Intelligent analysis processing method for agricultural information

Abstract

The invention relates to an intelligent analysis processing method for agricultural information. The method comprises the following steps: acquiring a plant picture and problem information uploaded by a farmer; according to the problem information, identifying an analysis area in the plant picture; extracting a first feature of the analysis area and a second feature of the problem information; and performing intelligent analysis according to the first feature and the second feature to obtain an analysis result. According to the method provided by the invention, accurate agricultural information intelligent analysis can be carried out based on plant pictures and problem information uploaded by farmers, so that auxiliary decision-making reference is provided for work and life of users.

Description

Technical Field

[0001] The present invention relates to the field of computer technology, and in particular to an intelligent analysis and processing method for agricultural information. Background Art

[0002] Agricultural information is complex in nature, and there are great differences in agricultural information in different regions.

[0003] Currently, users are often required to analyze agricultural information based on their own knowledge reserves, but users' knowledge reserves are limited after all, and they are likely to obtain incorrect analysis and processing conclusions.

[0004] Therefore, how to conduct accurate intelligent analysis of agricultural information and provide auxiliary decision-making references for users' work and life has become a problem of great concern at present. Summary of the invention

[0005] 1. Technical issues to be resolved

[0006] In order to solve the above problems, the present invention provides an intelligent analysis and processing method for agricultural information.

[0007] (II) Technical solution

[0008] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0009] An intelligent analysis and processing method for agricultural information, the method comprising:

[0010] Obtain plant pictures and problem information uploaded by farmers;

[0011] Based on the question information, identify the analysis area in the plant image;

[0012] Extracting a first feature of the analysis area and a second feature of the problem information;

[0013] An intelligent analysis is performed based on the first feature and the second feature to obtain an analysis result.

[0014] Optionally, based on the question information, an analysis area in the plant image is identified, including:

[0015] Perform semantic recognition on the question information to obtain the first keyword of the question information;

[0016] Determine a synonym of the first keyword, and form a first keyword set with the first keyword and the synonym of the first keyword;

[0017] Segment the question information, remove invalid segmentations, and form a segmentation set with the removed segmentations;

[0018] Determine the weight of each word in the word set according to the first keyword set;

[0019] Perform object recognition on plant pictures to obtain the objects of the plant pictures;

[0020] For any object, determine the second keyword of any object, determine the synonym of the second keyword, and form a second keyword set with the second keyword and the synonym of the second keyword; determine the analysis probability of any object according to the second keyword set, the first keyword set, the word set and the weight of each word;

[0021] An analysis region is determined from the object according to the analysis probability.

[0022] Optionally, determining the weight of each word in the word set according to the first keyword set includes:

[0023] For any word in the word set, determine the degree of synonymy between any word and each element in the first keyword set Among them, i is the word segmentation identifier, j is the element identifier in the first keyword set, is the degree of synonymy between participle i and element j in the first keyword set;

[0024] Determine the criticality of any word Among them, J is the total number of elements in the first keyword set, α j is the weight of element j in the first keyword set;

[0025] Determine the weight of any word Among them, H i is the left-right entropy of word i.

[0026] Optionally,

[0027] Among them, β j is the part-of-speech coefficient of element j in the first keyword set, sim j is the synonymy degree between element j and the first keyword in the first keyword set, The synonymy degree of the first keyword in the first keyword set is not less than sim j And the number of elements that are not the first keyword.

[0028] Optionally, if the element j in the first keyword set has the same part of speech as the first keyword, then β j = 1, if the element j in the first keyword set has a different part of speech from the first keyword, then is the number of different parts of speech of all elements in the first keyword set.

[0029] Optionally, determining the analysis probability of any object according to the second keyword set, the first keyword set, the word segmentation set, and the weight of each word segmentation includes:

[0030] The second keyword set of any object forms a second vector, and the first keyword set forms a first vector;

[0031] Calculate the distance d between the first vector and the second vector u ; Where u is the object identifier;

[0032] For any word in the word set, determine the degree of synonymy between any word and each element in the second keyword set Among them, i is the word segmentation identifier, k is the element identifier in the second keyword set, is the degree of synonymy between participle i and element k in the second keyword set;

[0033] Determine the degree of analysis of each segmentation for any object according to the degree of synonymy between each segmentation and each element in the second keyword set;

[0034] Determine the probability of analysis of any object Among them, i is the total number of elements in the word segmentation set, w i is the weight of word i, is the degree of analysis of word i for object u, γ u is the agricultural attribute value of object u.

