Analysis method and system based on artificial intelligence

Abstract

The invention discloses an artificial intelligence-based analysis method and system, and relates to the technical field of artificial intelligence, and the artificial intelligence-based analysis method comprises the following steps: S1, receiving an analysis request, analyzing the analysis request, obtaining a plurality of pieces of correction push data, and sending the correction push data; s2, receiving correction selection data sent according to the multiple pieces of correction push data, taking recommended acquisition time in the correction selection data as verification acquisition time, and taking an allowable difference value in the correction selection data as a correction verification value; s3, acquiring actual growth data when verification acquisition time arrives, analyzing the actual growth data by using the to-be-analyzed abnormal data and the correction verification value to obtain a correction analysis result, and if the correction analysis result is failure, acquiring a new analysis request, and executing S1; and if the correction analysis result is successful, ending. The correction success rate of the abnormal plant growth condition is improved.

Description

Technical Field

[0001] The present application relates to the field of artificial intelligence technology, and in particular to an analysis method and system based on artificial intelligence. Background Art

[0002] Artificial Intelligence (AI) is a technical science that uses digital computers or machines controlled by digital computers to simulate, extend and expand human intelligence. It is a theory, method, technology and application system that can perceive the environment, acquire knowledge and use knowledge to obtain the best results.

[0003] Among them, the analysis method and system based on artificial intelligence is one of the important applications of artificial intelligence. With the increasing demand for health and comfort of living environment, many people are keen on plant cultivation, but most growers lack professional planting knowledge. Therefore, how to help people analyze abnormal conditions in the growth process of plants and push corresponding correction plans in time has become a complex and practical problem. However, the existing artificial intelligence analysis methods and systems have the following problems:

[0004] (1) Existing AI analysis methods and systems usually directly identify text descriptions or single images uploaded by growers, thereby providing a large number of messy similar images or a large number of messy network data. Growers need to find the required data from the large amount of messy data. Existing AI analysis methods and systems cannot accurately analyze plant varieties and abnormal plant growth conditions, nor can they directly push accurate correction plans to growers based on the analysis results of abnormal plant growth.

[0005] (2) Existing artificial intelligence analysis methods and systems are unable to timely verify the actual growth conditions of plants after correction according to the correction plan, and are unable to guarantee the success rate of correcting abnormal plant growth conditions. Summary of the invention

[0006] The purpose of the present application is to provide an analysis method and system based on artificial intelligence, which can accurately analyze plant varieties and abnormal plant growth conditions, accurately push correction plans based on the analysis results of abnormal plant growth, and timely verify the actual growth conditions of plants after correction according to the correction plan, thereby improving the success rate of correction of abnormal plant growth conditions.

[0007] To achieve the above-mentioned purpose, the present application provides an analysis method based on artificial intelligence, comprising the following steps: S1: receiving an analysis request, analyzing the analysis request, obtaining multiple correction push data, and sending them; wherein the analysis request at least includes: variety data to be analyzed, abnormal data to be analyzed, and request time; the variety data to be analyzed at least includes: plant name, and image data and / or video data when the plant grows normally; the abnormal data to be analyzed at least includes: text data describing abnormal plant growth, and image data and / or video data when the plant grows abnormally; S2: receiving correction selection data sent according to multiple correction push data, using the recommended acquisition time in the correction selection data as the verification acquisition time, and using the allowable difference value in the correction selection data as the correction verification value; S3: when the verification acquisition time is reached, obtaining actual growth data, and using the abnormal data to be analyzed and the correction verification value to analyze the actual growth data, to obtain a correction analysis result, wherein the correction analysis result is failure or success; if the correction analysis result is failure, obtaining a new analysis request and executing S1; if the correction analysis result is success, ending.

[0008] As above, the sub-steps of analyzing the analysis request and obtaining multiple correction push data are as follows: S11: determining the variety target data from multiple plant information data in the plant database according to the variety data to be analyzed in the analysis request; S12: determining the abnormal target data from multiple abnormal sub-data of the variety target data according to the abnormal data to be analyzed in the analysis request, and using the abnormal category code in the abnormal target data as the target code; S13: traversing the abnormal correction database according to the target code, and using the abnormal correction data whose abnormal category code is consistent with the target code as the abnormal correction target data; S14: processing the image data and / or video data when the plant growth in the abnormal data to be analyzed is abnormal, and obtaining the current abnormality degree; S15: using the correction sub-data corresponding to the abnormality degree range containing the current abnormality degree in the abnormal correction target data as the abnormal correction target sub-data, and using the multiple correction schemes in the correction target sub-data as the correction push data.

