JP7504036B2 - Pig breeding support device, pig breeding support method, and pig breeding support program - Google Patents

Description

The present invention relates to a pig breeding support device, a pig breeding support method, and a pig breeding support program.

Systems that detect abnormalities in livestock are known. For example, various sensors are attached to individual livestock, and if the output is a value that is impossible in a healthy state, the individual is determined to be abnormal (see, for example, Patent Document 1).

JP 2017-201930 A

When raising pigs, the method of group rearing is often adopted, generally in cages called pens or separate areas. In group rearing, pigs that feel very stressed may attack and injure other pigs. Such pigs are particularly prone to biting the tails of other pigs. Keepers need to provide treatment for pigs whose tails have been bitten as soon as possible and investigate the cause of the stress.

It is not realistic to attach a sensor to each pig in a pen to detect abnormalities, and it is also difficult to prepare the huge amount of analysis data required for this purpose. In particular, it is not easy to detect tail injuries from abnormal behavior. Furthermore, pig farms often have multiple pens in their pig houses, making it difficult for keepers to keep an even watch over each individual pig.

The present invention was made to solve these problems, and aims to provide a pig breeding support device that can notify breeders in a timely manner, without excessive cost or labor, whether any pigs raised in a pen have injured tails.

The pig breeding support device according to the first aspect of the present invention includes an acquisition unit that acquires image data of an image captured by a camera installed facing a pen in which pigs are bred in groups, and a detection unit that extracts the pig's buttocks from the image of the image data and detects whether the pig is injured based on the red area present in the buttocks.

The pig breeding support method according to the second aspect of the present invention includes an acquisition step of acquiring image data of an image captured by a camera installed facing a pen in which pigs are bred in groups, and a detection step of extracting the pig's buttocks from the image of the image data and detecting whether the pig is injured based on the red area present in the buttocks.

The pig breeding support program in the third aspect of the present invention causes a computer to execute an acquisition step of acquiring image data of an image captured by a camera installed facing a pen in which pigs are bred in groups, and a detection step of extracting the pig's buttocks from the image of the image data and detecting whether or not the pig is injured based on the red area present in the buttocks.

The present invention provides a pig breeding support device that can notify breeders in a timely manner, without excessive cost or labor, whether any pigs raised in a pen have injured tails.

FIG. 1 is a diagram showing an overall picture of a pig farming environment in which a pig raising support device according to an embodiment of the present invention is used. FIG. 2 is a diagram showing the hardware configuration of the pig breeding assistance device and peripheral devices. FIG. 1 is a diagram illustrating the difference between normal pigs and injured pigs in this embodiment. 11 is a diagram illustrating a procedure for extracting a verification area using a learning model. FIG. FIG. 13 is a diagram illustrating a procedure for detecting an injured pig using a learning model. FIG. 13 is a diagram showing an example of the display on a zookeeper terminal that has received a detection notification. FIG. 4 is a flowchart illustrating a processing procedure of a calculation unit. FIG. 11 is a flowchart illustrating a processing procedure of a calculation unit according to another embodiment. FIG. 13 is a diagram showing an overall picture of a pig farming environment in which a pig farming support device according to yet another embodiment is adopted. FIG. 10 is a diagram illustrating a procedure for detecting an injured pig in the embodiment of FIG. 9.

The present invention will be described below through embodiments of the invention, but the invention according to the claims is not limited to the following embodiments. Furthermore, not all of the configurations described in the embodiments are necessarily essential as means for solving the problem.

Figure 1 is a diagram showing an overall picture of a pig farming environment in which a pig breeding support device according to this embodiment is used. The pig farm has multiple pens 301 separated by walls and fences. Each pen 301 houses multiple pigs 302 (e.g., about 10 pigs) that are bred in groups. The number of pigs 302 bred in each pen 301 can be adjusted depending on the breed of pigs 302, the breeding environment, etc.

A camera unit 210 for observing the housed pigs 302 is installed in each pen 301. The camera unit 210 is installed, for example, suspended from the ceiling, facing the pen 301 to be observed so that the entire pen 301 can be imaged from above. The camera unit 210 converts the captured image into image data and transmits it to the server 100 via the network 200. Specifically, a wireless unit 230 installed in the facility is connected to the network 200, and the camera unit 210 can transmit the image data to the server 100 by establishing wireless communication with the wireless unit 230. The network 200 connecting the camera unit 210 and the server 100 may use the Internet or an intranet, or may use short-range wireless communication when the management facility in which the server 100 is installed is installed in a pig farm.

