Pig health early warning system and method based on image recognition

Abstract

The present invention belongs to the field of image recognition technology, and specifically relates to a pig health early warning system and method based on image recognition. The system includes: a data acquisition unit, a data processing unit, a data coupling judgment unit and a data analysis unit; the data acquisition unit is configured to collect pig farm operation data and pig image data; the pig farm operation data at least includes: temperature, humidity, light intensity and number of microorganisms. The system not only uses image recognition to warn of pig health, but also uses the collected pig farm operation data for correction, which greatly improves the accuracy of pig health early warning; at the same time, the idea of ​​data grouping and coupling is used to make multiple judgments, and then the judgment is made from the overall perspective, which further improves the accuracy.

Description

Background Art

[0002] According to statistics, 84% of rural animals rely on themselves to treat their health, while 96% rely on vaccination for disease prevention. At present, there are two main ways to diagnose and treat animal health: one is to draw a "diagnosis result" based on experience based on clinical symptoms and related information and take preventive and treatment measures in a timely manner; the other is to achieve the purpose of diagnosis through high-tech means such as laboratories in combination with clinical symptoms. Although the second method has a high accuracy rate, most farms do not have the conditions and it takes a long time to diagnose, so it is not suitable for actual production. This method can only be used to diagnose the case and take corresponding measures based on the diagnosis results. It is widely used in highly contagious diseases. More than 90% of animal health diagnoses are based on clinical experience, of which 84% are done by oneself and 16% are done by veterinarians or other colleagues. Relying on others to diagnose and treat animal health often delays time due to various factors, easily misses the best time for diagnosis and treatment, and wastes a lot of manpower, material resources, and financial resources. The breeding industry is originally a high-risk, low-profit industry, and we should strive to reduce costs. Although relying on self-diagnosis and treatment is timely and economical, the diagnosis and treatment levels vary greatly, and the diagnosis effect is difficult to guarantee, which will cause immeasurable losses to production.

[0003] The patent number CN201810175030.4A discloses a method for constructing a medical image atlas based on image segmentation and convolutional neural network. The convolutional neural network method is used to perform fine segmentation of medical images, image segmentation and recognition are performed on the collected medical images, and medical knowledge units are segmented for medical diagnosis knowledge to form knowledge units of different medical image pictures; the image segmentation image relationship is analyzed to establish the connection between different image knowledge units and connect the additional medical disease knowledge units, and the knowledge base coupling relationship is established based on the connection between medical image recognition and medical diagnosis knowledge; the knowledge units after image segmentation are feature matched to form entities with similar attributes as entity nodes in the knowledge graph, and the knowledge base entities are integrated to complete entity alignment; the knowledge graph is established using the constructed entities and the coupling relationship between entities. It is based on image segmentation and knowledge base connection to perform health diagnosis. Although it can realize healthy intelligent diagnosis, its coupling and connection process has high requirements on system resources, and there is no correction of image recognition, which makes the accuracy rate low. Summary of the invention

[0004] In view of this, the main purpose of the present invention is to provide a pig health early warning system and method based on image recognition. The system not only uses image recognition to warn of pig health, but also uses the collected pig farm operation data for correction, thereby greatly improving the accuracy of pig health early warning; at the same time, the idea of ​​data grouping and coupling is used to make multiple judgments, and then to make judgments from the overall perspective, which further improves the accuracy.

[0005] To achieve the above object, the technical solution of the present invention is achieved as follows:

[0006] A pig health early warning system based on image recognition, the system comprises: a data acquisition unit, a data processing unit, a data coupling judgment unit and a data analysis unit; the data acquisition unit is configured to collect pig farm operation data and pig image data; the pig farm operation data at least includes: temperature, humidity, light intensity and microbial quantity; the data processing unit is configured to perform data preprocessing on the acquired pig farm operation data to obtain pig farm preprocessing data, and perform image preprocessing on the pig image data to obtain image preprocessing data; the data coupling judgment unit is configured to couple and judge the pig farm preprocessing data and the image preprocessing data based on a preset data coupling judgment result model to obtain a data coupling judgment result. if the data coupling judgment result determines that the pig farm preprocessing data and the image preprocessing data are coupled, the pig farm preprocessing data and the image preprocessing data are retained; if the data coupling judgment result determines that the pig farm preprocessing data and the image preprocessing data are not coupled, the pig farm preprocessing data and the image preprocessing data are discarded; the data analysis unit is configured to use a preset pig farm health data analysis model to perform data analysis on the pig farm preprocessing data to obtain a pig farm data analysis result, and at the same time use a preset pig health image analysis model to analyze the image preprocessing data to obtain an image data analysis result, and use the pig farm data analysis result to correct the image data analysis result to obtain a final early warning result.