[0035] Optionally, determining the degree of analysis of each segmentation for any object according to the degree of synonymy between each segmentation and each element in the second keyword set includes:

[0036] For any participle, determine the mean and standard deviation of the synonymy between any participle and each element in the second keyword set;

[0037] The analytical degree of any word for any object is determined as the quotient of the mean and the standard deviation.

[0038] Optionally, if the object u is agricultural, then γ u =1;

[0039] If object u is non-agricultural, then when object u is a foreground object, When object u is a background object,

[0040] Among them, n 3 is the total number of foreground objects, n 4 is the total number of objects in the plant image.

[0041] Optionally, determining an analysis area from the object according to the analysis probability includes:

[0042] The objects whose analysis probability is greater than a preset probability threshold are regarded as candidate objects;

[0043] Determine the analysis object from the candidate objects;

[0044] The area corresponding to the analysis object is determined as the analysis area.

[0045] Optionally, determining the analysis object from the candidate objects includes:

[0046] If there are candidate objects with agricultural attributes, the candidate objects with agricultural attributes are taken as analysis objects;

[0047] If there are no candidate objects with agricultural attributes, the mean analysis probability of the candidate objects is determined, and the candidate objects with analysis probabilities greater than the mean analysis probability are taken as analysis objects.

[0048] (III) Beneficial effects

[0049] The present invention relates to an intelligent analysis and processing method for agricultural information, the method comprising: obtaining plant pictures and problem information uploaded by farmers; identifying an analysis area in the plant picture according to the problem information; extracting a first feature of the analysis area and a second feature of the problem information; performing intelligent analysis according to the first feature and the second feature to obtain an analysis result. The method provided by the present invention can perform accurate intelligent analysis of agricultural information based on the plant pictures and problem information uploaded by farmers, thereby providing auxiliary decision-making references for users' work and life. BRIEF DESCRIPTION OF THE DRAWINGS

[0050] Figure 1 A flowchart of an intelligent analysis and processing method for agricultural information provided in one embodiment of the present application;

[0051] Figure 2 A schematic diagram of a plant picture provided in one embodiment of the present application;

[0052] Figure 3 A schematic diagram of another plant picture provided in an embodiment of the present application. DETAILED DESCRIPTION

[0053] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0054] At present, users are often required to analyze agricultural information based on their own knowledge reserves, but the user's knowledge reserves are limited after all, and it is very likely to get wrong analysis and processing conclusions. Therefore, how to conduct accurate intelligent analysis of agricultural information and provide auxiliary decision-making references for users' work and life has become a more concerned issue.

[0055] In order to solve the above problems, the present invention relates to an intelligent analysis and processing method for agricultural information, the method comprising: obtaining plant pictures and problem information uploaded by farmers; identifying the analysis area in the plant picture according to the problem information; extracting the first feature of the analysis area and the second feature of the problem information; performing intelligent analysis according to the first feature and the second feature to obtain the analysis result. The method provided by the present invention can perform accurate intelligent analysis of agricultural information based on the plant pictures and problem information uploaded by farmers, thereby providing auxiliary decision-making reference for users' work and life.

[0056] See also Figure 1 This embodiment provides an intelligent analysis and processing method for agricultural information, and its implementation process is as follows:

[0057] 101, obtain plant pictures and problem information uploaded by farmers.

[0058] When farmers need to perform intelligent analysis and processing of agricultural information, they will upload the plant pictures and question information they need to process to the device that executes the intelligent analysis and processing method of agricultural information through the human-machine interface. The intelligent analysis and processing method of agricultural information will obtain the plant pictures and question information uploaded by farmers through the human-machine interface on the device where it is located.