The keeper who takes care of the pig 302 may carry a keeper terminal 220. The keeper terminal 220 is, for example, a tablet terminal or a smartphone, and can transmit and receive various information to and from the server 100 via the wireless unit 230 and the network 200. The keeper can, for example, input breeding records into the keeper terminal 220 and transfer them to the server 100, or can call up breeding records stored in the server 100. In addition, when the keeper terminal 220 receives a detection notification, which will be described later, it displays that fact.

In the management facility, a server 100 is installed as a pig breeding support device. The server 100 is connected to a network 200. The server 100 sequentially acquires image data from a camera unit 210 installed in each pen 301, and detects injured pigs 302 based on the image data. The server 100 can display the results on a display monitor 150 in response to a request from a manager or a zookeeper. The display monitor 150 is, for example, a monitor equipped with a liquid crystal panel. The server 100 is also connected to an input device 160 that accepts operations from the manager or the zookeeper. The input device 160 is composed of a keyboard, a mouse, a touch panel superimposed on the display surface of the display monitor 150, etc.

Now, in the case of pigs being raised in a group in a pen, pigs that feel strong stress may attack and injure other pigs. Such pigs tend to bite the tails of other pigs in particular. Also, the more stressed a pig is, the more frequently it bites the tails of other pigs. The zookeeper must quickly treat the pig whose tail has been bitten and investigate the cause of the stress. However, since pig farming is generally large in scale and pigs are often raised in groups in multiple pens set up in a pig house, it is a task that requires a great deal of effort and skill on the part of the zookeeper to constantly observe a large number of pigs and distinguish blood that changes over time from dirt. The pig breeding support device in this embodiment supports the breeding work by detecting whether there is a pig with an injured tail in the pen 301 and notifying the zookeeper.

When there are multiple pens 301 in a pig farm as shown in FIG. 1, the keeper can reduce a lot of work just by being informed by the pig breeding support device of which pen 301 contains an injured pig. In other words, the keeper can reduce the frequency of observing the pens, and when notified, can simply search for the injured pig in that specific pen. Even an inexperienced keeper can find the injured pig relatively easily among the limited number of pigs in one pen.

Figure 2 shows the hardware configuration of the server 100 and peripheral devices as a pig breeding support device. As described above, the server 100 can be connected to the display monitor 150, the input device 160, the camera unit 210, and the breeder terminal 220.

The server 100 is mainly composed of a calculation unit 110, a memory unit 120, and a communication unit 130. The calculation unit 110 is a processor (CPU: Central Processing Unit) that controls the server 100 and executes programs. The processor may be configured to work in conjunction with a calculation processing chip such as an ASIC (Application Specific Integrated Circuit) or a GPU (Graphics Processing Unit). The calculation unit 110 reads out the pig breeding support program stored in the memory unit 120 and executes various processes related to pig breeding support.

The storage unit 120 is a non-volatile storage medium, and is configured, for example, by a HDD (Hard Disk Drive). In addition to the program that controls and executes the processing of the server 100, the storage unit 120 can store various parameter values, functions, display element data, lookup tables, etc. used for control and calculation. In particular, the storage unit 120 stores a detection neural network 121. When an image captured by the camera unit 210 is input, the detection neural network 121 outputs the number of injured pigs present in the image. This will be described in detail later. The storage unit 120 may be configured by multiple hardware components, and for example, the storage medium that stores the program and the storage medium that stores the detection neural network 121 may be configured by separate hardware components.

The communication unit 130 includes, for example, a LAN unit, and transmits the imaging control signal generated by the calculation unit 110 to the camera unit 210 via the network 200, and passes the image data sent from the camera unit 210 to the calculation unit 110. It also relays the exchange of data between the keeper terminal 220 and the calculation unit 110. The communication unit 130 can also relay the exchange of data and control signals with other external devices. For example, it can also be used when importing update data for the pig breeding support program or the detection neural network 121 from an external server.

The calculation unit 110 also serves as a functional calculation unit that executes various calculations in response to the processing instructed by the pig breeding support program. The calculation unit 110 can function as an acquisition unit 111 and a detection unit 112. The acquisition unit 111 mainly acquires image data of images captured by the camera unit 210 and passes it over to the detection unit 112. The detection unit 112 mainly extracts the pig's buttocks from the image data received from the acquisition unit 111 and detects whether or not the pig is injured based on the proportion of the red area relative to the buttocks.