[0007] Furthermore, the data processing unit includes a method for preprocessing the acquired pig farm operation data, including: setting a time interval, dividing the acquired humidity, temperature, light intensity and microorganism quantity data in chronological order according to the set time interval, to obtain multiple temperature data groups, humidity data groups, light intensity data groups and microorganism quantity data groups, and classifying the temperature data groups, humidity data groups, light intensity data groups and microorganism quantity data groups in the same time interval into one category to obtain multiple operation data categories, thereby completing the pig farm operation data preprocessing.

[0008] Furthermore, the data processing unit performs image preprocessing on the pig image data to obtain image preprocessing data, and the method includes: dividing the acquired image data according to the set time interval to obtain multiple image data groups, and completing the pig image data preprocessing.

[0009] Furthermore, the data coupling judgment unit performs coupling judgment on the pig farm preprocessing data and the image preprocessing data based on a preset data coupling judgment result model, and the method for obtaining the data coupling judgment result includes: grouping the operation data class and the image data in the same time interval and performing coupling judgment using a preset data coupling judgment model; the data coupling judgment model is expressed using the following formula: ;in, is the number of data in the running data class, is the number of pixel values ​​in the image data group, wherein the number of pixel values ​​is defined as the number of categories of unequal pixel values; Grouping image data; ; is the adjustment coefficient, and its value range is 3~6; the calculated coupling value is compared with the preset threshold range, and according to the comparison result, it is judged whether the pig farm preprocessing data and the image preprocessing data are coupled, so as to obtain the data coupling judgment result.

[0010] Furthermore, the data analysis unit uses a preset pig farm health data analysis model to perform data analysis on the pig farm preprocessing data to obtain the pig farm data analysis results. The method performs the following steps: Use the preset pig farm health data analysis to perform data analysis on each operating data class respectively, specifically including: unifying the data format of each data in the operating data class to obtain target data, and normalizing each data form in the target data according to attributes to generate a target matrix; selecting an attribute set for the target matrix through the pig farm health data analysis model, and screening out the target attribute set to form a sample matrix; selecting an attribute set for the sample matrix again through the pig farm health data analysis model to obtain sample characteristics, comparing the sample characteristics with the historical health sample characteristics, calculating the difference between the two, and using the difference as a correction value; the correction value is the final pig farm data analysis result.

[0011] Furthermore, the pig farm health data analysis model is expressed using the following formula: ;in, is the target data, expressed in matrix form, is the classification matrix, is the matrix transpose operation, is the order of the matrix, is the sample matrix.

[0012] Furthermore, the data analysis unit uses a preset pig health image analysis model to analyze the image preprocessing data to obtain the image data analysis result, and the method performs the following steps: first, use the pig health image analysis model to perform a first feature extraction on the image preprocessing data to obtain a first feature; then perform secondary feature extraction and secondary feature normalization extraction on the first feature respectively; based on the results of the secondary feature extraction and the secondary feature normalization extraction results, respectively compare the differences with the historical image sample features, and then take the average value according to the difference ratio between the two to obtain the image data analysis result.

[0013] Furthermore, the pig health image analysis model is expressed using the following formula: ;in, is the image preprocessing data or the first feature, is the result of secondary feature extraction or secondary feature normalization; t is the adjustment coefficient. When the value range of t is 1~5, is the result of secondary feature extraction. When the value of t is between 6 and 10, It is the extraction result of secondary feature normalization.

[0014] Furthermore, the method in which the data analysis unit uses the pig farm data analysis result to correct the image data analysis result includes: multiplying the image data analysis result with the pig farm data analysis result to obtain a correction result, thereby completing the correction of the image data analysis result.

