A behavioral analysis method and system for yak breeding

Abstract

The present invention relates to a behavioral analysis method and system for yak farming, comprising: extracting yak objects from captured images; labeling the yak objects to obtain yak objects with identity tags, which are recorded as characteristic yak objects; recording the movement trajectories and behavioral characteristics of the characteristic yak objects; performing similarity analysis on the movement trajectories and behavioral characteristics of the characteristic yak objects in a time dimension to obtain similarity analysis results; and grouping the characteristic yak objects according to the similarity analysis results to obtain normal and abnormal groups. The behavioral analysis method and system for yak farming disclosed in the present invention utilizes local multi-point recognition combined with group difference analysis to perform consistent behavioral analysis of yaks. This improves the data richness of individual yaks by improving data consistency, while simultaneously identifying distinguishable individuals through group difference analysis.

Description

Technical Field

[0001] The present invention relates to the technical field of image processing, and in particular to a behavior analysis method and system for yak breeding. Background Art

[0002] Yak farming can be broadly divided into free-range (non-winter) and confined (winter) farming. Confined farming aims to fatten yaks, addressing traditional winter yak farming practices that often result in yak sluggishness, inability to market, and high mortality rates. Current research focuses on feed rationing, water management, and breast milk replacement. Free-range farming, a method that combines natural grazing with scientific management, is particularly suitable for alpine pastoral areas such as the Qinghai-Tibet Plateau. Current research focuses on pasture rotation, individual group management, and disease prevention and control.

[0003] Another research direction of free-range farming is to analyze individuals to evaluate the growth of yaks. Currently, the more commonly used method is image analysis, but the coverage area of ​​the free-range farming area is large, and the coverage of fixed equipment is insufficient, and it is easy to disturb the yak herd. Therefore, the use of drones for monitoring has been explored, but drones need to be used for long-distance monitoring.

[0004] A major problem facing long-distance monitoring is the insufficient individual recognition rate, which leads to poor consistency of individual data and insufficient reference value of the analysis results. At the same time, in terms of behavioral analysis, due to the insufficient richness of data collected at a distance, the current data samples are small and lack representativeness, so the behavioral analysis results also have the problem of insufficient reference value. Summary of the Invention

[0005] The present invention provides a behavioral analysis method and system for yak breeding, which uses local multi-point recognition combined with group difference analysis to perform consistent behavioral analysis on yaks, improves the data richness of individual yaks by improving data consistency, and finds distinguishable individuals through group difference analysis.

[0006] The above-mentioned object of the present invention is achieved through the following technical solutions:

[0007] In a first aspect, the present invention provides a behavior analysis method for yak breeding, comprising:

[0008] In response to the acquired image, extracting a yak object in the image;

[0009] Marking the yak object to obtain a yak object with an identity mark, which is recorded as a feature yak object;

[0010] Record the movement trajectory and behavioral characteristics of the characteristic yak object;

[0011] Conduct similarity analysis on the movement trajectory and behavior characteristics of the characteristic yak object in the time dimension to obtain similarity analysis results;

[0012] According to the similarity analysis results, the characteristic yak objects were grouped into normal group and abnormal group.

[0013] In a possible implementation of the first aspect, marking the yak object includes:

[0014] The image corresponding to the yak object is divided into regions to obtain multiple sub-images, and the sub-images are arranged in the form of an MxN matrix, where M and N are both natural numbers greater than zero;

[0015] Extract spots in a sub-image, and when there are no spots in a sub-image, fuse the sub-image into an adjacent sub-image with spots;

[0016] Create regions of interest based on blobs;

[0017] Optimize the spots within the region of interest, including position optimization and scale optimization;

[0018] The optimized spot regions are used to establish feature groups, which serve as markers for the marking process.

[0019] In a possible implementation manner of the first aspect, extracting spots from the sub-image includes:

[0020] Performing noise reduction on the sub-image, the noise reduction process includes representing each row or column of pixels in the sub-image with a curve, and then performing equal-length smoothing on the curve;

[0021] Use the extraction operator to extract spots on the sub-image;

[0022] Use the regions of interest created based on the spots to filter the spots and retain the spots that meet the requirements;

[0023] When filtering spots, the spots are filtered based on the brightness difference between the spots and the regions of interest. The brightness of the spots is the maximum brightness or minimum brightness of the spots, and the brightness of the regions of interest is the average brightness of the regions of interest.