Figure 3 is a diagram explaining the difference between a normal pig and an injured pig in this embodiment. The pig breeding support device in this embodiment detects an injured pig with an injured tail. As described above, a pig whose tail has been bitten becomes injured and bleeds. The blood spreads and adheres to the buttocks.

Figure 3(A) shows examples of the backs of normal pigs, and in all cases no blood is found on the buttocks. In this embodiment, such pigs are considered to be normal pigs even if blood is found on other parts of the body. Figure 3(B) shows examples of the backs of injured pigs, and in all cases blood is found on the buttocks. In the image, blood appears as a red area on the buttocks, so the detection unit 112 uses this red area near the buttocks as a clue to detect whether the pig in the image is an injured pig. Specifically, the detection unit 112 sequentially performs an extraction process to extract an area in which the back of the pig is captured from the overhead image looking down on the pen 301 as a verification area, and a detection process to detect whether the pig captured in the verification area is an injured pig.

Figure 4 is a diagram explaining the procedure for extracting a verification area using a learning model. When the detection unit 112 receives an acquired image, which is an overhead image of the pen 301, from the acquisition unit 111, the detection unit 112 inputs the acquired image to the first neural network 121a constituting the detection neural network 121, which is the learning model.

The first neural network 121a is created in advance by supervised learning, in which a rectangular area surrounding a pig showing its back in an overhead image looking down on the pen 301 is used as a correct answer, and a considerable amount of training data linked to the overhead image is provided for learning. The training data is created, for example, by an operator extracting a pig showing its back, where the buttocks can be seen, for each pig that appears in the overhead image, and surrounding it with a rectangular area to provide the correct answer. When there are multiple target pigs in a single overhead image, the operator designates each of them as the correct answer by surrounding them with a rectangular area.

The first neural network 121a, which is the learning model thus generated, is incorporated into the server 100, which is a pig breeding support device, as a part of the detection neural network 121 and is used. Specifically, for example, if eight pens 301 to be observed are provided in a pig farm, image data is sequentially sent to the server 100 from eight camera units 210 that overlook each of the pens 301. Images img ₁ to img ₈ of the image data are sequentially input to the first neural network 121a. The first neural network 121a executes a calculation process each time an image is input, and extracts an area in which a pig is seen from behind as a verification area from the image. In this way, for example, as shown in the figure, verification areas "area01", "area02", ... are extracted from the image img ₁ of the first pen.

Fig. 5 is a diagram for explaining the procedure for detecting an injured pig using a learning model. When the first neural network 121a extracts a verification area, the detection unit 112 cuts out each area to generate an area image, and inputs the area image to the second neural network 121b constituting the detection neural network 121. When multiple area images are generated, they are input to the second neural network 121b in sequence. Note that Fig. 5 shows an example of the area image of the verification area extracted from the image img ₁ of the image data shown in Fig. 4.

The second neural network 121b is created in advance by supervised learning, in which a considerable amount of training data is provided to the network, and the network is trained by linking correct answers to red areas around the buttocks that are likely to be caused by blood if they are present, and incorrect answers to incorrect answers if they are not. The training data is created by an operator, for example, by providing correct and incorrect answers while taking into consideration the size of the red areas relative to the buttocks and the color that is likely to be caused by blood.

The second neural network 121b, which is the learning model thus generated, is incorporated into the server 100, which is the pig breeding support device, as a part of the detection neural network 121 and is used. Specifically, for example, when four area images (area01, area02, area03, area04) are sequentially input as shown in the figure, the second neural network 121b calculates the probability P _n (P ₁ = 0.023, P ₂ = 0.921, P ₃ = 0.017, P ₄ = 0.146) that the pig in the image is an injured pig for each area. If the calculated probability P _n exceeds a preset threshold (0.85 in the example shown in the figure), the pig in the area image is recognized as an injured pig, and the total number of recognized injured pigs is output as the number of detected pigs (1 in the example shown in the figure). The threshold value is set to a value adjusted in advance by trial and error according to the output accuracy of the learning model and the probability of being an injured pig. It is also possible to set a threshold value according to the degree of injury by classifying injured pigs into classes such as severely injured pigs, moderately injured pigs, and mildly injured pigs according to the degree of injury and having the system learn the classes. When the second neural network 121b outputs a detected number of one or more pigs, the detection unit 112 outputs a detection notification to the keeper terminal 220 to notify the keeper of the presence of an injured pig. The detection notification is accompanied by information regarding the pen in which the injured pig was detected. In this embodiment, the pen number is added as the pen information.