[0015] A pig health early warning method based on image recognition, the method performs the following steps: step 1: collecting pig farm operation data and pig image data; the pig farm operation data at least includes: temperature, humidity, light intensity and microbial quantity; step 2: performing data preprocessing on the acquired pig farm operation data to obtain pig farm preprocessing data, and performing image preprocessing on the pig image data to obtain image preprocessing data; step 3: based on a preset data coupling judgment result model, coupling and judging the pig farm preprocessing data and the image preprocessing data to obtain a data coupling judgment result; if the data coupling judgment result judges that the pig farm preprocessing data If the data coupling judgment result determines that the pig farm preprocessing data and the image preprocessing data are coupled, the pig farm preprocessing data and the image preprocessing data are retained; if the data coupling judgment result determines that the pig farm preprocessing data and the image preprocessing data are not coupled, the pig farm preprocessing data and the image preprocessing data are discarded; Step 4: Use the preset pig farm health data analysis model to analyze the pig farm preprocessing data to obtain the pig farm data analysis results, and at the same time use the preset pig health image analysis model to analyze the image preprocessing data to obtain the image data analysis results, and use the pig farm data analysis results to correct the image data analysis results to obtain the final early warning results.

[0016] The pig health early warning system and method based on image recognition of the present invention has the following beneficial effects: the system not only uses image recognition to warn pig health, but also uses the collected pig farm operation data for correction, which greatly improves the accuracy of pig health early warning; at the same time, it uses the idea of ​​data grouping and coupling to make multiple judgments, and then makes judgments from the overall perspective, which further improves the accuracy. It is mainly achieved through the following processes: 1. Data coupling: In the prior art, it is often easy to misdiagnose when using a single category of data to warn animals about health; because many health conditions will not be reflected in certain specific data in certain situations, if multiple data cannot be combined for judgment, the accuracy will be reduced. The present invention can significantly improve the accuracy by combining multiple data for judgment. However, if multiple data are analyzed and judged, although the accuracy can be improved in theory, the types of data obtained increase, and the instruments or equipment for obtaining data may cause errors due to their own errors or the influence of environmental factors when obtaining data, thereby reducing the accuracy of the data. At this time, even if multiple types of data are used for analysis and judgment, the accuracy is still not ideal. Therefore, the present invention conducts coupling judgment on the data. The essence of coupling judgment is to judge whether two different data obtained at the same time or the same period are reliable, because in the present invention, image data is used to obtain the same type of data. 1. The value obtained in the same time period of the data and the pig farm operation data will be correlated, because the health status of the pigs will be reflected in the pig farm operation data and the pig image data. If the two data obtained do not meet this coupling relationship, it can be judged that the two sets of data cannot be used for pig health analysis and early warning. If this coupling relationship is met, it can be proved that the data reliability of the two meets the requirements, and the two sets of data can be used for subsequent health analysis and early warning, thereby improving the accuracy rate; 2. Image recognition based on multiple feature extraction: The image recognition using a single feature in the prior art often reduces the recognition accuracy because the single feature cannot fully reflect all the features of the image. However, the present invention performs feature extraction twice. When performing the second feature extraction, different feature extraction methods are used, and then the two different secondary features obtained are cross-compared to find the feature difference, and then feature matching recognition is performed based on this feature difference. Compared with the prior art, the computational complexity of this method is only manifested as the second feature extraction, but the process of direct image matching recognition is omitted. In the case of a small increase in complexity, the accuracy rate can be significantly improved. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] Figure 1 A schematic diagram of the system structure of a pig health early warning system based on image recognition provided by an embodiment of the present invention;

[0018] Figure 2 A schematic diagram of a method flow of a pig health early warning method based on image recognition provided by an embodiment of the present invention;

[0019] Figure 3 A structural schematic diagram of image feature extraction of a pig health early warning system and method based on image recognition provided by an embodiment of the present invention;

[0020] Figure 4 A schematic diagram showing the effect of the accuracy of the pig health early warning system and method based on image recognition provided by an embodiment of the present invention changing with the number of experiments. DETAILED DESCRIPTION

[0021] The method of the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments of the present invention.

