Pig behavior rhythm analysis method

An analysis method and behavioral technology, applied in the field of distributed remote communication, can solve problems such as difficulties in pig behavior analysis, and achieve the effect of simple structure and reduced labor intensity

Inactive Publication Date: 2017-01-04
SOUTH CHINA AGRI UNIV
6 Cites 20 Cited by

AI-Extracted Technical Summary

Problems solved by technology

In the past, due to limited conditions, it w...
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
View more

Abstract

The invention discloses a pig behavior rhythm analysis method, which comprises the steps of S1, installing a monitoring camera at the pig production field, and marking pigs so as to differentiate the pigs; S2, acquiring monitoring video of the pig production field in real time, positioning individual pigs, and determining the movement distance of the pigs; S3, judging pig behaviors according to the movement distance of the pigs in unit time; S4, and building a pig behavior rhythm model by combining behavior data of pig individual growth cycle and behavior data between groups. The pig behavior rhythm analysis method is simple in structure, convenient, easy to use, and suitable for being deployed and applied in a large scale to the breeding industry.

Application Domain

Technology Topic

Surveillance cameraGrowth cycle +3

Image

  • Pig behavior rhythm analysis method
  • Pig behavior rhythm analysis method
  • Pig behavior rhythm analysis method

Examples

  • Experimental program(1)
  • Effect test(1)

Example Embodiment

[0041] Example
[0042] like Figure 1-Figure 4 Shown, a kind of pig behavior rhythm analysis method, comprises the steps:
[0043] S1 Install a monitoring camera at the pig production site. The pig production site is within the field of view of the camera, and the pigs are marked to distinguish the pigs;
[0044] S2 as figure 2 As shown, the real-time monitoring video of the pig production site is obtained. The module first divides the image to obtain a binary image, and then fits the pig ellipse to locate the individual pig and determine the moving distance of the pig;
[0045] Specifically:
[0046] Using the reference object method, specifically, placing a calibration plate on the pig production site to obtain the mapping relationship between the image pixel distance and the actual distance, and then applying the method based on the combination of color features and morphological features to segment the pigs with mild adhesion and adopt The least squares method is used to fit the ellipse to obtain the individual pigs in each frame of image, and determine the moving distance of the pigs, that is, the displacement.
[0047] S3 judges the pig's behavior according to the pig's movement distance per unit time; the details are as follows: image 3 and Figure 4 As shown, the present invention divides the behaviors of pigs into two types, a and b, wherein type a represents basic behaviors, and type b represents complex behaviors. The judgment of pig behavior must be one of the behaviors of type a, and the behavior of type b exists side by side with the behavior of type a as an additional behavior. One, the pig behaviors include type a behaviors and type b behaviors, the type a behaviors include exercise and rest, and the type b behaviors include drinking water, feeding, excreting, pig biting pens, pig tail biting and fighting ;
[0048] The specific pig behavior is described in the following table:
[0049]
[0050]By analyzing the movement distance of the pig per unit time, the movement and rest behaviors of the pigs are classified as Type A; the head position of the pig is analyzed by the Euclidean distance between the center of mass of the pig and the outline point, so that according to the position of the head and the food The location of troughs, water tanks, and defecation areas can be used to judge the type B behaviors of pigs such as feeding, drinking, and excretion;
[0051] Judging type B behaviors such as pig biting pens, pig tail biting, and fighting by pigs' center of mass, movement speed, movement acceleration, distance between groups, and behavior duration. Bayesian training is performed based on a part of the samples in the database, without manual intervention, to obtain the threshold of behavior discrimination, and the dynamic adaptive discrimination module combines the historical database and the current pig movement state to adjust the results of behavior discrimination.
[0052] The specific judgment process is as follows:
[0053] S3.1 First identify the type a behavior:
[0054] Motion recognition, if the pig’s movement displacement is greater than or equal to the set threshold Smove from time t to time t+1, then it is judged that the pig is moving, and it continues to enter into the judgment of type b behavior in S3.3;
[0055] Rest identification, if the pig’s movement displacement is less than the set threshold Sstay from time t to time t+1, it is considered that the pig is resting, and continues to enter the judgment of type b behavior in S3.2;
[0056] S3.2 Assuming that the drinking trough, food trough, and excretion trough have been delineated, and the drinking trough, food trough, and excretion trough are all rectangular, then the head of the pig is included in the rectangle of the drinking trough and the food trough, and it is determined that the pig It is only for drinking or feeding behavior, eliminating the error that the tail of the pig is judged as drinking or feeding behavior within the rectangle. On the other hand, only when the tail of the pig is included in the rectangle of the excretion trough can it be judged as the excretion behavior, which eliminates the error that the pig's head is judged as the excretion behavior within the rectangle r.
[0057] The preceding behaviors of drinking, feeding and excretion are resting behaviors. That is, only after the pig is judged to be resting, will it continue to judge whether it is drinking, feeding or excreting.
[0058] The specific judgment process is:
[0059] Assume that mCount is the total number of contour edge points of the current pig i, among which points (C x_n t ,C y_n t )(n i t under (x t i ,y t i ) is the coordinates of the current pig's centroid. On the premise of obtaining the complete outline of the pig (that is, the binary image including the complete head, ears and body of the pig), calculate the distance from the centroid to the edge point, and calculate the distance from the pig's centroid to the pig The distance between the contour edge points and the coordinates corresponding to the minimum and maximum values ​​can be obtained to obtain three feature points corresponding to the left ear, nose, and right ear respectively.
[0060] Assuming that the total number of pig area pixels at the current moment is mArea, the rectangular area threshold of the drinking trough is wArea, the rectangular area threshold of the food trough is eArea, and the rectangular area threshold of the excretion trough is pArea, then the pig enters the drinking trough, food trough, excretion Only when the area of ​​the trough is greater than these three thresholds can drinking, feeding and excretion be performed.
[0061] Drinking, feeding, and excreting behaviors all need to last for a certain period of time. Assuming that the time threshold is t, the pigs can only become drinking, feeding, and excreting behaviors when they stay in the drinking trough, food trough, and excretion trough for a time greater than the time threshold.
[0062] S3.3 If the distance between the pig and the fence is less than the distance threshold and the duration is greater than the time threshold, it is considered that the pig is biting the fence;
[0063] Whether the two pigs are in a fight can be determined by the distance between the two pigs' centroids, the speed of movement, the acceleration of the movement, and the duration of the behavior. If the distance between two pigs is less than the distance threshold and the duration is greater than the time threshold, then the two pigs are considered to be fighting.
[0064] If two pigs find the feature points in the pig outline by the distance from the center of mass of the pig to the edge point, such as the three feature points representing the head, and the feature point that is farthest from the center of mass is the feature point representing the tail, then To judge whether there is pig bite disease between two pigs, if
[0065] The distance between the head of a pig and the tail of another pig is less than the set threshold, and the duration is longer than the set threshold, which is pig tail biting.
[0066] Because the manual intervention setting threshold is too limited, the behavioral parameters of individual pigs are different, so the threshold Smove, threshold Sstay, distance threshold and time threshold of the present invention are specifically Bayesian threshold training and dynamic self-adaptive discrimination mechanism. All thresholds for behavior judgment do not use artificially set values, but are obtained directly through Bayesian training on the data set; in order to enhance the accuracy of the algorithm for behavior recognition, it is necessary to add dynamic The self-adaptive behavior category judgment mechanism dynamically and adaptively adjusts the threshold according to individual differences during the long-term analysis of pig behavior. The specific method is as follows:
[0067] ① Bayesian iterative training
[0068] Select a small collection of videos as the training set, and manually divide the training set into eight behavior categories: exercise, rest, drinking water, feeding, excretion, pig biting pen, pig tail biting, and fighting. The training set is used to calculate the (a) prior probability P(s) of this behavior category for a certain behavior category; (b) the prior probability P(c) of each group of motion features appearing in the training set; (c) belongs to The prior probability P(c|s) of each set of motion features for this behavior category. Next, the probability P(s|c) of this behavior category can be calculated by Bayesian rule:
[0069] P ( s | c ) = P ( c | s ) * P ( s ) P ( c )
[0070] After the possibility of each group of motion features belonging to a certain behavior category is determined, the probability P(s|c)>Tmax will be considered as this behavior category. In addition, based on the temporal continuity of the behavior, the two frames before and after P(s|c)>Tmin and Tmin
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

