KR102462098B1 - System and method for determining the optimal pig shipping time using prediction model based on big data learning - Google Patents

Abstract

Disclosed are a system and method for determining an optimal pig shipping time using a prediction model based on big data learning. The system for determining a pig shipping time comprises: a camera part provided in a pig room in which pigs corresponding to a pre-specified fattening period are reared; a reception part for receiving photo information and weight measurement information for each of pigs from a weight measurement part; and a prediction part for applying the photo information for each of the pigs reared in the pig room to a prediction model pre-trained by machine learning to determine whether there is a candidate pig for determining the presence of an abdominal fat forming pig, if the candidate pig exists, applying a change trend in weight information measured and collected so far for the candidate pig to the prediction model, which has learned the weight gain trend pattern of an abdominal fat forming pig in advance using machine learning, to determine whether the candidate pig is an abdominal fat forming pig, and if the candidate pig is an abdominal fat forming pig, allowing the prediction model to predict when the candidate pig, which is an abdominal fat forming pig, will gain weight up to a predetermined minimum shipping weight in reference to the weight gain trend pattern of an abdominal fat forming pig learned in advance by machine learning and the change trend in weight information measured and collected for the candidate pig so far, and to generate shipping time information.

Description

System and method for determining the optimal pig shipping time using prediction model based on big data learning

The present invention relates to a system and method for determining the optimal pig shipping time using a prediction model based on big data learning.

The present invention is the result of the smart farm multi-ministerial package innovation technology development project (task number: 421029-04, research project name: smart farm farming consulting expert system development) of the Agricultural and Forestry Food Technology Planning and Evaluation Institute and Smart Farm Research and Development Team .

Pigs, which are finisher pigs for producing meat in the pig industry, are born after a gestation period of about 16 weeks.

That is, finisher pigs consume breast milk from sows during the lactation period, which is up to about 4 weeks after birth, and go through a piglet period in which they receive a formulated feed away from sows until about 8 weeks after the feeding period. It goes through a growing period in which muscles are generated by receiving feed with a high protein content until the age of 26, and then goes through a fattening period in which a feed with a high fiber content is supplied so that intramuscular fat adheres to the production of high-quality meat until about the 26th week of life. When the fattening period has elapsed and about 26 weeks of age have elapsed, the finishing pigs are shipped and slaughtered.

The carcass grade and meat quality score of the shipped pigs may vary greatly depending on what kind of feed is supplied during the pig breeding period and how the breeding environment is managed.

In particular, in order to produce high-quality meat, it is very important how pigs are bred and managed during the fattening period, which is about the 22nd to the 26th week of life.

The fat accumulation order of fattening pigs consists of subcutaneous fat, intermuscular fat, intramuscular fat, and abdominal fat in the order of body composition. Generally, based on 100 kg of live weight, abdominal fat is about 5%, subcutaneous fat is 70 to 75%, and muscle Fat is estimated to be about 20%, and intramuscular fat is 1 to 2%.

Here, in the case of finisher pigs having too much abdominal fat, the carcass grade and meat quality score are evaluated very low, and therefore the shipping price is also evaluated very low.

During the fattening period of the pig breeding period, there was no way to determine which pigs among the pigs to be shipped had a large amount of abdominal fat accumulation, although the commercial value was decreasing due to excessive accumulation of abdominal fat in some pigs.

For this reason, pigs are shipped after the specified fattening period for pig breeding has elapsed, but in the case of pigs whose feed intake is accumulating as abdominal fat, it is not possible to secure shipment income as much as the feed cost input during the fattening period, resulting in breeding costs. only cause it to increase.

Korean Patent Registration No. 10-0426947

The present invention estimates whether intramuscular fat is adhered or abdominal fat is excessively formed by the fed feed based on the body type and weight of the pig corresponding to the fattening period, and provides an appropriate shipping time according to the fattening situation of the pig. It is to provide a system and method for determining the optimal pig shipping time using a predictive model based on big data learning that can be used.

According to the present invention, in the case of finisher pigs in which abdominal fat is excessively formed, it is possible to reduce feed costs and secure good meat quality by shipping them at an earlier point in time compared to normal finisher pigs. Big data learning that can improve the profits of livestock farmers An object of the present invention is to provide a system and method for determining the optimal pig release time using a predictive model based on the present invention.

