. The disease diagnosis engine pre-identifies the plant image, and two disease pre-identification results are obtained, which are A disease (such as yellow leaves-overwatering) and B disease (such as yellow leaves-lack of water), and The confidence levels of the two disease pre-identification results are both less than the first preset value (0.9), thus triggering the output of an interactive question associated with the disease pre-identification result:

"Is the plant soil remaining moist at all times?

Stick your finger into the soil up to your first knuckle and feel if there’s moisture present.” That is, “Is the plant soil always kept moist? Dip your fingers into the soil up to your first knuckles and feel for moisture. "

Preferably, please refer to FIG. 4 , the interactive question includes at least two alternative branches, and the answer to the interactive question is selected from at least two of the alternative branches. For the above example, two options: 1-Yes; 2-No can be provided, and the user will choose an answer from the two options. Corresponding to the answers of these two selection branches, if 1-Yes is selected, then A disease (leaves turn yellow-overwatering) is confirmed as disease result; if 2-No is selected, B disease (leaves turn yellow) - lack of water) is confirmed as a result of the condition.

Preferably, the interactive question may include at least two levels, and the different selection branches at the upper level correspond to different branch questions at the lower level. In some embodiments, the interactive question may include not only one level, but at least two levels. For example, if A is selected in the selection branch of the upper level, the branch question based on A will pop up further. , for example to pop up the branches of A1 and A2 to select branches. 2. If the user chooses answer B in the selection branch of the upper level, then the branch questions of the next level based on B will pop up, for example, the branch question selection branches of B1 and B2 will pop up. In this way, further auxiliary judgment information can be provided, so as to further narrow down the scope of the diagnosis result.

Preferably, each of the disease pre-identification results is associated with at least one of the interactive questions, or at least two of the disease pre-identification results are associated with at least one of the interactive questions. In some embodiments, each symptom pre-identification result can be associated with 1, 2 or more interactive questions, and the interactive questions can have 1, 2 or more levels, one or some of these levels Relational 1, 2 or more branching issues. For example, the pre-identification result of a certain disease is associated with 2 interactive questions: 1. Whether there is too much watering, 2. Whether there is too little light.

In some other embodiments, at least two disease pre-identification results are associated with at least one interactive question, it can be that at least two disease pre-identification results are associated with the same interactive question, or at least two disease pre-identification results are associated with 2 or more The interaction problem is not limited in this embodiment. For example, some two disease pre-identification results are associated with the same interactive question: whether there is too much watering; or some two disease pre-identification results are also associated with two interactive questions: 1. Is there too much watering, 2. Whether there is too little light.

Optionally, in step S1, before using the disease diagnosis engine to pre-identify the plant image, the method for diagnosing plant diseases and insect pests further includes: obtaining a species identification result corresponding to the plant image; After the engine pre-identifies the plant image, the pre-identification result of the disease is screened out according to the second preset condition.

In some cases, the purpose of uploading plant images by users may be mainly to identify plant species, and they do not actively want to diagnose plant diseases and insect pests, or identify plant species without knowing that plants have plant diseases and insect pests. At this time, pest diagnosis is in a passive diagnosis mode. The species recognition result can be obtained by using a species recognition engine to recognize the plant image. A pre-trained species recognition engine can be utilized to identify candidate species based on plant images. Wherein, the species recognition engine may include a neural network model, for example.

Generally speaking, in the passive diagnosis mode, the main purpose of the user is not to diagnose the disease itself, but, for example, to determine the species information of the plant and the like. In this case, only diseases with high accuracy and reliability can be output, so as to help users discover plant diseases in time and avoid additional troubles to users due to inaccurate output diseases. Therefore, in some embodiments, the pre-identification result of the disease can be screened out according to the second preset condition, so as to improve the accuracy rate.

When the plant image is identified by the species identification engine, one or more species identification results (for convenience of description, hereinafter referred to as candidate species) may be obtained. At this point, the disease diagnosis engine can obtain one or more disease pre-identification results for each candidate species (for ease of description, hereinafter referred to as candidate diseases), of course, it can only determine candidate diseases for some candidate species. It can be understood that the species identification result also has its own confidence level, and the output diagnostic information may be different according to the confidence level of the disease result.