[0059] 102, identifying the analysis area in the plant image according to the question information.

[0060] Since the plant pictures are uploaded by farmers, the quality of the plant pictures collected by different farmers may be different. For example, if a farmer asks for information about a plant, the pictures he collects may all be of the plant. Figure 2 As shown in , it is also possible that most of the collected pictures are other irrelevant content, such as Figure 3 If we process the plant images directly, similar to Figure 3 The poor quality images shown will reduce the accuracy of the analysis and processing results. Therefore, in step 102, the analysis area in the plant image will be identified based on the problem information. The analysis area can directly reflect the farmer's problem, so that subsequent intelligent analysis based on the analysis area can obtain accurate analysis results.

[0061] The implementation process of step 102 is as follows:

[0062] 102-1, semantic recognition is performed on the question information to obtain the first keyword of the question information.

[0063] This step uses an existing semantic recognition method to perform semantic recognition on the question information to obtain the semantics of the question information, which is the first keyword.

[0064] 102-2, determine the synonyms of the first keyword, and form a first keyword set with the first keyword and the synonyms of the first keyword.

[0065] This step uses the existing synonym recognition method to obtain the synonyms of the first keyword. The first keyword can be effectively expanded through the synonyms of the first keyword, avoiding the incompleteness of subsequent processing caused by the one-sided use of a single first keyword.

[0066] The first keyword set includes the first keyword and its synonyms, which can describe the semantics of the problem information more comprehensively from multiple angles.

[0067] 102-3, segment the question information, remove invalid segmentations, and form a segmentation set with the removed segmentations.

[0068] This step segments the question information using the existing segmentation method.

[0069] In addition, invalid participles are participles that cannot represent actual meaning, such as auxiliary words, conjunctions, adverbs, interjections, prepositions, modal particles, punctuation marks, etc.

[0070] The final set of segmented words is the segmented words that can represent the actual meaning in the question information.

[0071] 102-4, determine the weight of each word in the word set according to the first keyword set.

[0072] Although the word set is the word that can represent the actual meaning in the question information, different word segments have different degrees of representation of the actual meaning, and different degrees of representation have different values ​​for intelligent analysis and processing. This step will use the weight of each word in the word set to represent the value of each word to the intelligent analysis and processing.

[0073] For example, for any word i in the word set, its weight w is determined by the following steps: i :

[0074] 1. Determine the degree of synonymy between any participle and each element in the first keyword set

[0075] Among them, i is the word segmentation identifier, j is the element identifier in the first keyword set, is the degree of synonymy between word i and element j in the first keyword set.

[0076] For example, when determining the synonyms of the first keyword, the degree of synonymy is calculated, and finally whether it is a synonym of the first keyword is determined based on the degree of synonymy. The degree of synonymy is also calculated using this method.

[0077] 2. Determine the criticality of any participle

[0078] Among them, J is the total number of elements in the first keyword set, α j is the weight of element j in the first keyword set,

[0079] Among them, β j is the part-of-speech coefficient of element j in the first keyword set, sim j is the synonymy degree between element j and the first keyword in the first keyword set, The synonymy degree of the first keyword in the first keyword set is not less than sim j And the number of elements that are not the first keyword.

[0080] If the element j has the same part of speech as the first keyword, then β j = 1, if the element j has a different part of speech from the first keyword, then is the number of different parts of speech of all elements in the first keyword set.

[0081] β j Characterizes the similarity of parts of speech. If the parts of speech are the same, then β j =1, if they are not the same, then the more elements with different parts of speech, the lower the degree of similarity of parts of speech. j It represents the degree of synonymy between element j and the first keyword. The larger the value, the more synonymous they are. It represents the number of other keywords in the first keyword set that are more synonymous than element j and are not the first keyword. The larger the value, the more keywords element j is similar to in the first keyword set. That is, if element j is used as the dividing point, the words in the first keyword set are more able to represent the first keyword, and the more important element j is in the first keyword set. Therefore, α j Characterizes the weight of element j in the first keyword set.