[0009] As above, the sub-steps of determining the variety target data from multiple plant information data in the plant database according to the variety data to be analyzed in the analysis request are as follows: S111: extracting features from the variety data to be analyzed in the analysis request, obtaining a variety feature set to be analyzed, and using the variety feature set to be analyzed and the plant name as the current variety data; S112: accessing the plant database, using the current variety data to analyze the standard variety data of each plant information data respectively, to obtain the plant variety attribution value; S113: if all plant variety attribution values ​​are less than or equal to the preset plant variety attribution threshold, a manual analysis instruction is generated; if there is one or more plant variety attribution values ​​greater than the preset plant variety attribution threshold, the plant information data corresponding to the maximum value of all plant variety attribution values ​​is used as the variety target data.

[0010] As above, wherein, according to the abnormal data to be analyzed in the analysis request, the abnormal target data is determined from multiple abnormal sub-data of the variety target data, and the abnormal category code in the abnormal target data is used as the target code. The sub-steps are as follows: S121: feature extraction is performed on the text data describing the plant growth abnormality in the abnormal data to be analyzed, and a text abnormal feature set to be analyzed is obtained; S122: feature extraction is performed on the image data and / or video data when the plant growth is abnormal in the abnormal data to be analyzed, and a set of image abnormal features to be analyzed is obtained; S123: each abnormal sub-data in the variety target data is analyzed using the text abnormal feature set to be analyzed and the image abnormal feature set to be analyzed, respectively, to obtain a plant abnormality attribution value; S124: if all the plant abnormality attribution values ​​are less than or equal to a preset plant abnormality attribution threshold, a manual analysis instruction is generated; if there is one or more plant abnormality attribution values ​​greater than the preset plant abnormality attribution threshold, the abnormal sub-data corresponding to the maximum value of all the plant abnormality attribution values ​​is used as the abnormal target data, and the abnormal category code in the abnormal target data is used as the target code.

[0011] As above, the image data and / or video data when the plant growth is abnormal in the abnormal data to be analyzed are processed, and the sub-steps for obtaining the current abnormality degree are as follows: S141: each image in the image data when the plant growth is abnormal and / or each image in the video data is divided into regions to obtain abnormal growth regions and normal growth regions; S142: the sub-abnormality degree of each image is calculated based on the abnormal growth region and the normal growth region; S143: the current abnormality degree is obtained based on all the sub-abnormality degrees.

[0012] As above, the expression of the current abnormality level is as follows:

[0013] Among them, Pdq is the current abnormality level; Pzy kis the sub-abnormality degree of the k-th image, k∈[1,K], K is the total number of sub-abnormality degrees; Pzy min is the minimum value among the K sub-abnormal degrees; Pzy max is the maximum value among the K sub-abnormality levels.

[0014] As above, the sub-steps of analyzing the actual growth data using the abnormal data to be analyzed and the correction check value to obtain the correction analysis result are as follows: S31: Processing the actual growth data to obtain the actual abnormality degree; S32: Analyzing the current abnormality degree and the actual abnormality degree obtained according to the abnormal data to be analyzed to obtain the difference ratio; S33: Using the correction check value to judge the difference ratio to generate the correction analysis result; if the difference ratio is less than or equal to the correction check value, the generated correction analysis result is successful; if the difference ratio is greater than the correction check value, the generated correction analysis result is a failure.

[0015] As above, the sub-steps of processing the actual growth data to obtain the actual abnormality degree are as follows: S311: dividing each image in the actual growth data into regions to obtain the actual growth abnormality region and the actual growth normal region; S312: calculating the actual sub-abnormality degree of each image according to the actual growth abnormality region and the actual growth normal region; S313: obtaining the actual abnormality degree according to all the actual sub-abnormality degrees; wherein the expression of the actual abnormality degree is as follows:

[0016] Among them, Psq is the actual abnormality degree; Pxz h is the actual sub-abnormality level of the h-th image, h∈[1,H], H is the total number of actual sub-abnormality levels; Pxz min is the minimum value among the H actual sub-abnormal degrees; Pxz max is the maximum value among the H actual sub-abnormal degrees.

[0017] As above, where the expression of the difference ratio is as follows: Among them, Bzb is the difference ratio; Pdq is the current abnormality level; Psq is the actual abnormality level.

[0018] The present application also provides an analysis system based on artificial intelligence, including: a service platform and an artificial intelligence analysis center; wherein the service platform: sends an analysis request and receives multiple correction push data; selects one from the multiple correction push data as correction selection data and sends it; when the verification collection time arrives, collects actual growth data and sends it; the artificial intelligence analysis center: is used to execute the above-mentioned artificial intelligence-based analysis method.