PUM

no PUM

Description & Claims & Application Information

We can also present the details of the Description, Claims and Application information to help users get a comprehensive understanding of the technical details of the patent, such as background art, summary of invention, brief description of drawings, description of embodiments, and other original content. On the other hand, users can also determine the specific scope of protection of the technology through the list of claims; as well as understand the changes in the life cycle of the technology with the presentation of the patent timeline. Login to view more.
the structure of the environmentally friendly knitted fabric provided by the present invention; figure 2 Flow chart of the yarn wrapping machine for environmentally friendly knitted fabrics and storage devices; image 3 Is the parameter map of the yarn covering machine
Login to view more

Similar technology patents

Intelligent fault diagnosis method for gear box

InactiveCN101660969AReduce labor intensityOvercoming the shortcomings of manual diagnosisMachine gearing/transmission testingSystems engineeringMissed diagnosis
Owner:BEIJING UNIV OF TECH

Classification and recommendation of technical efficacy words

  • Simple structure
  • Reduce labor intensity

Microlithographic projection exposure apparatus

InactiveUS20050068499A1Simple structureReliable and low-maintenance operationProjectorsPhotomechanical exposure apparatusCamera lensPhysics
Owner:CARL ZEISS SMT GMBH
Who we serve
  • R&D Engineer
  • R&D Manager
  • IP Professional
Why Eureka
  • Industry Leading Data Capabilities
  • Powerful AI technology
  • Patent DNA Extraction
Social media
Try Eureka
PatSnap group products