Broiler abnormality monitoring method, system, medium and terminal based on data analysis

Abstract

The present invention discloses a broiler abnormality monitoring method, system, medium, and terminal based on data analysis, relating to the field of data processing technology. The method primarily aims to address the accuracy and effectiveness issues of existing broiler abnormality monitoring. The method comprises: collecting multiple broiler image data from a broiler breeding facility using cameras located at different monitoring angles; parsing location region identifiers in the broiler image data, and performing pixel segmentation on the broiler image data according to the location region identifiers to obtain multiple block-level image data; performing feature extraction on the multiple block-level image data to obtain feature images, and comparing the multiple feature images according to the comparison time length to determine the abnormal area; and generating the broiler abnormality monitoring result if the abnormal area matches a preset broiler abnormality contour range.

Description

Technical Field

[0001] The present invention relates to the field of data processing technology, and in particular to a broiler abnormality monitoring method, system, medium and terminal based on data analysis. Background Art

[0002] As people's living standards continue to improve, their demand for nutritious food is also increasing, especially for foods rich in essential nutrients such as chicken and eggs. Traditional free-range chicken farms struggle to meet this demand. Consequently, more and more modern chicken houses are being used. Although housing conditions are improving, a single house still needs to accommodate thousands of chickens to achieve large-scale farming. To ensure the healthy breeding of broiler chickens, real-time monitoring of their growth is necessary.

[0003] At present, the existing monitoring of broiler abnormalities is usually based on real-time monitoring of captured videos, that is, manual verification of the collected video streams. However, workers are required to screen large amounts of video data, which consumes a lot of manpower and material resources. Due to the large number of broilers, manual verification will generate a lot of time-consuming workload and it is easy to overlook small abnormalities, which greatly reduces the accuracy and effectiveness of broiler abnormality monitoring. Summary of the Invention

[0004] In view of this, the present invention provides a broiler abnormality monitoring method, system, medium and terminal based on data analysis, the main purpose of which is to solve the problems of poor accuracy and effectiveness of existing broiler abnormality monitoring.

[0005] According to one aspect of the present invention, a method for monitoring broiler abnormalities based on data analysis is provided, comprising:

[0006] Collect multiple broiler chicken image data in a broiler chicken breeding facility through cameras located at different monitoring angles;

[0007] parsing the position area identifiers in the broiler chicken image data, and performing pixel splitting on the broiler chicken image data according to the position area identifiers to obtain a plurality of block-level image data;

[0008] Extracting features from the plurality of block-level image data to obtain feature images, and comparing the plurality of feature images according to a comparison time length to determine an abnormal area;

[0009] If the abnormal area matches the preset broiler abnormal contour range, the broiler abnormal monitoring result is generated.

[0010] Furthermore, the parsing of the position area identifier in the broiler chicken image data and pixel splitting of the broiler chicken image data according to the position area identifier to obtain a plurality of block-level image data includes:

[0011] Determine a location area identifier based on matching at least one area identifier stored in a location identifier library with a location identifier in the broiler image data, wherein the location identifier library stores area identifiers obtained by the camera at different shooting angles and location areas corresponding to the area identifiers;

[0012] Analyze the pixel features of the area corresponding to at least one of the position area identifiers to determine the broiler density coefficient, and determine the number of blocks, block-level pixels, and block-level areas for pixel splitting according to the broiler density coefficient;

[0013] The broiler chicken image data is pixel-segmented according to the block-level number, the block-level pixels, and the block-level area.

[0014] Furthermore, analyzing the pixel features of the area corresponding to at least one of the position area identifiers to determine the broiler density coefficient includes:

[0015] If there are multiple position area identifiers, constructing the area outlines between the position area identifiers by combining the image data corresponding to the position area identifiers, calculating the color similarity of the pixel points in the area outlines according to the pixel colors, and determining the broiler density coefficient based on the difference between the color similarities;

[0016] If the position area identifier is one, a region outline is constructed between the image edge of the broiler image data and the position area identifier, and the color similarity of the pixel points in the region outline is calculated according to the pixel color, and the broiler density coefficient is determined based on the difference between the color similarities.

[0017] Furthermore, determining the broiler density coefficient based on the difference between the color similarities includes:

[0018] Determining a plurality of target pixel points to be compared for difference values ​​according to a first preset pixel multiple of the area outline, and calculating the difference values ​​of the color similarities of the target pixel points;

[0019] Determine the broiler density coefficient based on the number of target pixel points in the area contour at which the difference is obtained and the second preset pixel multiple;

[0020] Wherein, the first preset pixel multiple and the second preset pixel multiple are less than 1.