The present invention aims to improve the profits of livestock farmers by collecting pig industry outlook information and the like and using the predicted pig shipment information to ship finisher pigs in a state in which abdominal fat is excessively formed at a time when the shipment price is expected to be high The purpose of this study is to provide a system and method for determining the optimal pig shipping time using a predictive model based on big data learning.

Objects other than the present invention will be easily understood through the following description.

According to one aspect of the present invention, there is provided a computer program stored in a computer-readable medium for performing a method for determining when to release pigs, the computer program causing the computer to perform the following steps, the steps comprising: ) by applying the photo information received for each of the pigs raised in the pig room corresponding to the pre-specified fattening period to the machine-learned predictive model in advance to determine whether there is a candidate pig for determining the presence of abdominal fat-forming pigs step; (b) If the candidate pig exists, the change in weight information measured and collected for the candidate pig is applied to the predictive model in which the weight gain trend pattern of the abdominal fat-forming pig is machine-learned in advance, so that the candidate pig is determining whether the pig is celiac fat forming; and (c) when the candidate pig is a celiac fat forming pig, the weight gain trend pattern of the celiac fat forming pig in which the predictive model is machine-learned in advance and the change in weight information measured and collected so far for the candidate pig. and generating shipment time information by predicting when the candidate pig, which is a celiac fat forming pig, will gain up to a predetermined minimum shipment weight, wherein the minimum shipment weight is a target at which normal pigs will gain weight upon expiration of the fattening period. A computer-readable medium, characterized in that the weight is relatively small compared to the optimal shipping weight, which is the weight, and the shipping date information of the celiac fat-forming pig is pre-limited to be designated as a date after the start of the fattening period and before expiration. A computer program stored in the .

In the step (a), the predictive model interprets (a-1) the height value of the pig and the height value of the pig's trunk from the side photo information among the received photo information, and analyzes the width of the pig's trunk from the top photo information. and then calculating a body shape information value that is a calculated value obtained by dividing a pig's trunk circumference, which is estimated by the trunk height value and the trunk width value, by the height value of the pig; and (a-2) determining whether a pig having the calculated body type information value equal to or greater than a predetermined threshold value is the candidate pig for determining the presence of abdominal fat-forming pigs.

In step (b), the predictive model determines that if the change in weight information measured and collected so far for the candidate pig corresponds to the weight gain trend pattern of the abdominal fat-forming pig previously machine-learned, the candidate pig forms abdominal fat. It can be considered a pig.

The computer program may further include the step of (d) generating a pig shipment report on a shipment time for each of the pigs raised in the pig room corresponding to the fattening period. Here, in the pig shipment report generated in step (d), i) in the case of normal pigs, the optimal shipping weight is reached by the predictive model that machine-learned the end of the fattening period or the normal pig weight gain trend pattern in advance. The time point may be included as information on the shipping time, and ii) in the case of a pig with abdominal fat formation, the time point at which the weight increases to the minimum shipping weight may be included as information on the shipping time.

If it is determined that the weight cannot be increased to the minimum shipment weight until the end of the fattening period based on the weight gain trend pattern of the abdominal fat-forming pigs, the pig shipment report generated in step (d) includes the breeding of the candidate pigs at the current time. It may also include information suggesting that it be stopped.

The computer program may include: extracting forecast information on pig shipments by day from pig industry forecast information collected using a web crawler function; and selecting, as recommended shipment days, days with relatively less pig shipments compared to the previous day or the weekly average value, with reference to the extracted forecast information on pig shipments by day. Here, the pig shipment report further includes information on the selected recommended shipment dates, and the recommended shipment dates included in the pig shipment report are the expiration date of the fattening period from the time the candidate pig has gained weight to the minimum shipment weight. It may be limited to the days falling within the range up to.