[0019] The artificial intelligence-based analysis method and system of the present application can accurately analyze plant varieties and abnormal plant growth conditions, can accurately push correction plans based on the analysis results of abnormal plant growth, and can timely verify the actual growth conditions of plants after correction according to the correction plan, thereby improving the success rate of correction of abnormal plant growth conditions. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present application 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 recorded in the present application. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0021] Figure 1 A schematic diagram of the structure of an embodiment of an analysis system based on artificial intelligence;

[0022] Figure 2 The present invention is a flowchart of an embodiment of an analysis method based on artificial intelligence. DETAILED DESCRIPTION

[0023] The following is a clear and complete description of the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are 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 those skilled in the art without creative work are within the scope of protection of the present invention.

[0024] like Figure 1 As shown, the present application provides an analysis system based on artificial intelligence, including: a service platform 1 and an artificial intelligence analysis center 2.

[0025] Among them, service platform 1: sends an analysis request and receives multiple correction push data; selects one from the multiple correction push data as correction selection data and sends it; when the verification collection time arrives, collects actual growth data and sends it.

[0026] Artificial Intelligence Analysis Center 2: used to execute the following artificial intelligence-based analysis methods.

[0027] Furthermore, the artificial intelligence analysis center 2 includes at least: an analysis unit, a verification unit and a storage unit.

[0028] Among them, the analysis unit: receives the analysis request, analyzes the analysis request, and obtains multiple correction push data.

[0029] Verification unit: receiving correction selection data sent according to multiple correction push data, taking the recommended collection time in the correction selection data as the verification collection time, and taking the allowable difference value in the correction selection data as the correction verification value; when the verification collection time arrives, obtaining the actual growth data, and using the abnormal data to be analyzed and the correction verification value to analyze the actual growth data, to obtain the correction analysis result.

[0030] Storage unit: used to store plant database and abnormality correction database.

[0031] The plant database includes: multiple plant information data, one plant information data corresponds to one standard variety data, the standard variety data includes: a plant name set and a plant variety feature set; the plant information data includes: multiple abnormal sub-data, each abnormal sub-data includes: a text abnormal feature set, an image abnormal feature set and an abnormal category code. The abnormal category code of each abnormal sub-data is different.

[0032] Specifically, the plant name collection includes a variety of plant names.

[0033] The same plant has different names in different regions and / or different groups of people. A different name corresponds to a recorded name of a plant. For example, potato, sweet potato, ground pear, Irish yam, potato, yam egg, Solanum tuberosum L., etc. are all different names for the same plant.

[0034] The plant variety feature set includes: a plurality of image features that can clearly and accurately characterize the plant variety to which the plant belongs.

[0035] The text abnormality feature set includes: a plurality of text features that can clearly and accurately characterize the abnormal growth conditions of the plant of this type.

[0036] The image abnormality feature set includes: a plurality of image features that can clearly and accurately characterize the abnormal growth conditions of the plant of this type.

[0037] The abnormality correction database includes: multiple abnormality correction data, one abnormality correction data corresponds to an abnormality category code, the abnormality correction data includes: multiple correction sub-data, one correction sub-data corresponds to an abnormality degree range, the correction sub-data includes: multiple correction schemes, each correction scheme corresponds to an allowable difference value and a recommended collection time.

[0038] Specifically, the abnormal category code of each abnormal correction data is different. Different correction sub-data corresponds to a type of abnormal situation that needs to be corrected. Since there are multiple solutions to the same type of abnormal situation that needs to be corrected, different correction schemes in the same correction sub-data are different solutions to the same type of abnormal situation that needs to be corrected, which can provide more and richer choices for growers who make analysis requests.

[0039] The correction plan at least includes: the abnormal factors that need to be corrected, and the specific correction method for each abnormal factor. The specific correction method is obtained based on research results and / or actual experience.

[0040] For example: when the correction plan is for pests and diseases, the abnormal factors that need to be corrected are specific pest categories, and the specific correction methods are: use corresponding drugs to eliminate pests, recommend the name of an insecticide, the number of times the insecticide is used, the interval between each use, and the amount of medicine used each time. Among them, different correction sub-data correspond to abnormal conditions caused by a type of pests and diseases. Different correction plans within the same correction sub-data are different solutions to abnormal conditions caused by the same type of pests and diseases, for example, different drugs are used, different specific means are used, etc.

[0041] When the correction plan is for nutrient deficiency, the abnormal factor that needs to be corrected is the specific missing nutrient category, and the specific correction method is: use the corresponding fertilizer, recommend the name of a fertilizer, the number of times the fertilizer is used, the interval between each use, and the amount of fertilizer used each time.

[0042] When the correction plan is for inappropriate temperature, the abnormal factor to be corrected is temperature, and the specific correction method is: use corresponding instruments or methods to adjust the temperature within an appropriate temperature range.