[0021] Furthermore, the step of extracting features from the plurality of block-level image data to obtain a feature image includes:

[0022] The feature extraction model based on the completed model training is used to extract features from the block-level image data to obtain the feature image. The feature extraction model is constructed based on a three-layer convolutional neural network and is trained based on block-level image samples marked with different features.

[0023] Furthermore, comparing the plurality of feature images according to the comparison time length to determine the abnormal area includes:

[0024] Determining a frame image capturing time of the camera, and determining the comparison time length based on a preset multiple of the frame image capturing time;

[0025] The characteristic images are compared according to the comparison time length to determine the abnormal area.

[0026] Furthermore, the method further comprises:

[0027] If the abnormal area does not match the preset broiler abnormal contour range, an area abnormality monitoring result is generated to instruct a cleaning prompt for the area corresponding to the position area identifier.

[0028] According to another aspect of the present invention, a broiler abnormality monitoring system based on data analysis is provided, comprising:

[0029] An acquisition module is used to collect multiple broiler chicken image data in a broiler chicken breeding facility through cameras located at different monitoring angles;

[0030] a parsing module, configured to parse the position area identifiers in the broiler chicken image data, and perform pixel splitting on the broiler chicken image data according to the position area identifiers to obtain a plurality of block-level image data;

[0031] An extraction module is used to extract features from the plurality of block-level image data to obtain feature images, and compare the plurality of feature images according to the comparison time length to determine an abnormal area;

[0032] A generation module is used to generate the broiler abnormality monitoring result if the abnormal area matches the preset broiler abnormality contour range.

[0033] Furthermore, the parsing module is used to match at least one area identifier stored in a location identifier library with the location identifier in the broiler image data to determine the location area identifier, wherein the location identifier library stores the area identifiers obtained by shooting at different shooting angles of the camera and the location area corresponding to the area identifier; parse the pixel features of the area corresponding to at least one of the location area identifiers to determine the broiler density coefficient, and determine the block-level number, block-level pixels, and block-level area of ​​pixel splitting according to the broiler density coefficient; and perform pixel splitting on the broiler image data according to the block-level number, the block-level pixels, and the block-level area.

[0034] Furthermore, the analysis module is specifically used to, if there are multiple position area identifiers, construct a region outline between the position area identifiers by combining the image data corresponding to the position area identifiers, calculate the color similarity of the pixel points in the region outline according to the pixel color, and determine the broiler density coefficient based on the difference between the color similarities; if there is one position area identifier, construct a region outline between the image edge of the broiler image data and the position area identifier, calculate the color similarity of the pixel points in the region outline according to the pixel color, and determine the broiler density coefficient based on the difference between the color similarities.

[0035] Furthermore, the analysis module is specifically used to determine multiple target pixel points whose differences are to be compared according to a first preset pixel multiple of the area contour, and calculate the difference in color similarity of the target pixel points; determine the broiler density coefficient based on the number of target pixel points whose differences are obtained in the area contour and the second preset pixel multiple; wherein the first preset pixel multiple and the second preset pixel multiple are less than 1.

[0036] Furthermore, the extraction module is specifically used to perform feature extraction on the block-level image data based on a feature extraction model that has completed model training to obtain the feature image. The feature extraction model is constructed based on a three-layer convolutional neural network and is trained based on block-level image samples marked with different features.

[0037] Furthermore, the extraction module is used to determine the frame image shooting time of the camera, and determine the comparison time length based on a preset multiple of the frame image shooting time; compare each of the feature images according to the comparison time length to determine the abnormal area.

[0038] Furthermore, the generation module is also used to generate a regional abnormality monitoring result if the abnormal area does not match the preset broiler abnormal contour range, to instruct to clean the area corresponding to the position area identifier.

[0039] According to another aspect of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the executable instruction enables a processor to execute operations corresponding to the above-mentioned broiler abnormality monitoring method based on data analysis.

[0040] According to another aspect of the present invention, there is provided a terminal, comprising: a processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus;

[0041] The memory is used to store at least one executable instruction, and the executable instruction enables the processor to execute operations corresponding to the above-mentioned broiler abnormality monitoring method based on data analysis.

[0042] By means of the above technical solution, the technical solution provided by the embodiment of the present invention has at least the following advantages:

[0043] The present invention provides a broiler abnormality monitoring method and system based on data analysis. Compared with the existing technology, the embodiment of the present invention collects multiple broiler image data in a broiler breeding place through cameras located at different monitoring angles; analyzes the position area identifiers in the broiler image data, and performs pixel splitting on the broiler image data according to the position area identifiers to obtain multiple block-level image data; performs feature extraction on the multiple block-level image data to obtain feature images, and compares the multiple feature images according to the comparison time length to determine the abnormal area; if the abnormal area matches the preset broiler abnormal contour range, the broiler abnormality monitoring result is generated, and real-time monitoring of broilers is achieved by performing data analysis on the image, which reduces the consumption of manpower and material resources, reduces the working time of the staff, greatly improves the effective monitoring of the small conditions of broilers, and thus improves the accuracy of chicken abnormality monitoring.