[0022] Example 1

[0023] like Figure 1 As shown, a pig health early warning system based on image recognition, the system includes: a data acquisition unit, a data processing unit, a data coupling judgment unit and a data analysis unit; the data acquisition unit is configured to collect pig farm operation data and pig image data; the pig farm operation data at least includes: temperature, humidity, light intensity and microbial quantity; the data processing unit is configured to perform data preprocessing on the acquired pig farm operation data to obtain pig farm preprocessing data, and perform image preprocessing on the pig image data to obtain image preprocessing data; the data coupling judgment unit is configured to couple and judge the pig farm preprocessing data and the image preprocessing data based on a preset data coupling judgment result model to obtain data coupling Judgment result; if the data coupling judgment result determines that the pig farm preprocessing data and the image preprocessing data are coupled, the pig farm preprocessing data and the image preprocessing data are retained; if the data coupling judgment result determines that the pig farm preprocessing data and the image preprocessing data are not coupled, the pig farm preprocessing data and the image preprocessing data are discarded; the data analysis unit is configured to use a preset pig farm health data analysis model to perform data analysis on the pig farm preprocessing data to obtain a pig farm data analysis result, and at the same time use a preset pig health image analysis model to analyze the image preprocessing data to obtain an image data analysis result, and use the pig farm data analysis result to correct the image data analysis result to obtain a final early warning result.

[0024] refer to Figure 3 and Figure 4 In the prior art, a single method is generally used to carry out health warning for pigs. One method uses an image warning method to carry out health warning. This warning method is based on the appearance image of the pig. The collected image is identified and judged using a recognition algorithm based on a neural network, and a health warning is carried out based on the identification and judgment results.

[0025] Both methods use the operating data of pig farms to make judgments. Generally, they collect data such as temperature, humidity, and number of microorganisms in the pig farms for data analysis in order to make health warnings or predictions.

[0026] However, both of the above methods have certain defects. Although the use of a single judgment standard is more efficient because the processed data and algorithm complexity are lower, it can easily lead to misjudgment or false alarm and has a low accuracy rate.

[0027] Figure 3 In the present invention, two feature extractions are used. In the second feature extraction, different feature extraction methods are used, and then difference comparison is performed based on the results. Compared with the traditional image feature extraction recognition method, this method has a lower algorithm complexity because the subsequent recognition process is omitted and is implemented by difference comparison. At the same time, because feature difference comparison is used, it is easy to find out whether the two differences with the matching template meet the set situation. Based on this judgment method, it will be possible to improve efficiency while ensuring accuracy.

[0028] refer to Figure 4 , Figure 4 The results of the present invention showing that the accuracy rate of the present invention is continuously improved with the number of experiments. Because the present invention makes judgments based on the combination of operation data and image data, the accuracy rate can be significantly improved. Moreover, as the number of experiments increases, the accuracy rate will be further improved because more data is collected from operation data and image data.

[0029] Example 2

[0030] On the basis of the previous embodiment, the data processing unit, the method for preprocessing the acquired pig farm operation data includes: setting a time interval, dividing the acquired humidity, temperature, light intensity and microorganism quantity data in chronological order according to the set time interval, to obtain multiple temperature data groups, humidity data groups, light intensity data groups and microorganism quantity data groups, and classifying the temperature data groups, humidity data groups, light intensity data groups and microorganism quantity data groups in the same time interval into one category to obtain multiple operation data categories, thereby completing the pig farm operation data preprocessing.

[0031] Specifically, there are many types of pig diseases, including infectious diseases, parasitic diseases, internal diseases, surgical diseases, obstetric diseases, nutritional metabolic diseases and toxic health. The most serious harm is infectious diseases, which often occur in large numbers, with high morbidity and mortality rates, seriously affecting the development of the pig industry and causing huge economic losses. Whether pigs can get sick is closely related to the individual's natural non-specific resistance. Strengthening pig feeding management, paying attention to environmental hygiene, implementing strict veterinary health systems, and enhancing pig health and resistance to external pathogenic factors are important conditions for preventing infectious diseases.

[0032] At the same time, we should also pay attention to the cleanliness of feed and drinking water. Do not feed rotten, moldy and spoiled feed. Clean the pens regularly to keep them clean and dry. Keep warm in winter and cool in summer. Keep feeding troughs and management utensils clean. These are all important aspects of health prevention that cannot be ignored. They are also the basic conditions for ensuring the growth and development of pigs, as well as their strong physique and disease resistance.

[0033] Example 3

[0034] Based on the previous embodiment, the data processing unit performs image preprocessing on pig image data to obtain image preprocessing data, including: dividing the acquired image data according to the set time interval to obtain multiple image data groups, and completing the pig image data preprocessing.