[0043] Each correction scheme corresponds to an allowable difference value and a recommended collection time. Due to differences in the environment and the growth status of different plants, under the correction of the same correction scheme, the growth conditions of different plants are different, and the results obtained after correction are also different. Therefore, the actual growth data of the plants corrected by the correction scheme obtained at the recommended collection time are also different. The allowable difference value is the maximum value of the difference between the actual growth data of the plants corrected by the correction scheme obtained at the recommended collection time and the actual growth data of the plants before correction, that is, the degree of abnormal growth of the plants after correction / the degree of abnormal growth of the plants before correction.

[0044] The recommended collection time is the average time it takes for plants corrected with the correction scheme to recover from abnormal growth to normal growth, obtained from multiple sets of actual data based on research results or actual experience.

[0045] Furthermore, the stored plant database and / or the abnormal correction database are updated according to the update conditions, wherein the update conditions at least include: reaching a preset update time node, new data appearing, and finding errors in the original data.

[0046] like Figure 2 As shown, the present application provides an analysis method based on artificial intelligence, comprising the following steps:

[0047] S1: receiving an analysis request, analyzing the analysis request, obtaining a plurality of correction push data, and sending the correction push data; wherein the analysis request at least includes: variety data to be analyzed, abnormal data to be analyzed, and request time; the variety data to be analyzed at least includes: plant name, and image data and / or video data when the plant grows normally; the abnormal data to be analyzed at least includes: text data describing abnormal plant growth, and image data and / or video data when the plant grows abnormally.

[0048] Specifically, the plant name is the name of the plant that needs to be analyzed and is filled in through the service platform.

[0049] The request time is the time when the service platform sends the analysis request to the artificial intelligence analysis center.

[0050] Further, the sub-steps of analyzing the analysis request and obtaining multiple correction push data are as follows:

[0051] S11: Determine variety target data from a plurality of plant information data in a plant database according to the variety data to be analyzed in the analysis request.

[0052] Further, the sub-step of determining the variety target data from the plurality of plant information data in the plant database according to the variety data to be analyzed in the analysis request is as follows:

[0053] S111: Extract features of the variety data to be analyzed in the analysis request to obtain a feature set of the variety to be analyzed, and use the feature set of the variety to be analyzed and the plant name as current variety data.

[0054] Specifically, feature extraction is performed on the data of the variety to be analyzed in the analysis request by using existing technology or a pre-trained artificial intelligence-based feature extraction module to obtain multiple features of the variety to be analyzed, and the multiple features of the variety to be analyzed constitute a set of features of the variety to be analyzed. The features of the variety to be analyzed are image features that can clearly and accurately characterize the plant variety to which the plant belongs.

[0055] S112: Accessing the plant database, using the current variety data to analyze the standard variety data of each plant information data, and obtaining the plant variety affiliation value.

[0056] Furthermore, the expression of the plant variety attribute value is:

[0057] G JZ i =λ1·Zmc i +λ2·Ztz i ;

[0058] Among them, Gpz i is the plant variety attribution value between the current variety data and the standard variety data of the i-th plant information data; Zmc i is the name association value. When the plant name set in the standard variety data of the i-th plant information data contains a plant name that is consistent with the plant name in the current variety data, Zmc i =A1, when there is no plant name in the plant name set in the standard variety data of the i-th plant information data that is consistent with the plant name in the current variety data, Zmc i =A2, and A1>A2; Ztz i is the variety feature similarity value between the plant variety feature set of the i-th plant information data and the variety feature set to be analyzed in the current variety data; λ1 is the weight of the name association; λ2 is the weight of the variety feature similarity.

[0059] Specifically, A1 and A2 are both preset fixed values, and the specific values ​​of A1 and A2 are set according to the actual situation. The specific values ​​of λ1 and λ2 are set according to the actual situation.

[0060] The variety feature similarity value between the plant variety feature set of the i-th plant information data and the variety feature set to be analyzed of the current variety data can be obtained through existing technology or a pre-trained artificial intelligence-based analysis model. The larger the variety feature similarity value, the more consistent the plant variety features corresponding to the plant variety feature set and the variety feature set to be analyzed.

[0061] Since there may be omissions in the plant name collection, there may be errors in the plant names entered in the analysis request through the service platform, and there may be certain similarities in the image data of different species of plants at different growth stages, the plant variety attribution value Gpz i The name relevance and variety characteristics similarity are comprehensively considered in the analysis, which can improve the accuracy of plant variety analysis. i The larger the value is, the more consistent the plant variety corresponding to the current variety data is with the plant variety corresponding to the plant information data.

[0062] S113: If all plant variety attributes are less than or equal to the preset plant variety attributes threshold, a manual analysis instruction is generated; if there are one or more plant variety attributes values ​​greater than the preset plant variety attributes threshold, the plant information data corresponding to the maximum value of all plant variety attributes values ​​is used as variety target data.