Figure 6 is a diagram showing an example of the display of a keeper terminal that has received a detection notification. As described above, when the detection unit 112 outputs a detection notification, the keeper terminal 220 receives the detection notification and displays the contents of the detection notification on the display panel. Specifically, as shown in the figure, the keeper terminal 220 refers to the pen information added to the detection notification and displays the pen number in which the injured pig was detected (in the example shown in the figure, "Pen 2"). At this time, the number of injured pigs detected is also displayed (in the example shown in the figure, "2"). In addition, if an indoor map of the pen layout of the pig farm is held, the map is also displayed so that the position of the corresponding pen can be recognized. The keeper terminal 220 may also issue a notification sound along with displaying the information in this way.

Next, the processing procedure of the pig breeding support method using the server 100 will be described. FIG. 7 is a flow diagram explaining the processing procedure of the calculation unit 110. The flow starts when an instruction to start observation is received.

In step S101, the detection unit 112 sets a variable n to 1. n represents a pen number, and here, the number of pens in the pig farm is m. Proceeding to step S103, the acquisition unit 111 acquires image data of img _n from the camera unit 210 pointed at the nth pen via the communication unit 130. If n=1, image data of img ₁ is acquired from the camera unit 210 pointed at the first pen. The acquisition unit 111 passes the acquired image data to the detection unit 112.

4 and 5, in step S103, the detection unit 112 inputs the image img _n of the received image data to the first neural network 121a of the detection neural network 121. Then, the output is input to the second neural network 121b of the detection neural network 121, thereby outputting the number of injured pigs captured in the image img _n (the number of detected pigs).

Proceeding to step S104, the detection unit 112 checks whether the number of detected pigs, which is the output of the second neural network 121b, is 1 or more, i.e., whether or not an injured pig is present. If it is confirmed that an injured pig is present, the process proceeds to step S105, where a detection notice is generated and output to the keeper terminal 220. At this time, the pen information is added to indicate that this is the nth pen, and the number of detected pigs, which is the output of the second neural network 121b, is added. After outputting the detection notice, the process proceeds to step S106. If the detection unit 112 confirms in step S104 that no injured pigs are present, it skips step S105 and proceeds to step S106.

In step S106, the detection unit 112 increments the variable n, and proceeds to step S107. When proceeding to step S107, it is determined whether or not the variable n has exceeded the number of pens m in the pig farm. If it is determined that the variable n has not exceeded the number m, the process returns to step S102, and the same process is performed on the incremented variable n. If it is determined that the variable n has exceeded the number m, the process proceeds to step S108.

When the detection unit 112 proceeds to step S108, it checks whether or not an instruction to end has been received. An instruction to end is generated when an instruction to end is received from the administrator via the input device 160, or when a condition preset in the program is satisfied. If it is determined that an instruction to end has not been received, the process returns to step S101 after an interval according to a preset cycle. If it is determined that an instruction to end has been received, the process ends.

According to the embodiment described above, the pig breeding support device is configured to sequentially acquire image data of each of the multiple pens 301, detect injured pigs from the images, and notify the keeper each time an injured pig is detected. However, depending on the management environment of the pig farm, it may be preferable for the keeper to be notified at a specific time or period rather than each time. In such cases, it is preferable to accumulate the detection results until a specified observation time has elapsed, and then notify the accumulated results together. Figure 8 is a flow diagram explaining the processing procedure of the calculation unit 110 according to such another embodiment. Note that the same processes as those in Figure 7 may be omitted from the description by assigning the same step numbers as those in Figure 7.

In this embodiment shown in Fig. 8, the calculation unit 110 has a counting unit as a functional calculation unit. The counting unit counts the number of injured pigs detected during the set observation time. Specifically, the counting unit counts the number of injured pigs using counters _C1 to _Cm prepared for each pen. The flow in Fig. 8 starts in synchronization with the start of the set observation time.