[0035] Specifically, after the image data are grouped, the image data in each group will be coupled with the operation data class, and the coupled operation data will be able to reflect the internal connection between the two, so as to improve the accuracy of correction during subsequent correction.

[0036] Example 4

[0037] On the basis of the previous embodiment, the data coupling judgment unit couples the pig farm preprocessing data and the image preprocessing data based on the preset data coupling judgment result model, and the method for obtaining the data coupling judgment result includes: grouping the operation data class and the image data in the same time interval and using the preset data coupling judgment model to perform coupling judgment; the data coupling judgment model is expressed by the following formula: ;in, is the number of data in the running data class, is the number of pixel values ​​in the image data group, wherein the number of pixel values ​​is defined as the number of categories of unequal pixel values; Grouping image data; ; is the adjustment coefficient, and its value range is 3~6; the calculated coupling value is compared with the preset threshold range, and according to the comparison result, it is judged whether the pig farm preprocessing data and the image preprocessing data are coupled, so as to obtain the data coupling judgment result.

[0038] Specifically, in the existing technology, it is often easy to misdiagnose animals by using a single category of data for health warning; because many health conditions will not be reflected by certain specific data in certain situations, if multiple data cannot be combined for judgment, the accuracy will be reduced. The present invention can significantly improve the accuracy by combining multiple data for judgment. However, if multiple data are analyzed and judged, although the accuracy can be improved in theory, the types of data obtained increase, and the instruments or equipment for obtaining data may cause errors due to their own errors or the influence of environmental factors when obtaining data, thereby reducing the accuracy of the data. At this time, even if multiple types of data are used for analysis and judgment, the accuracy is not ideal; therefore, the present invention performs coupling judgment on the data. The essence of coupling judgment is to judge whether two different data obtained at the same time or in the same period are reliable, because in the present invention, the image data and the pig farm operation data are values ​​obtained in the same time period, and there will be a correlation between the two, because the health status of pigs will be reflected in the pig farm operation data and the pig image data. If the two data obtained do not meet this coupling relationship, it can be judged that the two sets of data cannot be used for pig health analysis and early warning. If this coupling relationship is met, it can be proved that the reliability of the two data meets the requirements, and the two sets of data can be used for subsequent health analysis and early warning, thereby improving the accuracy.

[0039] Specifically, the data coupling judgment result will be able to reflect the coupling of data. This coupling will be able to provide a basis for subsequent health warnings. Although when one type of data reflects that the pig is healthy, there will be false alarms. If the two are combined, as well as the data coupling of each other, it is easy to achieve a more accurate pig health warning.

[0040] Example 5

[0041] On the basis of the previous embodiment, the data analysis unit uses a preset pig farm health data analysis model to perform data analysis on the pig farm preprocessing data to obtain the pig farm data analysis results. The method performs the following steps: Use the preset pig farm health data analysis to perform data analysis on each operating data class respectively, specifically including: unifying the data format of each data in the operating data class to obtain target data, and normalizing each data form in the target data according to attributes to generate a target matrix; selecting an attribute set for the target matrix through the pig farm health data analysis model, and screening out the target attribute set to form a sample matrix; selecting an attribute set for the sample matrix again through the pig farm health data analysis model to obtain sample characteristics, compare the sample characteristics with the historical health sample characteristics, calculate the difference between the two, and use the difference as a correction value; the correction value is the final pig farm data analysis result.

[0042] Specifically, image features mainly include color features, texture features, shape features and spatial relationship features of the image.

[0043] Color feature is a global feature that describes the surface properties of the scene corresponding to an image or an image region; texture feature is also a global feature that describes the surface properties of the scene corresponding to an image or an image region; there are two ways to represent shape features, one is contour feature and the other is regional feature. The contour feature of an image is mainly aimed at the outer boundary of an object, while the regional feature of an image is related to the entire shape region; spatial relationship feature refers to the spatial position or relative direction relationship between multiple targets segmented from an image. These relationships can also be divided into connection / adjacency relationships, overlap / overlapping relationships, and inclusion / inclusion relationships.

[0044] Example 6

[0045] Based on the previous embodiment, the pig farm health data analysis model is expressed using the following formula: ;in, is the target data, expressed in matrix form, is the classification matrix, is the matrix transpose operation, is the order of the matrix, is the sample matrix.