[0063] Specifically, the specific value of the preset plant variety attribution threshold is set according to the actual situation. If the plant variety attribution value is less than or equal to the preset plant variety attribution threshold, it means that the plant variety corresponding to the plant information data is different from the plant variety corresponding to the current variety data. If the plant variety attribution value is greater than the preset plant variety attribution threshold, it means that the plant variety corresponding to the plant information data is the same as the plant variety corresponding to the current variety data.

[0064] The manual analysis instruction at least includes: an execution object and an analysis request, and the execution object performs manual analysis on the analysis request. The execution object is an authoritative expert or staff with relevant professional knowledge.

[0065] S12: determining abnormal target data from a plurality of abnormal sub-data of the variety target data according to the abnormal data to be analyzed in the analysis request, and using the abnormal category code in the abnormal target data as the target code.

[0066] Further, the sub-steps of determining abnormal target data from multiple abnormal sub-data of the variety target data according to the abnormal data to be analyzed in the analysis request, and using the abnormal category code in the abnormal target data as the target code are as follows:

[0067] S121: Extract features from text data describing abnormal plant growth in the abnormal data to be analyzed, and obtain a set of abnormal features of the text to be analyzed.

[0068] Specifically, feature extraction is performed on text data describing abnormal plant growth in abnormal data to be analyzed by using existing technology or pre-trained artificial intelligence-based feature extraction modules to obtain multiple abnormal text features to be analyzed, and the multiple abnormal text features to be analyzed constitute a set of abnormal text features to be analyzed. The abnormal text features to be analyzed are text features that can clearly and accurately characterize the abnormal growth conditions of such plants.

[0069] S122: Extract features from the image data and / or video data of abnormal plant growth in the abnormal data to be analyzed, and obtain an abnormal feature set of the image to be analyzed.

[0070] Specifically, feature extraction is performed on the image data and / or video data of abnormal plant growth in the abnormal data to be analyzed by using the existing technology or a pre-trained artificial intelligence-based feature extraction module to obtain multiple abnormal image features to be analyzed, and the multiple abnormal image features to be analyzed constitute a set of abnormal image features to be analyzed. The abnormal image features to be analyzed are image features that can clearly and accurately characterize the abnormal growth conditions of such plants.

[0071] S123: Analyze each abnormal sub-data in the variety target data using the abnormal feature set of the text to be analyzed and the abnormal feature set of the image to be analyzed, and obtain the plant abnormality attribution value.

[0072] Furthermore, the expression of plant abnormality attribution value is:

[0073] Gy j =η1·Zwb j +η2·Zyt j ;

[0074] Among them, Gyc j is the plant anomaly attribution value between the jth abnormal sub-data in the variety target data and the text anomaly feature set to be analyzed and the image anomaly feature set to be analyzed; η1 is the weight of the text anomaly similarity; Zwb j is the text feature similarity value between the text anomaly feature set in the jth anomaly sub-data and the text anomaly feature set to be analyzed; η2 is the weight of image anomaly similarity; Zyt j is the image feature similarity value between the image abnormality feature set in the jth abnormal sub-data and the image abnormality feature set to be analyzed.

[0075] Specifically, the specific values ​​of η1 and η2 are set according to actual conditions.

[0076] The similarity value between the text abnormality feature set in the jth abnormal sub-data and the text abnormality feature set to be analyzed can be obtained through the existing technology or a pre-trained artificial intelligence-based analysis model. The similarity value between the image abnormality feature set in the jth abnormal sub-data and the image abnormality feature set to be analyzed can be obtained through the existing technology or a pre-trained artificial intelligence-based analysis model.

[0077] Since the text data describing plant growth abnormalities in the analysis request input through the service platform may have description differences, errors or omissions, and the image data of different types of plant growth abnormalities may also have certain similarities, the plant abnormality attribution value Gyc j The text anomaly similarity and image anomaly similarity are comprehensively considered in the paper, which can improve the accuracy of the analysis of the abnormal type of plant growth. Plant anomaly attribution value Gycj The larger the value is, the more consistent the growth abnormality type corresponding to the abnormal data to be analyzed is with the growth abnormality type corresponding to the abnormal sub-data.

[0078] S124: If all plant abnormality attribution values ​​are less than or equal to the preset plant abnormality attribution threshold, a manual analysis instruction is generated; if there is one or more plant abnormality attribution values ​​greater than the preset plant abnormality attribution threshold, the abnormal sub-data corresponding to the maximum value of all plant abnormality attribution values ​​is used as abnormal target data, and the abnormal category code in the abnormal target data is used as the target code.