[0024] In a possible implementation of the first aspect, optimizing the scale of the spots within the region of interest includes:

[0025] Performing clarity statistics on the acquired images in time series to obtain multiple clarity value groups;

[0026] Processing the clarity of the region of interest using the obtained clarity value group;

[0027] Re-identifying spots in the region of interest processed by sharpness and calculating the probability of the spots being re-identified;

[0028] The spots are sorted according to the probability and a set number or a set proportion of spots are retained in the order sequence, and the probability of the spots being re-determined decreases in the order sequence.

[0029] In a possible implementation of the first aspect, optimizing the position of the spot within the region of interest includes:

[0030] Setting multiple analysis reference lines in the spot area, wherein the multiple analysis reference lines are parallel to each other;

[0031] Use the pixel points on the analysis reference line to establish an analysis curve and obtain the lowest point of the analysis curve;

[0032] Use the lowest point of the obtained analytical curve to establish a central reference straight line;

[0033] Obtaining multiple central reference lines and determining a convergence point or convergence area of ​​the central reference lines;

[0034] After moving the center of the spot to the collection point or collection area, the edge of the spot is re-collected and the edge value of the extracted spot is determined;

[0035] The blobs are reconstructed using edge values.

[0036] In a possible implementation of the first aspect, performing similarity analysis on the movement trajectory of the characteristic yak object in the time dimension includes:

[0037] An analysis grid is established with the location of the characteristic yak object as the reference point, and the number of edges of each grid in the analysis grid is the same;

[0038] Calculate the average length of the grid edges;

[0039] The outliers are determined based on the average value, and the edge lengths of the grids corresponding to the outliers are all greater than the average value;

[0040] The outlier time of outliers is counted in the time dimension.

[0041] In a possible implementation of the first aspect, performing behavioral feature analysis on the movement trajectory of the characteristic yak object in the time dimension includes:

[0042] Determine the head identification feature of the characteristic yak object, where the head identification feature is the horn feature;

[0043] A width variation feature group of head recognition features is established in the time series, and the width variation feature group includes area value and time length of area value;

[0044] Compare the similarities between width variation feature groups;

[0045] Normal behavior individuals and abnormal behavior individuals in the characteristic yak objects are determined according to the similarity.

[0046] In a second aspect, the present invention provides a behavior analysis device for yak breeding, comprising:

[0047] an object extraction unit, configured to extract a yak object in the image in response to the acquired image;

[0048] An object marking unit is used to mark the yak object to obtain a yak object with an identity mark, which is recorded as a characteristic yak object;

[0049] A recording unit, used to record the movement trajectory and behavioral characteristics of the characteristic yak object;

[0050] An analysis unit, configured to perform similarity analysis on the movement trajectory and behavior characteristics of the characteristic yak object in a time dimension to obtain a similarity analysis result;

[0051] The grouping unit is used to group the characteristic yak objects according to the similarity analysis results to obtain normal groups and abnormal groups.

[0052] In a third aspect, the present invention provides a behavioral analysis system for yak breeding, the system comprising:

[0053] one or more memories for storing instructions; and

[0054] One or more processors, configured to call and execute the instructions from the memory to perform the method as described in the first aspect and any possible implementation of the first aspect.

[0055] In a fourth aspect, the present invention provides a computer-readable storage medium, the computer-readable storage medium comprising:

[0056] The program, when the program is executed by a processor, the method described in the first aspect and any possible implementation of the first aspect is executed.

[0057] In a fifth aspect, the present invention provides a computer program product comprising program instructions. When the program instructions are executed by a computing device, the method described in the first aspect and any possible implementation of the first aspect is executed.

[0058] In a sixth aspect, the present invention provides a chip system comprising a processor for implementing the functions involved in the above aspects, such as generating, receiving, sending, or processing the data and / or information involved in the above methods.

[0059] The chip system may be composed of chips, or may include chips and other discrete devices.

[0060] In one possible design, the chip system also includes a memory for storing necessary program instructions and data. The processor and the memory can be decoupled and provided on different devices, connected via wired or wireless means, or the processor and the memory can be coupled on the same device.