For example, in a specific example, according to one or some plant images, the determined candidate species include species 1, species 2 and species 3. Species 1, 2 and 3 are arranged in descending order of species confidence. For example, Species 1 has a species confidence of 0.8, Species 2 has a species confidence of 0.75, and Species 3 has a species confidence of 0.7. In the passive diagnosis mode, for each species, the diseases that can be output are determined according to the second preset condition. When screening candidate diseases for a certain species, whether to screen out a certain candidate disease can be determined one by one according to the second preset condition according to the order of the diagnostic confidence of the candidate diseases of this species from high to low. For example, there are 3 candidate diseases corresponding to species 2 and 2 candidate diseases corresponding to species 3. If the diagnostic confidence for condition 2-1 is 95%, the diagnostic confidence for condition 2-2 is 90%, and the diagnostic confidence for condition 2-3 is 82%, then, in the process of screening for species 2 , according to the order of diagnosis confidence from high to low, according to the order of disease 2-1, disease 2-2 and disease 2-3 for screening. If after screening out the disease 2-1, no remaining disease can be found that can be used for output, then continue to screen out the disease 2-2. If the remaining symptoms that can be used for output have been found after screening the symptoms 2-1, then the screening of the symptoms 2-2 may not be continued, so as to simplify the whole processing process. Of course, in some cases, it is also possible that after the screening of all candidate diseases, there is still no remaining disease that can be used for output, then the screening is stopped, and no disease pre-identification results are output in the subsequent steps, then Neither step S2 nor step S3 will be executed, that is, no interactive question will be output according to the plant image at this time, so as to avoid confusion for the user.

In the screening process, the second preset condition involved may be related to various factors, such as the diagnostic confidence of the current disease diagnosis engine, the type of candidate species, the diagnostic accuracy of a certain type of disease, and the candidate One or more of the degree of match between the species and the candidate condition, etc.

In one embodiment, the second preset condition includes:

It is judged whether the candidate species is in the preset species white list, and when the candidate species is not in the preset species white list, the candidate disease corresponding to the candidate species is screened out.

The species included in the preset species white list are generally common species or important species, and the diagnosis of diseases of these species generally has relatively high accuracy and reliability. That is to say, in this embodiment, only diseases corresponding to these species may be output in subsequent steps without being screened out, so as to avoid outputting inaccurate, unreliable or unimportant diseases to users as much as possible, In order to avoid causing additional troubles to users.

In one embodiment, the second preset condition includes:

Comparing the diagnostic confidence of the candidate disease with a fourth preset value; when the diagnostic confidence of the candidate disease is less than the fourth preset value, the candidate disease is screened out.

Among them, the diagnostic confidence can represent the reliability of the disease pre-identification result obtained by the disease diagnosis engine in a single diagnosis process. In a specific example, the fourth preset value may be set to 70%. That is to say, when the diagnostic confidence of the candidate disease is less than 70%, the candidate disease will be screened out instead of being output, so as to avoid causing confusion to the user because the output interactive question does not match the actual situation.

In one embodiment, the second preset condition includes:

Comparing the diagnostic accuracy of the candidate disease with the preset accuracy; when the diagnostic accuracy of the candidate disease is less than the fifth preset value, the candidate disease is screened out.

Among them, diagnostic accuracy reflects the overall accuracy of identifying a particular type of disease. For example, the diagnostic accuracy can be obtained according to the ratio of the number of correct diagnoses to the total number of times in a certain total number of diagnoses. For some diseases that are difficult to diagnose, the diagnostic accuracy is often low, so by screening out candidate diseases related to these diseases, it is possible to avoid outputting inaccurate diseases as much as possible.

In one embodiment, the second preset condition includes:

It is judged whether the candidate species is in the candidate species blacklist corresponding to the disease, and if the candidate species is in the candidate species blacklist, the candidate disease is screened out.

For a particular species, it may not suffer from certain types of illnesses at all or only very rarely. Therefore, according to the mutually exclusive relationship between such species and diseases, the corresponding candidate species blacklist can be set in advance, so that candidate species can be screened out, so as to improve the accuracy and reliability of the output.

It should be noted that the above specific methods on how to screen out candidate diseases satisfying the second preset condition can be combined with each other. For example, in a specific example, as long as the candidate disease satisfies that the diagnostic confidence is less than the fourth preset value, the candidate species is not in the preset species whitelist, the diagnostic accuracy is less than the fifth preset value, and the candidate species is within the corresponding to the candidate disease Any condition in the candidate species blacklist, the candidate disease will be excluded.

Optionally, in an example, with the third preset value as the threshold, when the confidence of the species recognition result is not less than the third preset value, the species recognition result is considered to be of high confidence, and when the species recognition result is When the confidence level of is less than the third preset value, the species recognition result is considered as low confidence level. If the identification confidence of the species identification result is not less than a third preset value, the method for diagnosing plant diseases and insect pests further includes: displaying the species identification result and common diseases of the corresponding plant species. Please refer to FIG. 5 , in an example, when the recognition confidence of the species recognition result is not less than the third preset value, common diseases of the plant can be displayed. In the example shown in Figure 5, the first column 101 shows "Common diseases for Golden pothos based on our 10 millions of plant disease cases. (Golden pothos) the first 4 data of common diseases, each common disease also supports to view more.

In other cases, the user has preliminarily believed that the plant may have pests and diseases and sought confirmation, and the purpose of uploading plant images is mainly for the diagnosis of pests and diseases. At this time, pest diagnosis is in an active diagnosis mode. Optionally, the species identification result is extracted from the identified species identification result page. In these application scenarios, users may enter the diagnosis of plant diseases and insect pests from the interface of a certain plant identification result. At this time, the plant image can be considered to have species identification information, and further verification of the species identification information can be performed.