[0046] Specifically, the "Classification Matrix" classifies all the cases in the model into different categories by determining whether the predicted value matches the actual value. All cases in each category are then counted and the totals are displayed in the matrix. The classification matrix is ​​a standard tool for evaluating statistical models and is sometimes called a "confusion matrix."

[0047] Each column of the confusion matrix represents the predicted category, and the total number of each column represents the number of data predicted to be in that category; each row represents the true category of the data, and the total number of data in each row represents the number of data instances in that category. The value in each column represents the number of true data predicted to be in that category. For example, 43 in the first column of the first row means that 43 instances that actually belong to the first category are predicted to be in the first category. Similarly, 2 in the second column of the first row means that 2 instances that actually belong to the first category are mistakenly predicted to be in the second category.

[0048] In the present invention, the actual value is the target data, and the predicted value is the historical data; the historical health sample characteristics in the present invention are obtained by extracting and analyzing the characteristics of the historical data.

[0049] Example 7

[0050] Based on the previous embodiment, the data analysis unit uses a preset pig health image analysis model to analyze the image preprocessing data, and the method for obtaining the image data analysis result performs the following steps: first, use the pig health image analysis model to perform a first feature extraction on the image preprocessing data to obtain a first feature; then perform secondary feature extraction and secondary feature normalization extraction on the first feature respectively; based on the results of the secondary feature extraction and the secondary feature normalization extraction results, perform difference comparison with the historical image sample features respectively, and then take the average value according to the difference ratio between the two to obtain the image data analysis result.

[0051] Specifically, image features mainly include color features, texture features, shape features and spatial relationship features of the image.

[0052] Color feature is a global feature that describes the surface properties of the scene corresponding to the image or image area.

[0053] Texture features are also global features, which also describe the surface properties of the scene corresponding to the image or image area. However, since texture is only a characteristic of the surface of an object and cannot fully reflect the essential properties of the object, it is impossible to obtain high-level image content using only texture features. Unlike color features, texture features are not based on pixel points, and they need to be statistically calculated in an area containing multiple pixels.

[0054] There are two types of representation methods for shape features, one is contour features and the other is region features. The contour features of an image are mainly aimed at the outer boundaries of the object, while the region features of an image are related to the entire shape area.

[0055] The so-called spatial relationship refers to the spatial position or relative direction relationship between multiple objects segmented from the image. These relationships can also be divided into connection / adjacency relationship, overlap / overlap relationship and inclusion / inclusion relationship. Usually, spatial position information can be divided into two categories: relative spatial position information and absolute spatial position information. The former relationship emphasizes the relative situation between objects, such as up, down, left, and right, etc., while the latter relationship emphasizes the distance and direction between objects.

[0056] Example 8

[0057] Based on the previous embodiment, the pig health image analysis model is expressed using the following formula: ;in, is the image preprocessing data or the first feature, is the result of secondary feature extraction or secondary feature normalization; t is the adjustment coefficient. When the value range of t is 1~5, is the result of secondary feature extraction. When the value of t is between 6 and 10, It is the extraction result of secondary feature normalization.

[0058] Example 9

[0059] Based on the previous embodiment, the method in which the data analysis unit uses the pig farm data analysis result to correct the image data analysis result includes: multiplying the image data analysis result with the pig farm data analysis result to obtain the correction result, thereby completing the correction of the image data analysis result.

[0060] Example 10

[0061] A pig health early warning method based on image recognition, the method performs the following steps: step 1: collecting pig farm operation data and pig image data; the pig farm operation data at least includes: temperature, humidity, light intensity and microbial quantity; step 2: performing data preprocessing on the acquired pig farm operation data to obtain pig farm preprocessing data, and performing image preprocessing on the pig image data to obtain image preprocessing data; step 3: based on a preset data coupling judgment result model, coupling and judging the pig farm preprocessing data and the image preprocessing data to obtain a data coupling judgment result; if the data coupling judgment result judges that the pig farm preprocessing data If the data coupling judgment result determines that the pig farm preprocessing data and the image preprocessing data are coupled, the pig farm preprocessing data and the image preprocessing data are retained; if the data coupling judgment result determines that the pig farm preprocessing data and the image preprocessing data are not coupled, the pig farm preprocessing data and the image preprocessing data are discarded; Step 4: Use the preset pig farm health data analysis model to analyze the pig farm preprocessing data to obtain the pig farm data analysis results, and at the same time use the preset pig health image analysis model to analyze the image preprocessing data to obtain the image data analysis results, and use the pig farm data analysis results to correct the image data analysis results to obtain the final early warning results.