[0061] The beneficial technical effects of the present invention are:

[0062] The behavioral analysis method and system for yak breeding disclosed in the present invention use local multi-point recognition combined with group difference analysis to conduct consistent behavioral analysis of yaks. This method can obtain individual data based on images taken by drones and discover abnormal individuals through statistical analysis of a yak herd, making it easier for herders to manage the yak herd and detect problems early. BRIEF DESCRIPTION OF THE DRAWINGS

[0063] Figure 1 This is a schematic flowchart of the steps of a behavior analysis method for yak breeding provided by the present invention.

[0064] Figure 2 This is a schematic diagram of extracting spots from a sub-image provided by the present invention.

[0065] Figure 3 This is a schematic diagram of establishing a feature group using spot areas provided by the present invention. DETAILED DESCRIPTION

[0066] The technical solution of the present invention is further described in detail below with reference to the accompanying drawings.

[0067] The present invention discloses a behavior analysis method for yak breeding, please refer to Figure 1 In some examples, the behavioral analysis method for yak farming disclosed in the present invention includes the following steps:

[0068] S101, in response to the acquired image, extracting a yak object in the image;

[0069] S102, marking the yak object to obtain a yak object with an identity mark, which is recorded as a characteristic yak object;

[0070] S103, recording the movement trajectory and behavioral characteristics of the characteristic yak object;

[0071] S104, performing similarity analysis on the movement trajectory and behavior characteristics of the characteristic yak object in the time dimension to obtain a similarity analysis result;

[0072] S105 , grouping the characteristic yak objects according to the similarity analysis result to obtain a normal group and an abnormal group.

[0073] Overall, the behavioral analysis method for yak breeding provided in this application is to determine the similarities and differences of individuals in a group through continuous analysis of the behavior of yak objects. In step S101, the yak objects in the image will be extracted first. The image here is generally taken by drones. The reason for using drones is that yaks are currently mostly raised in a free-range mode, which has a large coverage area and uses fixed image sensors. There are defects such as a large number of devices, difficulty in maintenance, and easy impact on the normal activities of yaks. The drone can move with the movement of the yak herd and can perform long-distance shooting, which can minimize the impact on the normal activities of the yak herd.

[0074] In step S102, the yak object is marked to obtain a yak object with an identity mark, which is recorded as a characteristic yak object. The purpose of marking the yak object is to achieve continuous tracking of a yak object.

[0075] It should be understood that when a drone photographs a yak object, it obtains images that are continuously arranged in a time series. The yak objects in these images need to be matched before a continuous movement trajectory of the yak object can be obtained and then analyzed.

[0076] In step S103 , the movement trajectory and behavior characteristics of the characteristic yak object are recorded. The movement trajectory here is generally represented by coordinates and line segments, that is, the movement trajectory is represented by multiple end-to-end connected line segments.

[0077] In this application, behavioral characteristics are divided into three categories: walking, resting and eating. Walking (moving state) can be represented by the movement trajectory, and resting (stationary state) and eating are both represented by time periods. Here, resting (stationary state) and eating require the drone to take side images, that is, to shoot in the area not directly above the yak object.

[0078] In step S104, a similarity analysis is performed on the movement trajectory and behavioral characteristics of the characteristic yak object in the time dimension to obtain a similarity analysis result. The similarity analysis result here includes two categories: movement trajectory and behavioral characteristics. The movement trajectory analysis is mainly to find out whether the characteristic yak object has outlier behavior. The analysis method here is to find out by comparing the similarity of the movement trajectory. The reference indicators are distance and time. These two indicators are set according to the specific environment of the use location and are not restricted here.

[0079] The main reasons for separation include failure in struggle, natural separation, stimulation and food shortage, etc. At this time, further analysis can be carried out through images taken by drones.

[0080] Behavioral characteristic analysis mainly compares the differences in rest time and eating time. For example, if the characteristic yak subjects have too long rest time or insufficient eating time, it indicates that there may be a disease.

[0081] Finally, in step S105 , the characteristic yak objects are grouped according to the similarity analysis results to obtain a normal group and an abnormal group.