At the start of observation, as an initial process, in step S201, the counting unit starts measuring time using a time timer T, resets all counters _Cn that count the number of injured pigs housed in the nth pen, and then proceeds to step S101.

In step S104, the detection unit 112 checks the number of injured pigs detected in the image img _n , and if it is confirmed that there is one or more injured pigs, the process proceeds to step S205. In step S205, the detection unit 112 passes the number of detected pigs to the counting unit. The counting unit updates C _n by adding the number of detected pigs received from the detection unit 112 to the counter value of C _n at that time, and then proceeds to step S106. If it is confirmed in step S104 that there are no injured pigs, the process skips step S205 and proceeds to step S106.

In step S107, the detection unit 112 judges whether or not the variable n has exceeded the number of pens m in the pig farm, and if it is judged that the variable n has not exceeded the number, the process returns to step S102, and executes the same process on the incremented variable n. If it is judged that the variable n has exceeded the number, the process proceeds to step S208 ..., where the counting unit judges whether or not the elapsed time timer T has exceeded the set observation time _Tc . If it is judged that the variable n has not exceeded the number, the process returns to step S101 after an interval according to a preset cycle. If it is judged that the variable n has exceeded the number, the process proceeds to step S209.

When the process proceeds to step S209, the counting unit passes the counter values from _C1 to _Cm at that time to the detection unit 112. The detection unit 112 checks these counter values, generates a detection notification for the pens showing a counter value of 1 or more, and outputs the detection notification to the keeper terminal 220. When the keeper terminal 220 receives multiple detection notifications, it may refer to the added pen information and display the fact that an injured pig has been detected by switching the screen for each pen, or may display the results in a list. Note that the detection unit 112 may generate a detection notification even if an injured pig is not detected in any pen, and display the fact on the keeper terminal 220. When the detection unit 112 finishes the process of step S209, the calculation unit 110 ends the series of processes. When performing continuous observation, the process starts again from step S201.

When injured pigs are repeatedly detected during the set observation time, the same injured pig may be detected multiple times and counted multiple times. This is fine if the main purpose is to check the presence or absence of injured pigs, but if you want to count the number of injured pigs accurately, you can have the counting unit determine whether they are the same pig or not, and if it is determined that they are the same pig, remove them from the count. Specifically, the area image in which the injured pig detected by the detection unit 112 is captured is stored in the memory unit 120, and when a new injured pig is detected, the area image in which the injured pig is captured is compared with the area image stored in the memory unit to determine whether they are the same pig or not. More specifically, the shapes of the red areas of the injured pig captured in each area image are compared by pattern matching, and if they are determined to match, it is determined that the two are the same pig.

In each of the above-described embodiments, one camera unit 210 is installed in each pen 301, but a camera unit that overlooks multiple pens 301 together may be installed. In that case, the acquisition unit 111 divides the image acquired from the camera unit along the boundaries of each pen 301 and sequentially transfers each divided image to the detection unit 112. Conversely, multiple camera units 210 may be installed for one pen 301. For example, by installing a camera unit 210 that captures an image of the pig 302 in the pen 301 from the side, false detection of an injured pig can be reduced. Furthermore, by installing multiple camera units 210 for one pen 301, the opportunities to capture images of the backs of pigs moving around in the pen 301 increase, making it possible to detect injured pigs with greater accuracy.

In each of the embodiments described above, injured pigs are detected based on an overhead image of each pen 301, but injured pigs may also be detected based on an image captured by pointing a camera at a specific location in the pen with an angle of view that captures approximately the back of one pig. Figure 9 shows an overall view of a pig farming environment that employs a pig breeding support device according to yet another embodiment of this invention. Elements similar to those in Figure 1 are given the same reference numbers and their description will be omitted.

The camera unit 210 captures an image of a specific position in the pen 102 from above. The camera unit 210 is installed so as to capture an image within a range of an angle of view Fr, for example, capturing an image of the pig 302 from behind while feeding with its head immersed in the feeding container 303. The camera unit 210 installed in this manner can capture an image in which the rump of the pig 302 while feeding is directly facing the camera.

The camera unit 210 does not have to be located near the feeding container 303 as the predetermined position, and may be located at any location that is expected to capture images of the buttocks of the pigs 302 moving around in the pen 301 from a frontal position. For example, the predetermined position may be near the water container. Feeding and watering areas are places where the housed pigs 302 come and go one after another, so they are convenient for randomly capturing images of the buttocks of many pigs 302.