[0082] k i It represents the degree of change of the similarity between the segmentation i and each element in the first keyword set. The larger the value, the greater the degree of change, that is, the greater the fluctuation of the synonymy degree of each element in the first keyword set. The smaller the value, the smaller the degree of change, that is, the smaller the fluctuation of the synonymy degree of each element in the first keyword set.

[0083] Since the elements in the first keyword set are all synonyms, theoretically, the similarity between the key word and the synonym is not much different, so k iThe smaller it is, the smaller the synonymy between word segment i and the elements in the first keyword set fluctuates, so word segment i is more critical to the first keyword set.

[0084] 3. Determine the weight of any word

[0085] Among them, H i is the left-right entropy of word i.

[0086] The left-right entropy is calculated using the existing scheme. The left-right entropy can be used to measure the possibility that a segmentation i is an independent representation of the substantive meaning, which is the value of the segmentation i for intelligent analysis and processing. i The bigger it is, the greater the value.

[0087] k i It represents the criticality of word i to the first keyword set. The more critical the word, the greater its value for intelligent analysis and processing. i The smaller.

[0088] Therefore, the weight w i The larger the value, the greater the value of word segmentation for intelligent analysis and processing.

[0089] 102-5, performing object recognition on the plant picture to obtain the object of the plant picture.

[0090] In this step, existing image recognition methods can be used to identify all objects in the plant picture, and all objects are objects in the plant picture.

[0091] 102-6, for any object, determine the second keyword of any object, determine the synonyms of the second keyword, and form a second keyword set with the second keyword and the synonyms of the second keyword. Determine the analysis probability of any object according to the second keyword set, the first keyword set, the segmentation set and the weight of each segmentation.

[0092] The second keyword is also obtained based on semantics. For example, for any object u, the existing recognition method is used to recognize the object u to obtain the semantics of the object u (such as barley, sun, cloud, etc.), which is the second keyword.

[0093] After obtaining the second keyword, the synonyms of the second keyword are obtained by using the existing synonym recognition method. The second keyword can be effectively expanded through the synonyms of the second keyword, avoiding the incompleteness of subsequent processing caused by the one-sided use of a single second keyword.

[0094] After obtaining the second keyword set, the analysis probability of any object is determined according to the second keyword set, the first keyword set, the word set and the weight of each word. The specific analysis probability determination process is:

[0095] 1. The second keyword set of any object is formed into a second vector, and the first keyword set is formed into a first vector.

[0096] 2. Calculate the distance d between the first vector and the second vector u .

[0097] Where u is the object identifier.

[0098] The distance d here u This can be achieved through existing distance calculation methods, such as distance d u is the Euclidean distance between the first vector and the second vector.

[0099] 3. For any word in the word set, determine the degree of synonymy between any word and each element in the second keyword set

[0100] Among them, i is the word segmentation identifier, k is the element identifier in the second keyword set, is the degree of synonymy between the participle u and the element k in the second keyword set.

[0101] The calculation scheme of synonymy degree is similar to The calculation scheme is the same as that of The implementation plan can be used.

[0102] 4. Determine the degree of analysis of each segmentation for any object based on the degree of synonymy between each segmentation and each element in the second keyword set.

[0103] For example, for any segmented word, the mean and standard deviation of the synonymy between any segmented word and each element in the second keyword set are determined. The degree of analysis of any segmented word for any object is determined to be the quotient of the mean and the standard deviation.

[0104] The analysis degree represents the degree of abnormality between any segmented word and the second keyword set. The larger the value, the more abnormal it is. The lower the value of the second keyword set for any segmented word, the lower the analysis degree.

[0105] 5. Determine the probability of analysis of any object

[0106] Among them, I is the total number of elements in the word segmentation set, w i is the weight of word i, is the degree of analysis of word i for object u, γ u is the agricultural attribute value of object u.

[0107] If object u is agricultural, then γ u =1.

[0108] If object u is non-agricultural, then when object u is a foreground object, When object u is a background object,

[0109] Among them, n 3 is the total number of foreground objects, n 4 is the total number of objects in the plant image.