[0082] It should be noted that there are currently some research modeling methods, which use statistical analysis and neural networks to perform differential analysis. However, due to insufficient sample capacity and differences between yak herds, the neural networks trained in this way lack stability and adaptability, and can often only be used to process one or more specific yak herds.

[0083] In order to solve this practical problem, the method used in this application is to conduct an inter-individual difference analysis on a yak herd. This method does not require modeling. By screening the obtained data (distance, time away from the herd, rest time and eating time, etc.), different individuals can be found. Then the staff can analyze the specific situation of this different individual and choose the appropriate treatment plan.

[0084] In some examples, the specific way to label the Yak object is as follows:

[0085] S201, dividing the image corresponding to the yak object into regions to obtain multiple sub-images, wherein the sub-images are arranged in an MxN matrix, where M and N are both natural numbers greater than zero;

[0086] S202, extracting spots from the sub-image, and when there are no spots in the sub-image, fusing the sub-image into an adjacent sub-image with spots;

[0087] S203, creating a region of interest based on the spots;

[0088] S204, optimizing the spots within the region of interest, including position optimization and scale optimization;

[0089] S205 , using the optimized spot area to establish a feature group, and the feature group is used as a mark for marking processing.

[0090] See also Figure 2In steps S201 to S205, spots are extracted from the divided sub-images, and then regions of interest are created based on the spots. Then, feature groups are established using the spot regions. The feature groups serve as markers for labeling.

[0091] It should be understood that the differences among individual yaks in a yak herd (characteristic yak objects) take priority and cannot be processed solely through facial recognition. At the same time, due to the limitations of the high-altitude operation method adopted by drones, it is impossible to obtain all the facial features of the characteristic yak objects.

[0092] In order to solve this problem, this application uses a processing method of matching using feature groups. The feature groups are established based on the images corresponding to the yak objects, which reflect some features on the body of the feature yak objects. Through the matching of these features, the feature yak objects at different time points can be matched to obtain the performance of the feature yak objects in the time series.

[0093] The specific method of extracting spots in the sub-image is as follows:

[0094] Performing noise reduction on the sub-image, the noise reduction process includes representing each row or column of pixels in the sub-image with a curve, and then performing equal-length smoothing on the curve;

[0095] Use the extraction operator to extract spots on the sub-image;

[0096] Use the regions of interest created based on the spots to filter the spots and retain the spots that meet the requirements;

[0097] When filtering spots, the spots are filtered based on the brightness difference between the spots and the regions of interest. The brightness of the spots is the maximum brightness or minimum brightness of the spots, and the brightness of the regions of interest is the average brightness of the regions of interest.

[0098] In the above method, the sub-image is first subjected to denoising to remove interference factors, and then an extraction operator is used to extract spots on the sub-image. The extraction operator here refers to the operator used in the edge extraction method. One difference of spots is that they are different in color from the surrounding area, so they can be extracted using an extraction operator.

[0099] After obtaining the spots, you need to use the regions of interest created based on the spots to filter them. The specific principle is: filter based on the brightness difference between the spots and the regions of interest. The brightness of the spots is the maximum or minimum brightness of the spots, and the brightness of the regions of interest is the average brightness of the regions of interest. The absolute value of the difference between the two must be greater than or equal to a set parameter. There is no restriction on the setting parameter here.

[0100] When there are multiple spots, the spot with the largest area or the largest brightness difference is retained. Generally, one spot is retained for each region of interest.

[0101] The purpose of spot screening is to remove some spots with insufficient distinction, because these spots with insufficient distinction may have defects that cannot be continuously displayed on the image during subsequent processing.

[0102] In the above content, we also mentioned the optimization of spots in the region of interest. The optimization includes two aspects: position optimization and scale optimization. The specific method of scale optimization is as follows:

[0103] S301, performing clarity statistics on the acquired images in a time series to obtain a plurality of clarity value groups;

[0104] S302, processing the clarity of the region of interest using the obtained clarity value group;

[0105] S303, re-determining spots in the region of interest after the clarity processing and calculating the probability of the spots being re-determined;

[0106] S304 , sorting the spots according to the probability and retaining a set number or a set proportion of spots in the sequence. In the sequence, the probability of the spots being re-determined decreases.