Furthermore, in the active diagnosis mode, since users subjectively want to diagnose pests and diseases, more candidate disease information can be included in the disease pre-identification results, and at the same time, the accuracy and accuracy of the generated disease results can be appropriately reduced. reliability requirements. Specifically, the second preset condition can also be appropriately relaxed accordingly, so as to reduce the candidate diseases that are screened out.

This embodiment also provides a readable storage medium on which a program is stored, and when the program is executed, the method for diagnosing plant diseases and insect pests as described above is realized. Further, this embodiment also provides a system for diagnosing plant diseases and insect pests, which includes a processor and a memory, and a program is stored in the memory, and when the program is executed by the processor, the above-mentioned diagnosis of plant diseases and insect pests is realized. method.

In summary, in the method for diagnosing plant diseases and insect pests, the system for diagnosing plant diseases and insect pests, and the readable storage medium provided by the present invention, the method for diagnosing plant diseases and insect pests includes: acquiring an image of a plant to be diagnosed, and using a disease diagnosis engine to analyze the image of the plant Perform pre-identification to obtain a disease pre-identification result; if the confidence of the disease pre-identification result is less than a first preset value, output an interactive question associated with the disease pre-identification result; and obtain an answer to the interactive question , and obtain the disease result of the plant image according to the answer, and output the diagnosis information. With such a configuration, by outputting the interactive question associated with the pre-identification result of the disease, the user selects the answer for processing, so that the disease result of the plant image is determined according to the answer, which is equivalent to the pre-identification result of the disease through additional input information. Further auxiliary confirmation effectively improves the diagnostic accuracy of plant diseases.

The above description is only a description of the preferred embodiments of the present invention, and does not limit the scope of the present invention. Any changes and modifications made by those of ordinary skill in the field of the present invention based on the above disclosures shall fall within the protection scope of the claims.

Claims

A method for diagnosing plant diseases and insect pests, comprising:

Obtaining a plant image to be diagnosed, and using a disease diagnosis engine to pre-identify the plant image to obtain a disease pre-identification result;

If the confidence of the disease pre-identification result is less than a first preset value, output an interactive question associated with the disease pre-identification result; and

The answer to the interactive question is obtained, and the disease result of the plant image is obtained according to the answer, and the diagnosis information corresponding to the disease result is output.
The method for diagnosing plant diseases and insect pests according to claim 1, wherein the interactive question includes at least two options, and the answer to the interactive question is selected from at least two options.
The method for diagnosing plant diseases and insect pests according to claim 2, wherein the interactive question includes at least two levels, and the different selected branches of the upper level correspond to different branch questions of the lower level.
The method for diagnosing plant diseases and insect pests according to claim 1, wherein the pre-identification results of the disease are at least two.
The method for diagnosing plant diseases and insect pests according to claim 1, wherein each of the pre-identification results of the disease is associated with at least one interactive question, or at least two of the pre-identification results of the disease are associated with at least one interactive question.
The method for diagnosing plant diseases and insect pests according to claim 1, wherein, if the confidence of the disease result is not less than a second preset value, the diagnostic information includes a reference map and/or a disease block diagram, and the reference map is A preset image corresponding to the disease result, the disease block diagram showing at least part of the plant image including the disease area, and marking the disease area.
The method for diagnosing plant diseases and insect pests according to claim 1, wherein, if the confidence of the disease result is not less than a second preset value, the diagnostic information includes diagnostic summary data and diagnostic detailed data.
The method for diagnosing plant diseases and insect pests according to claim 1, wherein if the confidence of the disease result is less than a second preset value, and/or, the image of the plant to be diagnosed does not meet the first preset condition, Then the diagnostic information includes diagnostic summary data and prompts to reacquire the plant image.
The method for diagnosing plant diseases and insect pests according to claim 1, wherein, before using the disease diagnosis engine to pre-identify the plant image, the method for diagnosing plant diseases and insect pests further comprises: acquiring the species corresponding to the plant image recognition result;

Furthermore, after the step of pre-identifying the plant image by using the disease diagnosis engine, the disease pre-identification result is screened out according to the second preset condition.
The method for diagnosing plant diseases and insect pests according to claim 9, wherein if the recognition confidence of the species identification result is not less than a third preset value, the method for diagnosing plant diseases and insect pests further comprises: displaying the species identification results and Common disorders of the plant species to which it corresponds.
The method for diagnosing plant diseases and insect pests according to claim 9, wherein the species identification result is obtained by identifying the plant image using a species identification engine, or the species identification result is obtained from the identified species identification result extracted from the page.
A readable storage medium on which a program is stored, wherein the method for diagnosing plant diseases and insect pests according to any one of claims 1-11 is realized when the program is executed.
A system for diagnosing plant diseases and insect pests, characterized in that it includes a processor and a memory, and a program is stored in the memory, and when the program is executed by the processor, the method according to any one of claims 1 to 11 is realized. Diagnostic methods for plant pests and diseases.