According to another aspect of the present invention, in the pig shipping time determination system, the photo information and weight measurement information for each of the pigs from the camera unit and the weight measurement unit provided in the pig room in which the pigs corresponding to the pre-specified fattening period are raised. a receiving unit for receiving; and applying photo information for each of the pigs raised in the pig room to a machine-learned prediction model in advance to determine whether there is a candidate pig for determining the presence of abdominal fat forming pig, and if the candidate pig exists, By applying the change in weight information measured and collected so far for the candidate pig to the predictive model in which the weight gain trend pattern of the abdominal fat forming pig is machine-learned in advance, it is determined whether the candidate pig is the abdominal fat forming pig, When the candidate pig is a celiac fat forming pig, the prediction model refers to the weight gain trend pattern of celiac fat forming pig machine-learned in advance and the change trend in weight information measured and collected for the candidate pig so far, and a prediction unit for generating shipment time information by predicting when the candidate pig, which is a forming pig, will gain up to a predetermined minimum shipment weight, wherein the minimum shipment weight is a target weight to which a normal pig will gain upon expiration of the fattening period. It is a relatively small weight compared to the weight, and the shipment date information of the celiac fat-forming pig is provided, characterized in that it is limited in advance to be designated as a date before expiration after the start of the fattening period.

In order to determine whether a candidate pig exists, the predictive model interprets the height value of the pig and the trunk height value of the pig in the side photo information among the received photo information, and analyzes the trunk width value of the pig in the top photo information. thereafter, calculating a body type information value that is a calculated value obtained by dividing a pig's trunk circumference, which is estimated by the trunk height value and the trunk width value, by the height value of the pig; and determining that a pig having the calculated body shape information value equal to or greater than a predetermined threshold value is the candidate pig for determining whether or not the abdominal fat forming pig exists.

The prediction model is implemented to determine that the candidate pig is an abdominal fat-forming pig when the change in weight information measured and collected so far for the candidate pig corresponds to the weight gain trend pattern of the abdominal fat-forming pig previously machine-learned. can

The pig shipment time determination system may further include a report generation unit that generates a pig shipment report regarding the shipment time for each of the pigs raised in the pig room corresponding to the fattening period. In the pig shipment report, i) in the case of normal pigs, the expiration time of the fattening period or the time when the optimal shipment weight is reached by the predictive model that machine-learned the normal pig weight gain trend pattern in advance is included as information on the shipment time. and ii) in the case of a pig with abdominal fat formation, the time point at which the weight is increased to the minimum shipment weight may be included as information about the shipment time.

If it is determined that the weight cannot be increased to the minimum shipment weight until the end of the fattening period based on the weight gain trend pattern of the abdominal fat-forming pigs, the pig shipment report includes information suggesting to stop breeding the candidate pigs at the current time. can be included.

The pig shipment time determination system includes: a web crawler for collecting pig industry outlook information using a web crawler function; and extracting forecast information on daily pig shipments from the collected pig industry forecast information, and referring to the extracted forecast information on daily pig shipments, recommending a day with relatively small pig shipments compared to the previous day or weekly average value. It may further include an analysis unit for selecting things. Here, the report generating unit further includes information on recommended shipment dates selected by the analysis unit in the pig shipment report, and the recommended shipment dates included in the pig shipment report are from the point in time when the candidate pig has increased to the minimum shipment weight. It may be limited to days falling within the range up to the expiration date of the fattening period.

Other aspects, features and advantages other than those described above will become apparent from the following drawings, claims, and detailed description of the invention.

According to an embodiment of the present invention, it is estimated whether intramuscular fat is adhered or abdominal fat is excessively formed by the fed feed based on the body shape and weight of the pig corresponding to the fattening period, and appropriate according to the fattening situation of the pig. It has the effect of suggesting the time of shipment.

In addition, in the case of finisher pigs in which abdominal fat is excessively formed, it is possible to reduce feed costs and secure good meat quality by shipping them at an earlier point in time compared to normal finisher pigs.

In addition, by collecting pig industry forecast information, etc. and using the predicted pig shipment information, it is proposed to ship finisher pigs with excessive abdominal fat at the time when the shipment price is expected to be high, thereby improving the profits of livestock farmers. There is an effect.

The effects obtainable in the present invention are not limited to the above-mentioned effects, and other effects not mentioned may be clearly understood by those of ordinary skill in the art from the following description. will be.

1 is a schematic block diagram of a system for determining the time of shipment of pigs according to an embodiment of the present invention.
Figure 2 is a view illustrating a pig room in which pigs corresponding to the fattening period according to an embodiment of the present invention are bred.
3 is a view for explaining a process of calculating a body shape information value of a pig according to an embodiment of the present invention.
4 is a graph showing the difference in weight increase between normal pigs and abdominal fat-forming pigs during the fattening period according to an embodiment of the present invention.
Figure 5 is a flow chart showing a method for determining the time of shipment of pigs according to an embodiment of the present invention.