[0107] In steps S301 to S304, the clarity of the acquired images is first statistically analyzed in a time series to obtain multiple clarity value groups. The clarity value groups here represent different clarity. The reason for statistical clarity is that during the image shooting process of the drone, it is impossible to guarantee that the clarity of each image is consistent. When the clarity of the image changes, some spots may have defects that are not displayed on the image, and these spots need to be excluded.

[0108] After obtaining the clarity value group, the clarity of the region of interest is processed using the obtained clarity value group, and then the spots are re-determined in the region of interest after the clarity processing and the probability of the spots being re-determined is counted. For example, there are a total of ten clarity value groups. During the ten processing processes, if the spots are re-determined eight times, then the probability of the spots being re-determined is 0.8. If the spots are re-determined ten times, then the probability of the spots being re-determined is 1.

[0109] Finally, the spots are sorted according to the probability and a set number or a set proportion of spots are retained in the sequence. In the sequence, the probability of the spots being re-determined decreases.

[0110] The position of the spots within the region of interest is optimized as follows:

[0111] S401, setting a plurality of analysis reference lines in the spot area, wherein the plurality of analysis reference lines are parallel to each other;

[0112] S402, establishing an analysis curve using the pixel points on the analysis reference line and obtaining the lowest point of the analysis curve;

[0113] S403, establishing a central reference straight line using the lowest point of the obtained analysis curve;

[0114] S404, obtaining multiple central reference lines and determining a convergence point or convergence area of ​​the central reference lines;

[0115] S405, after moving the center of the spot to the collection point or collection area, re-collect the edge of the spot and determine the edge value of the extracted spot;

[0116] S406: Reconstruct the spots using edge values.

[0117] In steps S401 to S406, multiple analysis reference lines are first set in the spot area (multiple analysis reference lines are parallel to each other), and then the pixel points on the analysis reference lines are used to establish an analysis curve. The specific method is to use the order of the pixel points as the horizontal coordinate and the value of the pixel points as the vertical coordinate to obtain discrete points, and then connect these discrete points together in sequence.

[0118] Determine the lowest point of the analysis curve and use the obtained lowest point of the analysis curve to establish a central reference straight line. The method of establishing the central reference straight line is to use a linear regression method to obtain a straight line, requiring that the sum of the distances from all the lowest points to this straight line is minimized.

[0119] The center reference line is obtained using the above method. During the analysis, the reference line will rotate. The multiple center reference lines obtained may have one or more intersections. The intersection point is called the convergence point, and the area where the multiple intersections are located is called the convergence area.

[0120] Then, the center of the spot is moved to the collection point or collection area, the edge of the spot is re-collected, the edge value of the extracted spot is determined, and the spot is reconstructed using the edge value.

[0121] The purpose of this method is to mathematically re-determine the center of the spot and then use the edge of the moved spot to determine the pixel values ​​that comprise the spot's edge. This is because edge extraction using an operator can introduce certain errors and ambiguity. Reconstructing the spot's edge after re-determining the spot's center can yield a more accurate spot edge.

[0122] For some examples, see Figure 3,The similarity analysis of the moving trajectory of the feature yak object in the time dimension is as follows:

[0123] An analysis grid is established with the location of the characteristic yak object as the reference point, and the number of edges of each grid in the analysis grid is the same;

[0124] Calculate the average length of the grid edges;

[0125] The outliers are determined based on the average value, and the edge lengths of the grids corresponding to the outliers are all greater than the average value;

[0126] The outlier time of outliers is counted in the time dimension.

[0127] In some examples, the specific method of analyzing the behavioral characteristics of the movement trajectory of the characteristic yak object in the time dimension is as follows:

[0128] Determine the head identification feature of the characteristic yak object, where the head identification feature is the horn feature;

[0129] A width variation feature group of head recognition features is established in the time series, and the width variation feature group includes area value and time length of area value;

[0130] Compare the similarities between width variation feature groups;

[0131] Normal behavior individuals and abnormal behavior individuals in the characteristic yak objects are determined according to the similarity.

[0132] The above method is to perform behavioral feature analysis through head recognition features, that is, horn features. The specific method is to establish a width change feature group of head recognition features in a time series. The width change feature group includes area values ​​and area value time lengths, that is, to count the area changes and area change time of the two horns of the yak. ​​This statistical method can reflect the time when the yak lowers its head, which can be regarded as the time when the yak eats and drinks water.