[0079] Specifically, the specific value of the preset plant abnormality attribution threshold is set according to the actual situation. If the plant abnormality attribution value is less than or equal to the preset plant abnormality attribution threshold, it means that the growth abnormality type corresponding to the abnormal data to be analyzed and the growth abnormality type corresponding to the abnormal sub-data are different. If the plant abnormality attribution value is greater than the preset plant abnormality attribution threshold, it means that the growth abnormality type corresponding to the abnormal data to be analyzed and the growth abnormality type corresponding to the abnormal sub-data are the same.

[0080] The manual analysis instruction at least includes: an execution object and an analysis request, and the execution object performs manual analysis on the analysis request. The execution object is an authoritative expert or staff with relevant professional knowledge.

[0081] S13: Traverse the abnormality correction database according to the target code, and use the abnormality correction data whose abnormality category code is consistent with the target code as abnormality correction target data.

[0082] S14: Processing the image data and / or video data of abnormal plant growth in the abnormal data to be analyzed to obtain the current abnormality degree.

[0083] Further, the image data and / or video data of abnormal plant growth in the abnormal data to be analyzed are processed to obtain the current abnormality level in the following sub-steps:

[0084] S141: Divide each image in the image data when the plant grows abnormally and / or each image in the video data into regions to obtain an abnormal growth region and a normal growth region.

[0085] Specifically, each image is divided into regions using existing image and / or video processing technology to obtain abnormal growth regions and normal growth regions. Alternatively, each image is divided into regions using a pre-trained artificial intelligence-based region division model to obtain abnormal growth regions and normal growth regions.

[0086] The abnormal growth area is: within the area where the plant is located in the image, the area where the abnormal growth of the plant can be displayed.

[0087] The normal growth area is: within the area where the plant is located in the image, the area where the plant can be shown to be growing normally.

[0088] S142: Calculate the sub-abnormality degree of each image according to the abnormal growth area and the normal growth area.

[0089] Furthermore, the expression of the sub-abnormality degree is as follows:

[0090]

[0091] Among them, Pzy k is the sub-abnormality degree of the kth image; Syc k is the area of ​​the abnormal growth region in the kth image; Szc k is the area of ​​the normal growth region in the kth image.

[0092] S143: Obtain the current abnormality level according to all sub-abnormality levels.

[0093] Furthermore, the expression of the current abnormality degree is as follows:

[0094]

[0095] Among them, Pdq is the current abnormality level; Pzy k is the sub-abnormality degree of the k-th image, k∈[1,K], K is the total number of sub-abnormality degrees; Pzy min is the minimum value among the K sub-abnormal degrees; Pzy max is the maximum value among the K sub-abnormality levels.

[0096] S15: taking the correction sub-data corresponding to the abnormality degree range including the current abnormality degree in the abnormality correction target data as the abnormality correction target sub-data, and taking the multiple correction schemes in the correction target sub-data as the correction push data.

[0097] S2: receiving correction selection data sent according to a plurality of correction push data, taking the recommended collection time in the correction selection data as the verification collection time, and taking the allowable difference value in the correction selection data as the correction verification value.

[0098] Specifically, after the artificial intelligence analysis center obtains multiple correction push data, it sends the correction push data to the service platform, which displays it. The service platform selects one from the multiple correction push data as correction selection data, and sends the correction selection data to the artificial intelligence analysis center.

[0099] S3: When the verification collection time is reached, the actual growth data is obtained, and the actual growth data is analyzed using the abnormal data to be analyzed and the correction verification value to obtain the correction analysis result, where the correction analysis result is failure or success; if the correction analysis result is failure, a new analysis request is obtained and S1 is executed; if the correction analysis result is success, the process ends.

[0100] Specifically, when the verification collection time is reached, the service platform sends the actual growth data to the artificial intelligence analysis center.

[0101] The actual growth data includes: image data and / or video data that can represent the current growth status of the plant.

[0102] Furthermore, the sub-steps of analyzing the actual growth data using the abnormal data to be analyzed and the correction check value to obtain the correction analysis result are as follows:

[0103] S31: Process the actual growth data to obtain the actual abnormality degree.

[0104] Furthermore, the sub-steps of processing the actual growth data to obtain the actual abnormality degree are as follows:

[0105] S311: Divide each image in the actual growth data into regions to obtain an actual growth abnormal region and an actual growth normal region.

[0106] Specifically, each image is divided into regions using existing image and / or video processing technology to obtain actual abnormal growth regions and actual normal growth regions. Alternatively, each image is divided into regions using a pre-trained artificial intelligence-based region division model to obtain actual abnormal growth regions and actual normal growth regions.

[0107] The actual abnormal growth area is: within the area where the plant is located in the image, the area where the abnormal growth of the plant can be displayed.

[0108] The actual normal growth area is: the area within the area where the plant is located in the image where the plant can be shown to be growing normally.