[0044] The above description is only an overview of the technical solution of the present invention. In order to more clearly understand the technical means of the present invention, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are specifically listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] Various other advantages and benefits will become apparent to those skilled in the art upon reading the detailed description of the preferred embodiment below. The accompanying drawings are for illustration purposes only and are not to be considered as limiting the present invention. The same reference symbols are used throughout the drawings to represent the same components. In the drawings:

[0046] Figure 1 A flow chart of a broiler abnormality monitoring method based on data analysis provided by an embodiment of the present invention is shown;

[0047] Figure 2 A schematic diagram of a broiler chicken farm provided by an embodiment of the present invention is shown;

[0048] Figure 3 A block diagram of a broiler abnormality monitoring system based on data analysis provided by an embodiment of the present invention is shown;

[0049] Figure 4 A schematic structural diagram of a terminal provided by an embodiment of the present invention is shown. DETAILED DESCRIPTION

[0050] Exemplary embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. Although exemplary embodiments of the present disclosure are shown in the accompanying drawings, it should be understood that the present disclosure can be implemented in various forms and should not be limited by the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the present disclosure to those skilled in the art.

[0051] The embodiment of the present invention provides a method for monitoring abnormalities of broiler chickens based on data analysis, such as Figure 1 As shown, the method includes:

[0052] 101. Collect multiple broiler chicken image data in a broiler chicken breeding facility through cameras located at different monitoring angles.

[0053] In the embodiment of the present invention, the current execution end, as the execution subject of broiler abnormality monitoring, can be a terminal device, or a cloud server, etc., and obtains the image of the broiler breeding place from the camera through data interaction with the camera, which is not specifically limited in the embodiment of the present invention. In the broiler breeding place, corresponding cameras can be installed in different locations, such as 4 cameras installed at the four corners of the breeding house, and the cameras shoot the broilers raised in the place according to the preset frame image shooting time to obtain broiler image data, and upload it to the current execution end through the camera. In addition, each camera can be installed according to different monitoring angles to obtain multiple broiler image data, which is not specifically limited in the embodiment of the present invention.

[0054] 102. Analyze the location area identifiers in the broiler chicken image data, and perform pixel splitting on the broiler chicken image data according to the location area identifiers to obtain a plurality of block-level image data.

[0055] In the embodiment of the present invention, after the current execution end obtains the broiler image data collected by the camera, it parses the location area identifier in the broiler image data. The location area identifier is used to represent different locations corresponding to different broiler isolation and feeding areas, such as Figure 2As shown, different location markers a, b, and c are configured at the edges of different areas, and cameras 1, 2, 3, and 4 are respectively installed at the four top corners of the broiler breeding pen, so that the current execution end can capture the location marker based on the camera and determine the location area marker to specifically analyze and identify the broilers in this area. After obtaining the location area marker, the current execution end performs pixel splitting on the broiler image data according to the location area marker, that is, dividing the pixels in a broiler image data into multiple block-level images. For example, according to the pixels of 800*300 of a broiler image data, each area in the broiler image data can be divided into blocks according to the location area marker to obtain multiple block-level image data, such as 30*50 pixels after the block-level image data is divided. This embodiment of the present invention does not make specific limitations.

[0056] In another embodiment of the present invention, for further definition and explanation, the step of parsing the location area identifier in the broiler chicken image data and performing pixel splitting on the broiler chicken image data according to the location area identifier to obtain multiple block-level image data includes:

[0057] Determine a location area identifier based on matching at least one area identifier stored in a location identifier library with a location identifier in the broiler image data;

[0058] Analyze the pixel features of the area corresponding to at least one of the position area identifiers to determine the broiler density coefficient, and determine the number of blocks, block-level pixels, and block-level areas for pixel splitting according to the broiler density coefficient;

[0059] The broiler chicken image data is pixel-segmented according to the block-level number, the block-level pixels, and the block-level area.