[0133] The change in the area of ​​the two corners refers to the maximum outline corresponding to the two corners. The change in the area of ​​this outline is because when the yak lowers its head, the area of ​​the maximum outline corresponding to the two corners will change synchronously.

[0134] In addition, by counting the number of times the area of ​​the maximum contour corresponding to the two corners changes, other abnormalities can also be discovered. For example, if the number of times the area of ​​a certain characteristic yak object changes significantly increases, it means that there may be a certain disease in the body.

[0135] The present invention also provides a behavior analysis device for yak breeding, comprising:

[0136] an object extraction unit, configured to extract a yak object in the image in response to the acquired image;

[0137] An object marking unit is used to mark the yak object to obtain a yak object with an identity mark, which is recorded as a characteristic yak object;

[0138] A recording unit, used to record the movement trajectory and behavioral characteristics of the characteristic yak object;

[0139] An analysis unit, configured to perform similarity analysis on the movement trajectory and behavior characteristics of the characteristic yak object in a time dimension to obtain a similarity analysis result;

[0140] The grouping unit is used to group the characteristic yak objects according to the similarity analysis results to obtain normal groups and abnormal groups.

[0141] Furthermore, the marking process for the yak object includes:

[0142] The image corresponding to the yak object is divided into regions to obtain multiple sub-images, and the sub-images are arranged in the form of an MxN matrix, where M and N are both natural numbers greater than zero;

[0143] Extract spots in a sub-image, and when there are no spots in a sub-image, fuse the sub-image into an adjacent sub-image with spots;

[0144] Create regions of interest based on blobs;

[0145] Optimize the spots within the region of interest, including position optimization and scale optimization;

[0146] The optimized spot regions are used to establish feature groups, which serve as markers for the marking process.

[0147] Furthermore, extracting spots in the sub-image includes:

[0148] Performing noise reduction on the sub-image, the noise reduction process includes representing each row or column of pixels in the sub-image with a curve, and then performing equal-length smoothing on the curve;

[0149] Use the extraction operator to extract spots on the sub-image;

[0150] Use the regions of interest created based on the spots to filter the spots and retain the spots that meet the requirements;

[0151] When filtering spots, the spots are filtered based on the brightness difference between the spots and the regions of interest. The brightness of the spots is the maximum brightness or minimum brightness of the spots, and the brightness of the regions of interest is the average brightness of the regions of interest.

[0152] Furthermore, the scale optimization of the spots in the region of interest includes:

[0153] Performing clarity statistics on the acquired images in time series to obtain multiple clarity value groups;

[0154] Processing the clarity of the region of interest using the obtained clarity value group;

[0155] Re-identifying spots in the region of interest processed by sharpness and calculating the probability of the spots being re-identified;

[0156] The spots are sorted according to the probability and a set number or a set proportion of spots are retained in the order sequence, and the probability of the spots being re-determined decreases in the order sequence.

[0157] Furthermore, optimizing the position of the spots within the region of interest includes:

[0158] Setting multiple analysis reference lines in the spot area, wherein the multiple analysis reference lines are parallel to each other;

[0159] Use the pixel points on the analysis reference line to establish an analysis curve and obtain the lowest point of the analysis curve;

[0160] Use the lowest point of the obtained analytical curve to establish a central reference straight line;

[0161] Obtaining multiple central reference lines and determining a convergence point or convergence area of ​​the central reference lines;

[0162] After moving the center of the spot to the collection point or collection area, the edge of the spot is re-collected and the edge value of the extracted spot is determined;

[0163] The blobs are reconstructed using edge values.

[0164] Furthermore, similarity analysis of the movement trajectory of the characteristic yak object in the time dimension includes:

[0165] An analysis grid is established with the location of the characteristic yak object as the reference point, and the number of edges of each grid in the analysis grid is the same;

[0166] Calculate the average length of the grid edges;

[0167] The outliers are determined based on the average value, and the edge lengths of the grids corresponding to the outliers are all greater than the average value;

[0168] The outlier time of outliers is counted in the time dimension.