Since the present invention can have various changes and can have various embodiments, specific embodiments are illustrated in the drawings and described in detail in the detailed description. However, this is not intended to limit the present invention to a specific embodiment, it should be understood to include all modifications, equivalents and substitutes included in the spirit and scope of the present invention.

1 is a schematic block diagram of a system for determining the time of shipment of pigs according to an embodiment of the present invention, and FIG. 2 is a diagram illustrating a pig farm in which pigs corresponding to the fattening period are bred according to an embodiment of the present invention. . 3 is a diagram for explaining a process of calculating a body type information value of a pig according to an embodiment of the present invention, and FIG. 4 is a weight increase of normal pigs and abdominal fat-forming pigs during the fattening period according to an embodiment of the present invention. This is a graph showing the difference.

Here, a normal pig means a pig in a state in which muscle and intramuscular fat are formed during the fattening period (eg, the period from 22 weeks to 26 weeks after birth of the pig) as intended, and abdominal fat-forming pigs are intended Contrary to this, it refers to pigs in a state in which abdominal fat is mainly formed during the fattening period.

Of course, the distinction between normal pigs and celiac fat-forming pigs does not depend on whether celiac fat is not produced at all. That is, even normal pigs can produce an appropriate amount of abdominal fat during the breeding process until shipment, and normal pigs and pigs with abdominal fat can produce abdominal fat at a level that can affect marketability, such as carcass grade and meat quality grade. It should be understood as being distinguished by whether or not

Referring to FIG. 1 , the pig shipping time determining system includes a shipping time determining device 110 , one or more individual recognition units 131 installed in pig rooms in which pigs corresponding to a fattening period are raised, a camera unit 133 , and A weight measurement unit 135 may be included.

The shipment time determination apparatus 110 may include a receiver 111 , a predictor 113 , a report generator 115 , a storage 117 , and a controller 119 . In addition, as will be described later, the shipment time determination apparatus 110 may further include a web crawler 121 and an analysis unit 123 .

The receiving unit 111 is connected to the object recognition unit 131, the camera unit 133, and the weight measurement unit 135 installed in the pig room corresponding to the fattening period by wire or wireless communication, Receive photo information and weight measurement information, respectively.

As illustrated in FIG. 2 , at least one camera unit 133a , 133b and a weight measurement unit 135 may be installed in the pig room 210 in which pigs in the fattening period are bred.

A plurality of feeders 220 may be installed in the pig room 210, and each feeder 220 is installed at a height at which a pig can ingest feed in a standing posture, and one pig picks up and feeds it. A partition 215 is installed to correspond to the feeding table 220 so that the feed supplied to the bed 220 can be ingested.

Camera parts 133a and 133b are installed to enter the area partitioned by the partition 215 and photograph the pigs ingesting the feed supplied to the feeding table 220 in a standing posture. At this time, the one or more first camera units 133a are installed to generate top photo information (refer to FIG. 3 ( b )) of a pig in a standing posture from the top, and at least one second camera unit is installed in a standing posture from the side. It may be installed to generate side photo information (refer to (a) of FIG. 3) photographing a pig.

Here, the camera units 133a and 133b may be cameras capable of acquiring depth information together when photographing a pig, for example.

In addition, the weight measuring unit 135 enters the area partitioned by the partition 215 and measures the weight of the pig eating the feed supplied to the feeding table 220 in a standing posture. It can be installed on the floor.

Also, although not shown, an entity recognition unit 131 for generating entity recognition information of a pig entering an area partitioned by a partition 215 may be disposed around the feeding table 220 . The object recognition unit 131 may be, for example, an RFID reader for recognizing an RFID tag attached to a pig.

In this way, in the process of a pig entering the pig room 210 and ingesting feed in a standing posture, individual recognition information of a pig ingesting feed may be generated, and photo information generated in relation to the pig and The weight measurement information may be managed corresponding to the entity recognition information.

Referring back to FIG. 1 , the prediction unit 113 applies the breeding history data of each pig (eg, photo information for each pig, weight measurement information, etc.) to the machine-learned prediction model in advance so that the corresponding pig is a normal pig. Alternatively, it can be determined whether it is a celiac fat-forming pig, and information about an optimal shipping time in the celiac fat forming pig can be generated.