[0060] In order to perform block-level division on the image to improve the accuracy of broiler image feature extraction, when the current execution end parses the position area identifier, specifically, first, based on at least one region identifier stored in the position identifier library and the position identifier in the broiler image data, the position area identifier is determined. The position identifier library stores region identifiers obtained by shooting at different shooting angles of the camera and the position areas corresponding to the region identifiers. Therefore, after the current execution end shoots the broiler image data, first, since different region edges are configured with different position identifiers, the corresponding position identifier can be captured in the broiler image data during shooting. Then, according to the corresponding relationship stored in the position identifier library, the position area corresponding to the position identifier and the matching region identifier are queried to obtain the position area identifier. At this time, the position area identifier is a combination of the region identifier and the position identifier, which is not specifically limited in the embodiment of the present invention. Furthermore, since the broiler image data obtained during the mobile shooting process of the camera may contain at least one location identifier, thereby parsing to obtain at least one location area identifier, the current execution end performs pixel feature recognition according to the area corresponding to the location area identifier to determine the broiler density coefficient of the broilers in this area. At this time, the broiler density coefficient is used to characterize the distribution of broilers in a unit area. The coefficient value is in the range of 0-1. The larger the coefficient value, the denser the distribution of broilers in the unit area. The embodiment of the present invention does not make specific limitations.

[0061] It should be noted that after obtaining the broiler density coefficient, the current execution end determines the block number, block-level pixels, and block-level area of ​​pixel splitting according to the broiler density coefficient. At this time, a query can be performed based on the preset block correspondence relationship. The preset block correspondence relationship matches the block number, block-level pixels, and block-level areas corresponding to different broiler density coefficients for pixel splitting. For example, if the broiler density is 0.5, it means that the broilers are evenly distributed in the farm. Therefore, the corresponding block number is 10 and the block-level pixels are 50*100 (the pixels of the broiler image data are greater than 10 50*100, and the shooting parameters of the camera can be configured in advance so that the broiler image data can be blocked). The block-level area is the area corresponding to the position area identifier, so that the area corresponding to the position area identifier is pixel split according to the block number of 10 and the block-level pixels of 50*100 to obtain multiple block-level images.

[0062] In another embodiment of the present invention, for further definition and explanation, the step of analyzing pixel features of an area corresponding to at least one of the position area identifiers to determine the broiler density coefficient includes:

[0063] If there are multiple position area identifiers, constructing the area outlines between the position area identifiers by combining the image data corresponding to the position area identifiers, calculating the color similarity of the pixel points in the area outlines according to the pixel colors, and determining the broiler density coefficient based on the difference between the color similarities;

[0064] If the position area identifier is one, a region outline is constructed between the image edge of the broiler image data and the position area identifier, and the color similarity of the pixel points in the region outline is calculated according to the pixel color, and the broiler density coefficient is determined based on the difference between the color similarities.

[0065] Since the camera can be mobile when filming a broiler chicken breeding facility, the position area identifiers parsed from the obtained broiler chicken image data may be one or multiple. To accurately and effectively parse the broiler chicken density coefficient and thereby improve the accuracy of dividing the block-level image data based on the broiler chicken density coefficient, in a specific implementation scenario, the current execution end first determines the number of position area identifiers. If there are multiple position area identifiers, a region outline between the position area identifiers is constructed by combining the image data corresponding to the position area identifiers. The multiple position area identifiers are connected to obtain a region outline. Simultaneously, color similarity is calculated for the pixels in the region outline according to the pixel color. In this case, the color value of each pixel in the region outline is extracted and then similarity is calculated. During the calculation, each pixel is selected to perform similarity calculation with the corresponding adjacent pixel point, so that the broiler chicken density coefficient is determined based on the difference between the calculated color similarity of each similarity point and the adjacent pixel point. The adjacent similarity points include eight adjacent similarity points: above, below, left, right, upper left, lower left, upper right, and lower right of the middle pixel point, which are not specifically limited in this embodiment of the present invention.

[0066] In another embodiment of the present invention, for further definition and explanation, the step of determining the broiler chicken density coefficient based on the difference between the color similarities includes:

[0067] Determining a plurality of target pixel points to be compared for difference values ​​according to a first preset pixel multiple of the area outline, and calculating the difference values ​​of the color similarities of the target pixel points;

[0068] The broiler density coefficient is determined based on the number of target pixel points in the area contour at which the difference is obtained and the second preset pixel multiple.