[0169] Furthermore, the behavioral feature analysis of the movement trajectory of the characteristic yak object in the time dimension includes:

[0170] Determine the head identification feature of the characteristic yak object, where the head identification feature is the horn feature;

[0171] A width variation feature group of head recognition features is established in the time series, and the width variation feature group includes area value and time length of area value;

[0172] Compare the similarities between width variation feature groups;

[0173] Normal behavior individuals and abnormal behavior individuals in the characteristic yak objects are determined according to the similarity.

[0174] In one example, the unit in any of the above devices can be one or more integrated circuits configured to implement the above method, such as: one or more application specific integrated circuits (ASICs), or, one or more digital signal processors (DSPs), or, one or more field programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.

[0175] For another example, when the units in the device can be implemented in the form of a processing element scheduling program, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor that can call programs. For another example, these units can be integrated together and implemented in the form of a system-on-a-chip (SOC).

[0176] Various objects such as various messages / information / equipment / network elements / systems / devices / actions / operations / processes / concepts that may appear in the present invention are named. It can be understood that these specific names do not constitute a limitation on the relevant objects. The names assigned may change with factors such as scenarios, contexts or usage habits. The understanding of the technical meaning of the technical terms in the present invention should be mainly determined from the functions and technical effects embodied / executed in the technical solutions.

[0177] Those skilled in the art will clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.

[0178] In the several embodiments provided by the present invention, it should be understood that the disclosed systems, devices and methods can be implemented in other ways. For example, the device embodiments described above are merely illustrative. For example, the division of the units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0179] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

[0180] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professionals and technicians can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of the present invention.

[0181] It should also be understood that in various embodiments of the present invention, the terms "first," "second," and so on are merely used to indicate that multiple objects are distinct. For example, the first time window and the second time window are merely used to indicate different time windows. These terms should not affect the time windows themselves, and the terms "first," "second," and so on should not limit the embodiments of the present invention in any way.

[0182] It should also be understood that in the various embodiments of the present invention, unless otherwise specified or there is any logical conflict, the terms and / or descriptions between different embodiments are consistent and can be referenced to each other, and the technical features in different embodiments can be combined to form new embodiments according to their inherent logical relationships.

[0183] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, or the part that contributes to the prior art, or the part of the technical solution, can be embodied in the form of a software product. The computer software product is stored in a computer-readable storage medium and includes several instructions for enabling a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the method described in each embodiment of the present invention. The aforementioned computer-readable storage medium includes various media that can store program codes, such as a USB flash drive, a mobile hard disk, a read-only memory (ROM), a random access memory (RAM), a magnetic disk, or an optical disk.

[0184] The present invention also provides a behavior analysis system for yak breeding, the system comprising:

[0185] one or more memories for storing instructions; and

[0186] One or more processors are used to call and execute the instructions from the memory to perform the method as described above.

[0187] The present invention also provides a computer program product, which includes instructions. When the instructions are executed, the terminal device and the network device perform operations of the terminal device and the network device corresponding to the above method.

[0188] The present invention also provides a chip system, which includes a processor for implementing the functions involved in the above content, such as generating, receiving, sending, or processing the data and / or information involved in the above method.

[0189] The chip system may be composed of chips, or may include chips and other discrete devices.

[0190] The processor mentioned in any of the above may be a CPU, a microprocessor, an ASIC, or one or more integrated circuits for executing a program for controlling the above-mentioned feedback information transmission method.

[0191] In one possible design, the chip system also includes a memory for storing necessary program instructions and data. The processor and the memory can be decoupled and provided on different devices, respectively, and connected via wired or wireless means to support the chip system in implementing the various functions of the above embodiments. Alternatively, the processor and the memory can be coupled on the same device.

[0192] Optionally, the computer instructions are stored in a memory.

[0193] Optionally, the memory is a storage unit within the chip, such as a register, cache, etc. The memory can also be a storage unit within the terminal located outside the chip, such as ROM or other types of static storage devices that can store static information and instructions, RAM, etc.

[0194] It can be understood that the memory in the present invention can be a volatile memory or a non-volatile memory, or can include both volatile and non-volatile memories.

[0195] The non-volatile memory may be ROM, programmable ROM (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), or flash memory.