[0109] S312: Calculate the actual sub-abnormality degree of each image according to the actual abnormal growth area and the actual normal growth area.

[0110] Furthermore, the expression of the actual sub-abnormality degree is as follows:

[0111]

[0112] Among them, Pxz h is the actual abnormality level of the hth image; Myc his the area of ​​the actual abnormal growth region in the hth image; Mzc h is the area of ​​the actual normal growth region in the hth image.

[0113] S313: Obtaining the actual abnormality degree according to all the actual sub-abnormality degrees.

[0114] Furthermore, the expression of the actual abnormality degree is as follows:

[0115]

[0116] Among them, Psq is the actual abnormality degree; Pxz h is the actual sub-abnormality level of the h-th image, h∈[1,H], H is the total number of actual sub-abnormality levels; Pxz min is the minimum value among the H actual sub-abnormal degrees; Pxz max is the maximum value among the H actual sub-abnormal degrees.

[0117] S32: Analyze the current abnormality degree and the actual abnormality degree obtained according to the abnormal data to be analyzed to obtain a difference ratio.

[0118] Specifically, the specific method for obtaining the current abnormality degree according to the abnormal data to be analyzed is the same as steps S141-S143, or the current abnormality degree obtained in steps S141-S143 is directly used.

[0119] Furthermore, the expression of the difference ratio is as follows:

[0120]

[0121] Among them, Bzb is the difference ratio; Pdq is the current abnormality level; Psq is the actual abnormality level.

[0122] S33: Use the correction check value to judge the difference ratio and generate a correction analysis result; if the difference ratio is less than or equal to the correction check value, the generated correction analysis result is successful; if the difference ratio is greater than the correction check value, the generated correction analysis result is failed.

[0123] The artificial intelligence-based analysis method and system of the present application can accurately analyze plant varieties and abnormal plant growth conditions, can accurately push correction plans based on the analysis results of abnormal plant growth, and can timely verify the actual growth conditions of plants after correction according to the correction plan, thereby improving the success rate of correction of abnormal plant growth conditions.

[0124] Although the preferred embodiments of the present application have been described, those skilled in the art may make additional changes and modifications to these embodiments once they are aware of the basic creative concepts. Therefore, the scope of protection of the present application is intended to be interpreted as including the preferred embodiments and all changes and modifications that fall within the scope of the present application. Obviously, those skilled in the art may make various changes and modifications to the present application without departing from the spirit and scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the protection scope of the present application and its equivalent technology, the present application is also intended to include these modifications and variations.

Claims

1. An analysis method based on artificial intelligence, characterized in that: The steps include: S1: receiving an analysis request, analyzing the analysis request, obtaining multiple correction push data, and sending the data; wherein the analysis request at least includes: variety data to be analyzed, abnormal data to be analyzed, and request time; the variety data to be analyzed at least includes: plant name, and image data and / or video data when the plant grows normally; the abnormal data to be analyzed at least includes: text data describing abnormal plant growth, and image data and / or video data when the plant grows abnormally; S2: receiving correction selection data sent according to a plurality of correction push data, taking the recommended collection time in the correction selection data as the verification collection time, and taking the allowable difference value in the correction selection data as the correction verification value; S3: When the verification collection time is reached, the actual growth data is obtained, and the actual growth data is analyzed using the abnormal data to be analyzed and the correction verification value to obtain the correction analysis result, where the correction analysis result is failure or success; if the correction analysis result is failure, a new analysis request is obtained and S1 is executed; if the correction analysis result is success, the process ends.

2. The artificial intelligence-based analysis method according to claim 1, characterized in that: The sub-steps for analyzing the analysis request and obtaining multiple correction push data are as follows: S11: determining variety target data from a plurality of plant information data in a plant database according to the variety data to be analyzed in the analysis request; S12: determining abnormal target data from a plurality of abnormal sub-data of the variety target data according to the abnormal data to be analyzed in the analysis request, and using the abnormal category code in the abnormal target data as the target code; S13: traversing the abnormality correction database according to the target code, and taking the abnormality correction data whose abnormality category code is consistent with the target code as the abnormality correction target data; S14: Processing the image data and / or video data of abnormal plant growth in the abnormal data to be analyzed to obtain the current abnormality degree; S15: taking the correction sub-data corresponding to the abnormality degree range including the current abnormality degree in the abnormality correction target data as the abnormality correction target sub-data, and taking the multiple correction schemes in the correction target sub-data as the correction push data.