[0069] Since the broiler density is used to characterize the distribution of broilers, in order to meet the effective calculation of the broiler density coefficient, the block-level division is performed based on the broiler density coefficient. The current execution end first determines the multiple target pixel points to be compared for the difference according to the first preset pixel multiple of the area contour. At this time, the first preset pixel multiple is less than 1, and can be configured according to the pixel value. If the pixel value is large, the first preset pixel multiple is large, such as 0.7 times, and more target pixel points are determined. The embodiment of the present invention does not make specific restrictions. In addition, when determining the multiple target pixel points to be compared for the difference according to the first preset pixel multiple, it can be determined by random sampling or proportional sampling. The embodiment of the present invention does not make specific restrictions. Furthermore, the color similarity is calculated according to the eight target pixel points: up, down, left, right, upper left, lower left, upper right, and lower right, and the difference between each determined target pixel point is calculated. The broiler density coefficient is determined based on the number of target pixel points of this difference and the second preset pixel multiple. Among them, the first preset pixel multiple and the second preset pixel multiple are both less than 1, and the first preset pixel multiple and the second preset pixel multiple can be the same or different. After obtaining the difference, first count the number of times this difference is greater than the preset difference threshold (the preset difference threshold can be configured based on the broiler color, venue brightness, etc., and the embodiment of the present invention does not make specific restrictions). At this time, if the counted number is greater than the number of pixels corresponding to the second pixel multiple, it means that the broilers in the image are densely distributed. Therefore, based on the sum of the number and the number of pixels divided by the number of pixels, it is used as the broiler density coefficient. If the counted number is less than or equal to the number of pixels corresponding to the second pixel multiple, it means that the broilers in the image are sparsely distributed. Therefore, based on the difference between the number and the pixel points divided by the number of pixels, it is used as the broiler density coefficient. The embodiment of the present invention does not make specific restrictions.

[0070] 103. Perform feature extraction on the plurality of block-level image data to obtain feature images, and compare the plurality of feature images according to comparison time lengths to determine abnormal areas.

[0071] In an embodiment of the present invention, the current execution end performs feature extraction on multiple block-level image data to obtain feature images. At this time, the feature images in multiple block-level image data corresponding to the same location area identifier can be compared according to the comparison time length to determine the abnormal area. The comparison time length can be determined according to the frame image shooting time. In this case, the frame image shooting time is less than the comparison time length. For example, the frame image shooting time is 1 / 24 second and the comparison time length is 1 second. Therefore, when comparing the comparison time length, multiple feature images can be compared to determine the abnormal area. This embodiment of the present invention does not specifically limit this.

[0072] In another embodiment of the present invention, for further definition and explanation, the step of comparing the plurality of feature images according to the comparison time length to determine the abnormal area includes:

[0073] Determining a frame image capturing time of the camera, and determining the comparison time length based on a preset multiple of the frame image capturing time;

[0074] The characteristic images are compared according to the comparison time length to determine the abnormal area.

[0075] In order to demarcate abnormal areas in broiler images and thereby utilize these areas to monitor the effectiveness of broilers, the current execution end, when determining the abnormal areas, first determines the frame image capture time of the camera. The frame image capture time is the time set for the camera to capture each frame, which can be 1 / 48 second or 1 / 24 second. Furthermore, the comparison time length is determined according to a preset multiple. The preset multiple can be configured based on the accuracy requirements for abnormal monitoring of broiler breeding scenes. The preset multiple is greater than 1. Preferably, the preset multiple can be 2x, 4x, etc., and is not specifically limited in the embodiments of the present invention. Furthermore, each feature image is compared according to the comparison time length, that is, the comparison time length is used as the time node, and the feature images obtained by parsing the block images of the corresponding blocks in the multiple broiler picture data within this time length are covered and compared, including the contour comparison of the feature image or the color value comparison of the feature image. When the pixel size of the contour difference or the pixel color difference is less than one tenth of the pixel in the feature image, it means that the broiler has been in this area for a long time. If the broiler remains motionless, this area is determined as an abnormal area, which is further matched with the preset broiler abnormal contour range to determine whether the broiler is abnormal.

[0076] In another embodiment of the present invention, for further definition and explanation, the step of extracting features from the plurality of block-level image data to obtain a feature image includes:

[0077] The feature extraction model based on the completed model training is used to extract features from the block-level image data to obtain the feature image. The feature extraction model is constructed based on a three-layer convolutional neural network and is trained based on block-level image samples marked with different features.

[0078] To effectively extract features from block-level image data and improve the accuracy of broiler anomaly monitoring, the current execution end pre-builds a three-layer convolutional network and obtains block-level image samples labeled with different feature labels. This is then used to train the three-layer convolutional neural network. During this process, the model parameters of the three-layer convolutional network can be optimized and adjusted during training to suit the purpose of extracting features from block-level images. This is not specifically limited in this embodiment of the present invention.

[0079] 104. If the abnormal area matches the preset broiler abnormal contour range, the broiler abnormal monitoring result is generated.

[0080] In an embodiment of the present invention, after obtaining the abnormal area, the current execution end compares the abnormal area with a preset broiler abnormal contour range to determine whether to generate an abnormal monitoring result for the broiler in order to avoid detecting fixed garbage or fixed objects. This is not specifically limited in this embodiment of the present invention. The preset broiler abnormal contour range is an image range divided based on the contours of different abnormal broilers, such as the contour image of a sick chicken or the contour image of a dead chicken. In this case, the preset broiler abnormal contour range can be pre-set for different broiler conditions, which is not specifically limited in this embodiment of the present invention.