[0062] It should be noted that the system provided in the above embodiment is only illustrated by the division of the above functional units. In practical applications, the above functions can be assigned to different functional units as needed, that is, the units or steps in the embodiments of the present invention can be decomposed or combined. For example, the units in the above embodiments can be combined into one unit, or further divided into multiple sub-units to complete all or unit functions described above. The names of the units and steps involved in the embodiments of the present invention are only for distinguishing the various units or steps and are not regarded as improper limitations of the present invention.

[0063] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process and related instructions of the storage device and processing device described above can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0064] Those skilled in the art should be able to appreciate that the units and method steps of each example described in conjunction with the embodiments disclosed herein can be implemented by electronic hardware, computer software, or a combination of the two, and the programs corresponding to the software units and method steps can be placed in random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs, or any other form of storage medium known in the technical field. In order to clearly illustrate the interchangeability of electronic hardware and software, the composition and steps of each example have been generally described in the above description according to the function. Whether these functions are performed in electronic hardware or software depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered to exceed the scope of the present invention.

[0065] The terms “first”, “another part” and the like are used to distinguish similar objects but are not used to describe or indicate a specific sequence or order.

[0066] The term "comprise" or any other similar term is intended to cover non-exclusive inclusion, such that a process, method, article, or unit / apparatus that includes a list of elements includes not only those elements but also other elements not explicitly listed, or also includes elements inherent to such process, method, article, or unit / apparatus.

[0067] So far, the technical solutions of the present invention have been described in conjunction with the preferred embodiments shown in the accompanying drawings. However, it is easy for those skilled in the art to understand that the protection scope of the present invention is obviously not limited to these specific embodiments. Without departing from the principle of the present invention, those skilled in the art can make equivalent changes or substitutions to the relevant technical marks, and the technical solutions after these changes or substitutions will fall within the protection scope of the present invention.

[0068] The above description is only a preferred embodiment of the present invention and is not intended to limit the protection scope of the present invention.

Claims

1. The pig health early warning system based on image recognition is characterized by: The system comprises: a data acquisition unit, a data processing unit, a data coupling judgment unit and a data analysis unit; the data acquisition unit is configured to collect pig farm operation data and pig image data; the pig farm operation data at least comprises: temperature, humidity, light intensity and microorganism quantity; the data processing unit is configured to perform data preprocessing on the acquired pig farm operation data to obtain pig farm preprocessing data, and perform image preprocessing on the pig image data to obtain image preprocessing data; the data coupling judgment unit is configured to perform coupling judgment on the pig farm preprocessing data and the image preprocessing data based on a preset data coupling judgment result model to obtain a data coupling judgment result; if the data coupling judgment result is a predetermined value, the data coupling judgment result is obtained; If the result of the data coupling judgment determines that the pig farm preprocessing data and the image preprocessing data are coupled, the pig farm preprocessing data and the image preprocessing data are retained; if the result of the data coupling judgment determines that the pig farm preprocessing data and the image preprocessing data are not coupled, the pig farm preprocessing data and the image preprocessing data are discarded; the data analysis unit is configured to use a preset pig farm health data analysis model to perform data analysis on the pig farm preprocessing data to obtain a pig farm data analysis result, and at the same time use a preset pig health image analysis model to perform image data analysis on the image preprocessing data to obtain an image data analysis result, and use the pig farm data analysis result to correct the image data analysis result to obtain a final early warning result; The data processing unit pre-processes the obtained pig farm operation data, including: setting a time interval, dividing the obtained humidity, temperature, light intensity and microorganism quantity data in chronological order and according to the set time interval, to obtain multiple temperature data groups, humidity data groups, light intensity data groups and microorganism quantity data groups, and classifying the temperature data groups, humidity data groups, light intensity data groups and microorganism quantity data groups in the same time interval into one category to obtain multiple operation data categories, thereby completing the pig farm operation data pre-processing; The data processing unit performs image preprocessing on the pig image data to obtain the image preprocessing data, and the method includes: dividing the acquired image data according to the set time interval to obtain a plurality of image data groups, and completing the pig image data preprocessing; The data coupling judgment unit performs coupling judgment on the pig farm preprocessing data and the image preprocessing data based on a preset data coupling judgment result model, and the method for obtaining the data coupling judgment result includes: grouping the operation data class and the image data in the same time interval and performing coupling judgment using a preset data coupling judgment model; the data coupling judgment model is expressed using the following formula: Among them, N is the number of data in the running data class, s is the number of pixel values ​​in the image data group, and the number of pixel values ​​is defined as the number of categories of unequal pixel values; T is the image data group; F is the coupling judgment value; α is the adjustment coefficient, and the value range is: 3 to 6; the calculated coupling value is compared with the preset threshold range, and according to the comparison result, it is judged whether the pig farm preprocessing data and the image preprocessing data are coupled, so as to obtain the data coupling judgment result.