[0196] Volatile memory can be RAM, which is used as an external cache memory. There are many different types of RAM, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double data rate synchronous DRAM (DDR SDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), and direct memory bus RAM.

[0197] The embodiments of this specific implementation method are all preferred embodiments of the present invention and are not intended to limit the scope of protection of the present invention. Therefore, any equivalent changes made based on the structure, shape, and principle of the present invention should be included in the scope of protection of the present invention.

Claims

1. A behavioral analysis method for yak breeding, characterized in that: include: In response to the acquired image, extracting a yak object in the image; Marking the yak object to obtain a yak object with an identity mark, which is recorded as a feature yak object; Record the movement trajectory and behavioral characteristics of the characteristic yak object; Conduct similarity analysis on the movement trajectory and behavior characteristics of the characteristic yak object in the time dimension to obtain similarity analysis results; According to the similarity analysis results, the characteristic yak objects were grouped into normal group and abnormal group; The marking process for the yak object includes: The image corresponding to the yak object is divided into regions to obtain multiple sub-images, and the sub-images are arranged in the form of an MxN matrix, where M and N are both natural numbers greater than zero; Extract spots in a sub-image, and when there are no spots in a sub-image, fuse the sub-image into an adjacent sub-image with spots; Create regions of interest based on blobs; Optimize the spots within the region of interest, including position optimization and scale optimization; Using the optimized spot regions, a feature group is established, and the feature group is used as a marker for the marking process; Optimizing the size of the spots within the region of interest involves: Performing clarity statistics on the acquired images in a time series to obtain multiple clarity value groups; Processing the clarity of the region of interest using the obtained clarity value group; Re-identifying spots in the region of interest processed by sharpness and calculating the probability of the spots being re-identified; Sort the spots according to the probability and retain a set number or a set proportion of spots in the sequence, and the probability of the spots being re-determined in the sequence decreases; Optimizing the position of the spots within the region of interest involves: Setting multiple analysis reference lines in the spot area, wherein the multiple analysis reference lines are parallel to each other; Use the pixel points on the analysis reference line to establish an analysis curve and obtain the lowest point of the analysis curve; Use the lowest point of the obtained analytical curve to establish a central reference straight line; Obtaining multiple central reference lines and determining a convergence point or convergence area of ​​the central reference lines; After moving the center of the spot to the collection point or collection area, the edge of the spot is re-collected and the edge value of the extracted spot is determined; The blobs are reconstructed using edge values.

2. The behavioral analysis method for yak breeding according to claim 1, characterized in that: Extracting spots in a sub-image involves: Performing noise reduction on the sub-image, the noise reduction process includes representing each row or column of pixels in the sub-image with a curve, and then performing equal-length smoothing on the curve; Use the extraction operator to extract spots on the sub-image; Use the regions of interest created based on the spots to filter the spots and retain the spots that meet the requirements; When filtering spots, the spots are filtered based on the brightness difference between the spots and the regions of interest. The brightness of the spots is the maximum brightness or minimum brightness of the spots, and the brightness of the regions of interest is the average brightness of the regions of interest.

3. The behavior analysis method for yak breeding according to claim 1, characterized in that: The similarity analysis of the movement trajectory of the characteristic yak object in the time dimension includes: An analysis grid is established with the location of the characteristic yak object as the reference point, and the number of edges of each grid in the analysis grid is the same; Calculate the average length of the grid edges; The outliers are determined based on the average value, and the edge lengths of the grids corresponding to the outliers are all greater than the average value; The outlier time of outliers is counted in the time dimension.

4. The behavior analysis method for yak breeding according to claim 1 or 3, characterized in that: The behavioral feature analysis of the movement trajectory of the characteristic yak object in the time dimension includes: Determine the head identification feature of the characteristic yak object, where the head identification feature is the horn feature; A width variation feature group of head recognition features is established in the time series, and the width variation feature group includes area value and time length of area value; Compare the similarities between width variation feature groups; Normal behavior individuals and abnormal behavior individuals in the characteristic yak objects are determined according to the similarity.

5. A behavior analysis system for yak breeding, characterized in that: The system comprises: one or more memories for storing instructions; and One or more processors, configured to call and execute the instructions from the memory to perform the method according to any one of claims 1 to 4.

Images