[0081] In another embodiment of the present invention, for further definition and explanation, the steps further include:

[0082] If the abnormal area does not match the preset broiler abnormal contour range, an area abnormality monitoring result is generated to instruct a cleaning prompt for the area corresponding to the position area identifier.

[0083] To meet the varying monitoring requirements for broiler chickens, the current execution end compares the abnormal area with a preset broiler abnormality contour range. If the abnormal area does not match the preset broiler abnormality contour range, it indicates that the abnormal area may not be a broiler object. Therefore, a regional abnormality monitoring result is generated, rather than a non-broiler abnormality monitoring result, to alert the user of the abnormality in this area and to initiate cleaning. The regional abnormality monitoring result and the broiler abnormality monitoring result can be sent to the user end or alarm end via the current execution end in the form of a text message, sound, or other form, and are not specifically limited in this embodiment of the present invention.

[0084] An embodiment of the present invention provides a broiler abnormality monitoring method based on data analysis. Compared with the existing technology, the embodiment of the present invention collects multiple broiler image data in a broiler breeding facility through cameras located at different monitoring angles; analyzes the position area identifiers in the broiler image data, and performs pixel splitting on the broiler image data according to the position area identifiers to obtain multiple block-level image data; performs feature extraction on the multiple block-level image data to obtain feature images, and compares the multiple feature images according to the comparison time length to determine the abnormal area; if the abnormal area matches the preset broiler abnormal contour range, the broiler abnormality monitoring result is generated, thereby realizing real-time monitoring of broilers by performing data analysis on images, reducing manpower and material resources consumption, reducing the work time of staff, and greatly improving the effective monitoring of small conditions of broilers, thereby improving the accuracy of chicken abnormality monitoring.

[0085] Furthermore, as a response to the above Figure 1 The embodiment of the present invention provides a broiler abnormality monitoring system based on data analysis, such as Figure 3 As shown, the system includes:

[0086] The acquisition module 21 is used to collect multiple broiler chicken image data in the broiler chicken breeding facility through cameras located at different monitoring angles;

[0087] a parsing module 22 for parsing the position region identifiers in the broiler image data, and performing pixel splitting on the broiler image data according to the position region identifiers to obtain a plurality of block-level image data;

[0088] An extraction module 23 is configured to extract features from the plurality of block-level image data to obtain feature images, and compare the plurality of feature images according to a comparison time length to determine an abnormal area;

[0089] The generating module 24 is configured to generate the broiler abnormality monitoring result if the abnormal area matches a preset broiler abnormality contour range.

[0090] Furthermore, the parsing module is used to match at least one area identifier stored in a location identifier library with the location identifier in the broiler image data to determine the location area identifier, wherein the location identifier library stores the area identifiers obtained by shooting at different shooting angles of the camera and the location area corresponding to the area identifier; parse the pixel features of the area corresponding to at least one of the location area identifiers to determine the broiler density coefficient, and determine the block-level number, block-level pixels, and block-level area of ​​pixel splitting according to the broiler density coefficient; and perform pixel splitting on the broiler image data according to the block-level number, the block-level pixels, and the block-level area.

[0091] Furthermore, the analysis module is specifically used to, if there are multiple position area identifiers, construct a region outline between the position area identifiers by combining the image data corresponding to the position area identifiers, calculate the color similarity of the pixel points in the region outline according to the pixel color, and determine the broiler density coefficient based on the difference between the color similarities; if there is one position area identifier, construct a region outline between the image edge of the broiler image data and the position area identifier, calculate the color similarity of the pixel points in the region outline according to the pixel color, and determine the broiler density coefficient based on the difference between the color similarities.

[0092] Furthermore, the analysis module is specifically used to determine multiple target pixel points whose differences are to be compared according to a first preset pixel multiple of the area contour, and calculate the difference in color similarity of the target pixel points; determine the broiler density coefficient based on the number of target pixel points whose differences are obtained in the area contour and the second preset pixel multiple; wherein the first preset pixel multiple and the second preset pixel multiple are less than 1.

[0093] Furthermore, the extraction module is specifically used to perform feature extraction on the block-level image data based on a feature extraction model that has completed model training to obtain the feature image. The feature extraction model is constructed based on a three-layer convolutional neural network and is trained based on block-level image samples marked with different features.

[0094] Furthermore, the extraction module is used to determine the frame image shooting time of the camera, and determine the comparison time length based on a preset multiple of the frame image shooting time; compare each of the feature images according to the comparison time length to determine the abnormal area.