[0110] Agricultural attributes refer to whether the object is related to agriculture. For example, if the object is the sun, it is not related to agriculture. Non-agricultural attributes, such as plants, are related to agriculture and are agricultural attributes. A whitelist can be pre-defined for what is related to agriculture. As long as the objects meet the whitelist, they are all agricultural attributes.

[0111] The larger the agricultural attribute value, the more relevant the object is to agricultural analysis. u The larger the value, the more similar the object is to the first keyword set. The larger the value, the greater the analysis value of the word segmentation. Therefore, the analysis probability PS of the object u The larger it is, the more likely it is that the object directly reflects the farmers' problems and the higher the accuracy of the analysis of the object.

[0112] 102-7, determining an analysis area from the object according to the analysis probability.

[0113] The implementation process of this step is:

[0114] 1. Objects whose analysis probability is greater than the preset probability threshold are regarded as candidate objects.

[0115] The objects whose analysis probability is greater than the preset probability threshold are objects that can directly reflect the problems of farmers, and the higher the accuracy of their analysis.

[0116] 2. Determine the analysis object from the candidate objects.

[0117] For example, if there are candidate objects with agricultural attributes, the candidate objects with agricultural attributes are taken as analysis objects. If there are no candidate objects with agricultural attributes, the mean analysis probability of the candidate objects is determined, and the candidate objects with analysis probabilities greater than the mean analysis probability are taken as analysis objects.

[0118] That is, objects related to agriculture among candidate objects are determined as analysis objects. If there are no objects related to agriculture, objects with analysis probabilities exceeding the mean analysis probabilities of the analysis objects are determined as analysis objects.

[0119] 3. Determine the area corresponding to the analysis object as the analysis area.

[0120] For example, the area where the minimum enclosing rectangle of the analysis object is located is determined as the analysis area.

[0121] 103, extracting a first feature of the analysis area and a second feature of the problem information.

[0122] This step uses existing feature extraction methods to extract the features of the analysis area and the problem.

[0123] 104. Perform intelligent analysis based on the first feature and the second feature to obtain an analysis result.

[0124] This step uses existing methods to perform intelligent analysis on the first feature and the second feature. For example, an intelligent analysis model is pre-trained, and the first feature and the second feature are input into the model to obtain the final analysis result.

[0125] Since the analysis area is obtained based on the question information, the part of the plant picture that best reflects the farmers' inquiry needs can more accurately reflect the farmers' inquiry intentions based on the characteristics of this part and the characteristics of the question information provided by the farmers, so that the analysis results are more accurate.

[0126] This embodiment provides an intelligent analysis and processing method for agricultural information, which obtains plant pictures and problem information uploaded by farmers; identifies the analysis area in the plant picture according to the problem information; extracts the first feature of the analysis area and the second feature of the problem information; performs intelligent analysis according to the first feature and the second feature to obtain the analysis result. The method provided in this embodiment can perform accurate intelligent analysis of agricultural information based on the plant pictures and problem information uploaded by farmers, thereby providing auxiliary decision-making reference for users' work and life.

[0127] It should be clear that the present invention is not limited to the specific configuration and processing described above and shown in the figures. For the sake of simplicity, a detailed description of the known method is omitted here. In the above embodiments, several specific steps are described and shown as examples. However, the method process of the present invention is not limited to the specific steps described and shown, and those skilled in the art can make various changes, modifications and additions, or change the order between the steps after understanding the spirit of the present invention.

[0128] It should also be noted that the exemplary embodiments mentioned in the present invention describe some methods or systems based on a series of steps or devices. However, the present invention is not limited to the order of the above steps, that is, the steps can be performed in the order mentioned in the embodiments, or in a different order from the embodiments, or several steps can be performed simultaneously.

[0129] Finally, it should be noted that the above-described embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the present invention. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some or all of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the scope of the technical solutions of the embodiments of the present invention.

Claims

1. An intelligent analysis and processing method for agricultural information, characterized in that: The method comprises: Obtain plant pictures and problem information uploaded by farmers; According to the question information, identifying the analysis area in the plant image; Extracting a first feature of the analysis area and a second feature of the problem information; An intelligent analysis is performed based on the first feature and the second feature to obtain an analysis result.