Breeding history data of each pig is generated by the camera unit 133 and the weight measurement unit 135 of the pig room in which the corresponding pig belonging to the fattening period is raised, received through the receiving unit 111, and stored in the storage unit 117. have.

The predictive model, for example, is the learning data using photo information continuously taken of pigs corresponding to the fattening period, ultrasound data measuring the body fat distribution of the pigs at each time point, and weight information continuously measured during the fattening period of the pigs. can be machine-learned, and when photo information and weight data for an arbitrary pig are received in the pig room 210, based on artificial intelligence to determine whether it is a pig with abdominal fat formation based on the body shape information value and the weight change trend of the pig It can be implemented as a built-in AI module of

In addition, the predictive model is based on machine learning using learning data about changes in body shape and weight of pigs with abdominal fat during the fattening period, and the weight of pigs with abdominal fat in the pig room 210 for breeding pigs during the fattening period is It may be implemented to further generate predictive information about the time of reaching a predetermined minimum shipping weight.

The predictive model can be learned, for example, by using, as learning data, breeding history data, which are previously accumulated photo information, ultrasound data, and weight information, for pigs corresponding to the fattening period in the corresponding livestock farm, as well as other livestock breeding. Breeding history data generated at a farm or collected through the Internet may also be set to be learned as learning data.

In addition, the predictive model is, for example, a few-shot learning method that can learn with only a small amount of training data, or a generative model such as a generative adversarial network (GAN) with a small amount of training data. It can be implemented to learn using a method of resolving the lack of learning data by increasing similar data generated in other sites in the same environment.

Predictive models include, for example, Fully Convolutional Neural Networks, Convolutional Neural Networks, Recurrent Neural Networks, Restricted Boltzmann Machines, Deep Belief Neural Networks. Network), etc., may be created with one or more deep learning-based models. Of course, it may be implemented as a machine learning technique other than the deep learning technique, or it may be created as a hybrid model in which the deep learning technique and the machine learning technique are combined.

In addition, a method of learning the predictive model may be various, such as supervised learning, unsupervised learning, reinforcement learning, and the like.

The body shape information value for determining whether a pig has abdominal fat formation uses, for example, the pig's height (A) and torso height (B) analyzed from the side picture information of the pig, and the body width (C) analyzed from the top picture information. to be calculated, and may be generated by the prediction model or may be previously designated to be generated by the prediction unit 113 .

That is, as illustrated in (a) of FIG. 3, the outline of the pig is extracted from the side photo information, and the lowest point (eg, the sole position) and the highest point (eg, the position of the sole) based on the extracted outline For example, the distance between the top of the back) is interpreted as the pig's height (A), the second lowest point (e.g., the lowest point of the belly) and the highest, located between the lowest points and consecutive over a predetermined length. The distance between the points can be interpreted as the pig's torso height value (B).

In addition, as illustrated in (b) of FIG. 3 , the outline of the pig is extracted from the top photo information, and the highest point (eg, the tip of the pig's tail) and the lowest point (eg, the pig's nose) The distance in the lateral direction from the predetermined ratio position corresponding to the belly position among the length of the pig's body can be interpreted as the body width value (C) of the pig.

As a method of extracting the outline of a pig from the side photo information and the top photo information of the pig, for example, a YOLO (You Only Look Once) outer rectangle utilization technique may be used. Pre-processing such as noise component removal may be performed to extract the pig's outline.

The body shape information value may be defined as, for example, a calculated value obtained by dividing the circumference length of the body estimated by the body height value B and the body width value C by the height value of the pig. That is, the body shape information value may be defined so that whether or not abdominal fat is formed can be determined in consideration of the pig's height rather than just the circumference of the pig's trunk.

Here, a method of estimating the circumference of the body, which is usually round, based on the body height value (B) and the body width value (C) can be easily predicted by those skilled in the art, and thus a detailed description thereof will be omitted.

The predictive model selects as a candidate pig a pig whose body type information value, which is a calculated value obtained by dividing the pig's trunk length by the pig's height, is greater than a predetermined threshold, and forms abdominal fat in which the weight change trend of the candidate pig during the fattening period is learned in advance. It can be determined whether the candidate pig is a normal pig or an abdominal fat-forming pig as whether it corresponds to the pig's weight gain trend pattern. It can be predicted and presented at the time of shipment.