[0095] Furthermore, the generation module is also used to generate a regional abnormality monitoring result if the abnormal area does not match the preset broiler abnormal contour range, to instruct to clean the area corresponding to the position area identifier.

[0096] An embodiment of the present invention provides a broiler abnormality monitoring system based on data analysis. Compared with the existing technology, the embodiment of the present invention collects multiple broiler image data in a broiler breeding facility through cameras located at different monitoring angles; analyzes the position area identifiers in the broiler image data, and performs pixel splitting on the broiler image data according to the position area identifiers to obtain multiple block-level image data; performs feature extraction on the multiple block-level image data to obtain feature images, and compares the multiple feature images according to the comparison time length to determine the abnormal area; if the abnormal area matches the preset broiler abnormality contour range, the broiler abnormality monitoring result is generated, thereby realizing real-time monitoring of broilers by performing data analysis on images, reducing manpower and material resources consumption, reducing the work time of staff, and greatly improving the effective monitoring of small conditions of broilers, thereby improving the accuracy of chicken abnormality monitoring.

[0097] According to one embodiment of the present invention, a storage medium is provided, wherein the storage medium stores at least one executable instruction, and the computer executable instruction can execute the broiler abnormality monitoring method based on data analysis in any of the above method embodiments.

[0098] Figure 4A schematic structural diagram of a terminal provided according to an embodiment of the present invention is shown. The specific embodiment of the present invention does not limit the specific implementation of the terminal.

[0099] like Figure 4 As shown, the terminal may include: a processor 302 , a communications interface 304 , a memory 306 , and a communication bus 308 .

[0100] The processor 302 , the communication interface 304 , and the memory 306 communicate with each other via a communication bus 308 .

[0101] The communication interface 304 is used to communicate with other devices such as clients or other servers.

[0102] The processor 302 is used to execute the program 310, and specifically can execute the relevant steps in the embodiment of the broiler abnormality monitoring method based on data analysis.

[0103] Specifically, the program 310 may include program codes, which include computer operation instructions.

[0104] Processor 302 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement embodiments of the present invention. The one or more processors included in the terminal may be processors of the same type, such as one or more CPUs, or processors of different types, such as one or more CPUs and one or more ASICs.

[0105] The memory 306 is used to store the program 310. The memory 306 may include a high-speed RAM memory, and may also include a non-volatile memory (non-volatile memory), such as at least one disk memory.

[0106] The program 310 may be specifically configured to cause the processor 302 to perform the following operations:

[0107] Collect multiple broiler chicken image data in a broiler chicken breeding facility through cameras located at different monitoring angles;

[0108] parsing the position area identifiers in the broiler chicken image data, and performing pixel splitting on the broiler chicken image data according to the position area identifiers to obtain a plurality of block-level image data;

[0109] Extracting features from the plurality of block-level image data to obtain feature images, and comparing the plurality of feature images according to a comparison time length to determine an abnormal area;

[0110] If the abnormal area matches the preset broiler abnormal contour range, the broiler abnormal monitoring result is generated.

[0111] Obviously, those skilled in the art will appreciate that the various modules or steps of the present invention described above can be implemented using a general-purpose computing system. They can be centralized on a single computing system or distributed across a network of multiple computing systems. Alternatively, they can be implemented using program code executable by a computing system, and thus, they can be stored in a storage system and executed by the computing system. In some cases, the steps shown or described herein can be performed in a different order than that shown, or they can be fabricated into separate integrated circuit modules, or multiple modules or steps can be fabricated into a single integrated circuit module. Thus, the present invention is not limited to any particular combination of hardware and software.

[0112] The foregoing description is merely a preferred embodiment of the present invention and is not intended to limit the present invention. Those skilled in the art will readily appreciate that various modifications and variations of the present invention are possible. Any modifications, equivalent substitutions, or improvements made within the spirit and principles of the present invention shall be included within the scope of protection of the present invention.