3. The artificial intelligence-based analysis method according to claim 2, characterized in that: The sub-steps of determining the variety target data from the multiple plant information data in the plant database according to the variety data to be analyzed in the analysis request are as follows: S111: extracting features from the variety data to be analyzed in the analysis request to obtain a feature set of the variety to be analyzed, and using the feature set of the variety to be analyzed and the plant name as current variety data; S112: Accessing the plant database, using the current variety data to analyze the standard variety data of each plant information data, and obtaining the plant variety attribution value; S113: If all plant variety attributes are less than or equal to the preset plant variety attributes threshold, a manual analysis instruction is generated; if there are one or more plant variety attributes values ​​greater than the preset plant variety attributes threshold, the plant information data corresponding to the maximum value of all plant variety attributes values ​​is used as variety target data.

4. The artificial intelligence-based analysis method according to claim 3, characterized in that: The sub-steps of determining abnormal target data from multiple abnormal sub-data of the variety target data according to the abnormal data to be analyzed in the analysis request, and using the abnormal category code in the abnormal target data as the target code are as follows: S121: extracting features from text data describing abnormal plant growth in the abnormal data to be analyzed, and obtaining a set of abnormal features of the text to be analyzed; S122: extracting features from the image data and / or video data of abnormal plant growth in the abnormal data to be analyzed, and obtaining an abnormal feature set of the image to be analyzed; S123: using the to-be-analyzed text abnormality feature set and the to-be-analyzed image abnormality feature set to analyze each abnormal sub-data in the variety target data, respectively, to obtain a plant abnormality attribution value; S124: If all plant abnormality attribution values ​​are less than or equal to the preset plant abnormality attribution threshold, a manual analysis instruction is generated; if there is one or more plant abnormality attribution values ​​greater than the preset plant abnormality attribution threshold, the abnormal sub-data corresponding to the maximum value of all plant abnormality attribution values ​​is used as abnormal target data, and the abnormal category code in the abnormal target data is used as the target code.

5. The artificial intelligence-based analysis method according to claim 4, characterized in that: The sub-steps of processing the image data and / or video data of abnormal plant growth in the abnormal data to be analyzed and obtaining the current abnormality degree are as follows: S141: dividing each image in the image data and / or each image in the video data when the plant grows abnormally into regions to obtain an abnormal growth region and a normal growth region; S142: Calculating the sub-abnormality degree of each image according to the abnormal growth area and the normal growth area; S143: Obtain the current abnormality level according to all sub-abnormality levels.

6. The artificial intelligence-based analysis method according to claim 5, characterized in that: The expression of the current abnormality level is as follows: Among them, Pdq is the current abnormality level; Pzy k is the sub-abnormality degree of the k-th image, k∈[1,K], K is the total number of sub-abnormality degrees; Pzy min is the minimum value among the K sub-abnormality levels; Pzy max is the maximum value among the K sub-abnormality levels.

7. The artificial intelligence-based analysis method according to claim 6, characterized in that: The sub-steps of analyzing the actual growth data using the abnormal data to be analyzed and the correction check value to obtain the correction analysis results are as follows: S31: Processing the actual growth data to obtain the actual abnormality degree; S32: Analyze the current abnormality degree and the actual abnormality degree obtained according to the abnormal data to be analyzed to obtain a difference ratio; S33: using the correction check value to judge the difference ratio and generate a correction analysis result; If the difference ratio is less than or equal to the correction check value, the generated correction analysis result is successful; if the difference ratio is greater than the correction check value, the generated correction analysis result is failed.

8. The artificial intelligence-based analysis method according to claim 7, characterized in that: The sub-steps for processing the actual growth data and obtaining the actual degree of abnormality are as follows: S311: performing region division on each image in the actual growth data to obtain an actual growth abnormal region and an actual growth normal region; S312: Calculating the actual sub-abnormality degree of each image according to the actual abnormal growth area and the actual normal growth area; S313: Obtaining an actual abnormality degree according to all actual sub-abnormality degrees; Among them, the expression of the actual abnormality degree is as follows: Among them, Psq is the actual abnormality degree; Pxz h is the actual sub-abnormality level of the h-th image, h∈[1,H], H is the total number of actual sub-abnormality levels; Pxz min is the minimum value among the H actual sub-abnormal degrees; Pxz max is the maximum value among the H actual sub-abnormal degrees.

9. The artificial intelligence-based analysis method according to claim 8, characterized in that: The expression of the difference ratio is as follows: Among them, Bzb is the difference ratio; Pdq is the current abnormality level; Psq is the actual abnormality level.

10. An analysis system based on artificial intelligence, characterized in that: include: Service platform and artificial intelligence analysis center; Among them, the service platform: sends an analysis request and receives multiple correction push data; selects one from the multiple correction push data as correction selection data and sends it; when the verification collection time arrives, collects actual growth data and sends it; Artificial intelligence analysis center: used to execute the artificial intelligence-based analysis method described in any one of claims 1-9.

Images