Figure 10 is a diagram illustrating the procedure for detecting an injured pig in the embodiment of Figure 9. The acquisition unit 111 acquires captured image data from the camera unit 210, and cuts out the area near the tail of the pig 302 from the captured image to generate a cut-out image. Specifically, the acquisition unit 111 finds the tail area from the target image, for example by performing a matching process with a large number of template images prepared in advance as tail images. The acquisition unit 111 then cuts out the found tail area and its surroundings in a preset image size to generate a cut-out image. The acquisition unit 111 passes the generated cut-out image to the detection unit 112.

The detection unit 112 performs edge processing on the cut-out image to calculate the number of pixels D _L from the center of the tail to the left body boundary and the number of pixels D _R to the right body boundary. As shown in the figure, when D _L =D _R , it is determined that the buttocks are facing forward, and this cut-out image is the buttocks image. In other words, the detection unit 112 extracts an image in which the tail is located at the center of the buttocks in the width direction, and sets it as the buttocks image. The detection unit 112 determines whether D _L and D _R are equal or not, taking into account the allowable error set according to the number of pixels in the width of the cut-out image.

The detection unit 112 extracts the red area and the buttocks area from the buttocks image extracted in this way and calculates their areas. It then calculates (area of red area) / (area of buttocks area) and, if this value exceeds a preset threshold, determines that blood is present in the buttocks and identifies the pig that appears in the captured image as an injured pig. The pig breeding support device can detect injured pigs even when employing such a method.

In the present embodiment described above through several examples, it is assumed that multiple pens 301 are installed in one pigpen, but the observation target may be pens 301 installed across multiple pigpens. The output destination of the detection notification is not limited to the zookeeper terminal 220. For example, a notification light connected to the server 100 via the wireless unit 230 and the network 200 may be installed adjacent to each pen 301, and the detection notification may be output to the notification light. For example, when the server 100 detects an injured pig in the fifth pen, the server 100 transmits a detection notification to the notification light installed adjacent to the fifth pen and turns on the notification light. The detection unit 112 may directly display the detection result on the display monitor 150 connected to the server 100. For example, a list of the number of injured pigs detected in each pen 301 may be displayed on the display monitor 150.

In the above embodiment, the server 100 functions as a pig breeding support device, but the hardware configuration is not limited to this. If the mobile terminal described as the keeper terminal 220 performs the same processing as the server 100, the mobile terminal can function as a pig breeding support device. Also, for example, if the keeper terminal 220 is configured to take on part of the processing of the server 100, a system in which the server 100 and the keeper terminal 220 work together can become a pig breeding support device.

100...server, 110...computing unit, 111...acquisition unit, 112...detection unit, 120...storage unit, 121...detection neural network, 130...communication unit, 150...display monitor, 160...input device, 200...network, 210...camera unit, 220...keeper terminal, 230...wireless unit, 301...pen, 302...pig, 303...feeding container

Claims (7)

An acquisition unit that acquires image data of an image captured by a camera installed facing a pen in which pigs are group-raised;
A pig breeding support device comprising a detection unit that extracts the pig's buttocks from the image of the image data and detects whether or not the pig is injured based on a red area present in the buttocks. The pig breeding support device according to claim 1, wherein the detection unit detects whether the injured pig is present in each of the multiple pens, and when it detects the presence of the injured pig, outputs pen information relating to the detected pen. The pig breeding support device according to claim 1 or 2, wherein the detection unit detects the injured pig using a learning model trained on a teacher image showing a pig with its tail exposed. The pig breeding support device according to any one of claims 1 to 3, further comprising a counting unit that counts the number of injured pigs detected during a set observation time. The pig breeding support device according to claim 4, wherein the counting unit determines whether the injured pigs detected are the same pig based on the shape of the red area, and does not count the injured pigs determined to be the same pig. An acquisition step of acquiring image data of an image captured by a camera installed facing a pen in which pigs are group-housed;
A pig breeding support method comprising a detection step of extracting the pig's buttocks from the image of the image data and detecting whether or not the pig is injured based on a red area present in the buttocks. An acquisition step of acquiring image data of an image captured by a camera installed facing a pen in which pigs are group-housed;
A pig breeding support program that causes a computer to execute a detection step of extracting the pig's buttocks from the image of the image data and detecting whether or not the pig is injured based on a red area present in the buttocks.

Images