2. The method according to claim 1, characterized in that The step of identifying the analysis area in the plant image according to the question information includes: Performing semantic recognition on the question information to obtain a first keyword of the question information; Determine a synonym of the first keyword, and form a first keyword set with the first keyword and the synonym of the first keyword; Segmenting the question information, removing invalid segmentations, and forming a segmentation set with the removed segmentations; Determining the weight of each word in the word set according to the first keyword set; Performing object recognition on the plant picture to obtain the object of the plant picture; For any object, determine a second keyword of the object, determine a synonym of the second keyword, and form a second keyword set with the second keyword and the synonym of the second keyword; determine the analysis probability of the object according to the second keyword set, the first keyword set, the segmentation set and the weight of each segmentation; An analysis region is determined from the object based on the analysis probability.

3. The method according to claim 2, characterized in that The step of determining the weight of each word in the word set according to the first keyword set includes: For any word in the word set, determine the degree of synonymy between the word and each element in the first keyword set Among them, i is the word segmentation identifier, j is the element identifier in the first keyword set, is the degree of synonymy between participle i and element j in the first keyword set; Determine the criticality of any of the above mentioned words Among them, J is the total number of elements in the first keyword set, α j is the weight of element j in the first keyword set; Determine the weight of any of the above words Among them, H i is the left-right entropy of word i.

4. The method according to claim 3, characterized in that Said Among them, β j is the part-of-speech coefficient of element j in the first keyword set, sim j is the synonymy degree between element j in the first keyword set and the first keyword, The synonymy degree of the first keyword in the first keyword set is not less than sim j And the number of elements that are not the first keyword.

5. The method according to claim 4, characterized in that If the element j in the first keyword set has the same part of speech as the first keyword, then β j = 1, if the element j in the first keyword set has a different part of speech from the first keyword, then is the number of different parts of speech of all elements in the first keyword set.

6. The method according to claim 2, characterized in that The determining the analysis probability of any object according to the second keyword set, the first keyword set, the word set and the weight of each word includes: The second keyword set of any one of the objects forms a second vector, and the first keyword set forms a first vector; Calculate the distance d between the first vector and the second vector u ; Where u is the object identifier; For any word in the word set, determine the degree of synonymy between the word and each element in the second keyword set Among them, k is the word segmentation identifier, k is the element identifier in the second keyword set, is the degree of synonymy between participle i and element k in the second keyword set; Determining the degree of analysis of each segmentation for the any one of the objects according to the degree of synonymy between each segmentation and each element in the second keyword set; Determine the probability of analysis of any of the objects Among them, I is the total number of elements in the word segmentation set, w i is the weight of word i, is the degree of analysis of word i for object u, γ u is the agricultural attribute value of object u.

7. The method according to claim 6, characterized in that Determining the degree of analysis of each segmentation for the any object according to the degree of synonymy between each segmentation and each element in the second keyword set includes: For any participle, determining the mean and standard deviation of the synonymy between the any participle and each element in the second keyword set; The analysis degree of any one of the word segments for any one of the objects is determined as the quotient of the mean and the standard deviation.

8. The method according to claim 6, characterized in that If object u is agricultural, then γ u =1; If object u is a non-agricultural attribute, then when object u is a foreground object, When object u is a background object, Among them, n 3 is the total number of foreground objects, n 4 is the total number of objects in the plant picture.

9. The method according to claim 2, characterized in that: Determining the analysis area from the object according to the analysis probability includes: The objects whose analysis probability is greater than a preset probability threshold are regarded as candidate objects; Determine the analysis object from the candidate objects; The area corresponding to the analysis object is determined as the analysis area.

10. The method according to claim 9, characterized in that The step of determining the analysis object from the candidate objects comprises: If there are candidate objects with agricultural attributes, the candidate objects with agricultural attributes are taken as analysis objects; If there are no candidate objects with agricultural attributes, the mean analysis probability of the candidate objects is determined, and the candidate objects with analysis probabilities greater than the mean analysis probability are taken as analysis objects.

Images