2. The system according to claim 1, characterized in that The data analysis unit uses a preset pig farm health data analysis model to perform data analysis on the pig farm preprocessing data to obtain the pig farm data analysis results. The method performs the following steps: uses the preset pig farm health data analysis to perform data analysis on each operation data class respectively, specifically including: unifying the data format of each data in the operation data class to obtain target data, and normalizing each data form in the target data according to attributes to generate a target matrix; selecting an attribute set for the target matrix through the pig farm health data analysis model, and screening out the target attribute set to form a sample matrix; selecting an attribute set for the sample matrix again through the pig farm health data analysis model to obtain sample characteristics, comparing the sample characteristics with the historical health sample characteristics, calculating the difference between the two, and using the difference as a correction value; the correction value is the final pig farm data analysis result.

3. The system according to claim 2, characterized in that The pig farm health data analysis model is expressed using the following formula: Among them, D is the target data, expressed in matrix form, H is the classification matrix, T is the transpose operation of the matrix, n is the order of the matrix, and Y is the sample matrix.

4. The system according to claim 3, characterized in that The data analysis unit uses a preset pig health image analysis model to analyze the image preprocessing data image data to obtain the image data analysis result. The method performs the following steps: firstly, the image preprocessing data is subjected to a first feature extraction using the pig health image analysis model to obtain a first feature; Then, the first feature is subjected to secondary feature extraction and secondary feature normalization extraction respectively; The results of secondary feature extraction and secondary feature normalization are compared with the historical image sample features respectively. The average value is taken according to the difference ratio between the two to obtain the image data analysis result.

5. The system according to claim 4, characterized in that The pig health image analysis model is expressed using the following formula: Among them, S is the image preprocessing data or the first feature, J is the result of secondary feature extraction or the extraction result of secondary feature normalization; t is the adjustment coefficient. When the value range of t is 1 to 5, J is the result of secondary feature extraction, and when the value range of t is 6 to 10, J is the extraction result of secondary feature normalization.

6. The system according to claim 5, characterized in that The method in which the data analysis unit uses the pig farm data analysis result to correct the image data analysis result includes: multiplying the image data analysis result with the pig farm data analysis result to obtain a correction result, thereby completing the correction of the image data analysis result; the pig farm data analysis result is.

7. A method of the pig health early warning system based on image recognition according to any one of claims 1 to 6, characterized in that: The method performs the following steps: Step 1: collecting pig farm operation data and pig image data; The pig farm operation data at least includes: temperature, humidity, light intensity and microorganism quantity; step 2: pre-processing the acquired pig farm operation data to obtain pig farm pre-processing data, and pre-processing the pig image data to obtain image pre-processing data; step 3: based on the preset data coupling judgment result model, coupling judgment is performed on the pig farm pre-processing data and the image pre-processing data to obtain the data coupling judgment result; if the data coupling judgment result judges that the pig farm pre-processing data and the image pre-processing data are coupled, the pig farm pre-processing data and the image pre-processing data are retained; if the data coupling judgment result judges that the pig farm pre-processing data and the image pre-processing data are not coupled, the pig farm pre-processing data and the image pre-processing data are discarded; step 4: using the preset pig farm health data analysis model to perform data analysis on the pig farm pre-processing data to obtain the pig farm data analysis result, and using the preset pig health image analysis model to analyze the image pre-processing data to obtain the image data analysis result, and using the pig farm data analysis result to correct the image data analysis result to obtain the final early warning result.

Images