Claims

1. A broiler abnormality monitoring method based on data analysis, characterized in that: include: Collect multiple broiler chicken image data in a broiler chicken breeding facility through cameras located at different monitoring angles; parsing the position area identifiers in the broiler chicken image data, and performing pixel splitting on the broiler chicken image data according to the position area identifiers to obtain a plurality of block-level image data; Extracting features from the plurality of block-level image data to obtain feature images, and comparing the plurality of feature images according to a comparison time length to determine an abnormal area; If the abnormal area matches the preset broiler abnormal contour range, a broiler abnormal monitoring result is generated; The parsing of the position area identifier in the broiler chicken image data and pixel splitting of the broiler chicken image data according to the position area identifier to obtain a plurality of block-level image data includes: Determine a location area identifier based on matching at least one area identifier stored in a location identifier library with a location identifier in the broiler image data, wherein the location identifier library stores area identifiers obtained by the camera at different shooting angles and location areas corresponding to the area identifiers; Analyze the pixel features of the area corresponding to at least one of the position area identifiers to determine the broiler density coefficient, and determine the number of blocks, block-level pixels, and block-level areas for pixel splitting according to the broiler density coefficient; Perform pixel splitting on the broiler chicken image data according to the block-level number, the block-level pixels, and the block-level area; The analyzing pixel features of the area corresponding to at least one of the position area identifiers to determine the broiler density coefficient includes: If there are multiple position area identifiers, constructing the area outlines between the position area identifiers by combining the image data corresponding to the position area identifiers, calculating the color similarity of the pixel points in the area outlines according to the pixel colors, and determining the broiler density coefficient based on the difference between the color similarities; If the position area identifier is one, a region outline is constructed between the image edge of the broiler image data and the position area identifier, and the color similarity of the pixel points in the region outline is calculated according to the pixel color, and the broiler density coefficient is determined based on the difference between the color similarities.

2. The method according to claim 1, characterized in that Determining the broiler chicken density coefficient based on the difference between the color similarities includes: Determining a plurality of target pixel points to be compared for difference values ​​according to a first preset pixel multiple of the area outline, and calculating the difference values ​​of the color similarities of the target pixel points; Determine the broiler density coefficient based on the number of target pixel points in the area contour at which the difference is obtained and a second preset pixel multiple; Wherein, the first preset pixel multiple and the second preset pixel multiple are less than 1.

3. The method according to claim 1, characterized in that The extracting features from the plurality of block-level image data to obtain a feature image comprises: The feature extraction model based on the completed model training is used to extract features from the block-level image data to obtain the feature image. The feature extraction model is constructed based on a three-layer convolutional neural network and is trained based on block-level image samples marked with different features.

4. The method according to claim 1, wherein The comparing the plurality of feature images according to the comparison time length to determine the abnormal area includes: Determining a frame image capturing time of the camera, and determining the comparison time length based on a preset multiple of the frame image capturing time; The characteristic images are compared according to the comparison time length to determine the abnormal area.

5. The method according to claim 1, characterized in that The method further comprises: If the abnormal area does not match the preset broiler abnormal contour range, an area abnormality monitoring result is generated to instruct a cleaning prompt for the area corresponding to the position area identifier.

6. A broiler abnormality monitoring system based on data analysis, characterized in that: include: An acquisition module is used to collect multiple broiler chicken image data in a broiler chicken breeding facility through cameras located at different monitoring angles; a parsing module, configured to parse the position area identifiers in the broiler chicken image data, and perform pixel splitting on the broiler chicken image data according to the position area identifiers to obtain a plurality of block-level image data; An extraction module is used to extract features from the plurality of block-level image data to obtain feature images, and compare the plurality of feature images according to the comparison time length to determine an abnormal area; A generating module, configured to generate a broiler abnormality monitoring result if the abnormal area matches a preset broiler abnormality contour range; The parsing module is used to match at least one area identifier stored in a location identifier library with the location identifier in the broiler image data to determine the location area identifier, wherein the location identifier library stores the area identifiers obtained by shooting at different shooting angles of the camera and the location areas corresponding to the area identifiers; parse the pixel features of the area corresponding to at least one of the location area identifiers to determine the broiler density coefficient, and determine the block-level number, block-level pixels, and block-level areas of pixel splitting according to the broiler density coefficient; and perform pixel splitting on the broiler image data according to the block-level number, the block-level pixels, and the block-level areas; The analysis module is specifically further used to, if there are multiple position area identifiers, construct the area outline between the position area identifiers by combining the image data corresponding to the position area identifiers, calculate the color similarity of the pixel points in the area outline according to the pixel color, and determine the broiler density coefficient based on the difference between the color similarities; if there is one position area identifier, construct the area outline between the image edge of the broiler image data and the position area identifier, calculate the color similarity of the pixel points in the area outline according to the pixel color, and determine the broiler density coefficient based on the difference between the color similarities.

7. A storage medium storing at least one executable instruction, wherein the executable instruction enables a processor to execute operations corresponding to the broiler abnormality monitoring method based on data analysis as described in any one of claims 1 to 5.

8. A terminal comprising: A processor, a memory, a communication interface, and a communication bus, wherein the processor, the memory, and the communication interface communicate with each other via the communication bus; The memory is used to store at least one executable instruction, and the executable instruction enables the processor to perform operations corresponding to the broiler abnormality monitoring method based on data analysis as described in any one of claims 1 to 5.

Images