Since muscle and fat have different densities, there is a big difference in the volume of fat almost twice that of muscle at the same weight. Therefore, compared to normal pigs in which muscle and intramuscular fat are formed, a pig with abdominal fat has a relatively large body type information value.

Accordingly, the predictive model may compare the body shape information value interpreted from the photo information of the pig with a preset threshold value, and select a pig having a body shape information value greater than the threshold value as a candidate pig.

Here, the predictive model is a candidate pig for discrimination of abdominal fat-forming pigs through machine learning using photo information continuously taken of pigs corresponding to the fattening period and ultrasound data measuring the body fat distribution of the pigs at each time point. You can set a threshold to select .

Of course, for machine learning regarding the setting of the threshold, the training data may further include determination information indicating that the pigs are abdominal fat-forming pigs or normal pigs for each of the ultrasound data of pigs.

In Figure 4, when the pigs are fed with the same amount and frequency, the weight gain trend of normal pigs in which muscle is formed and intramuscular fat is adhered during the fattening period, and the gain of abdominal fat-forming pigs in which abdominal fat is mainly formed during the fattening period The trend is shown.

Muscle and fat have different volumes, and in the case of the same volume, fat is measured with a relatively small weight. Therefore, as shown in FIG. 4 , compared to normal pigs in which muscle and intramuscular fat are formed, abdominal fat-forming pigs, in which abdominal fat is mainly formed, show a relatively small gain in weight.

The predictive model determines that the candidate pig is an abdominal fat-forming pig, and when the change in weight measured so far for the candidate pig matches the weight gain trend pattern of the abdominal fat forming pig derived by machine learning in advance, the candidate pig is the abdominal fat forming pig, and the weight gain trend pattern Based on this, it is possible to predict the point in time when the corresponding abdominal fat-forming pig will gain up to a pre-specified minimum shipping weight and present it as the shipping time.

At this time, if the fattening period is set from 22 weeks to 26 weeks after birth, the predictive model is limited in advance to predict the point in time to gain up to the minimum shipment weight to a time earlier than 26 weeks, based on the weight gain trend pattern of pigs with abdominal fat. When it is predicted that the minimum shipment weight will not be reached even after 26 weeks have elapsed, the predictive model may be set to generate information on the cessation of breeding of the celiac fat-forming pig at the current time point.

In order for the predictive model to predict when the abdominal fat-forming pigs will gain up to the minimum shipping weight based on the weight gain trend pattern, more information on the amount of feed consumed by the abdominal fat-forming pigs as well as weight measurement information may be needed. .

To this end, as shown in FIG. 1 , in the pig room 210 , the camera unit 133 , etc., as well as the feed intake amount of each of the pigs raised in the pig room 210 are measured and the shipment time determination device 110 . ) A feeding amount measurement unit 137 for transmission may be further included.

Feeding amount measurement unit 137 is integrally configured with the feeder 220 disposed in the pig room 210, or is provided as the feeder 220 to the feeder 220 to measure the amount of feed consumed by the pig. It may be installed adjacently.

Referring to FIG. 4 , normal pigs are bred up to 26 weeks and shipped in a situation where they have reached the optimal shipping weight (eg, 115 kg), but the predictive model is that abdominal fat forming pigs have a minimum shipment weight (eg, 100 kg). About 24.5 weeks of reaching , can be derived as the time of shipment.

In general, when pigs corresponding to the fattening period do not form muscle and intramuscular fat but mainly abdominal fat, there is a problem that shipment income is not generated as much as the input feed cost because the quality of meat is lowered and the shipment grade is evaluated low. .

By shipping these celiac fat-forming pigs at an appropriate time before the end of the fattening period, it is possible to secure a relatively good meat quality and secure a relatively high shipment grade, as well as reduce feed costs, thereby improving the profits of livestock farmers. There are advantages that can be

The report generating unit 115 may generate a pig shipment report on the breeding state and shipment time of pigs corresponding to the fattening period.

In the case of pigs determined to be normal pigs by the predictive model, the pig shipment report includes the 26th week when the pre-specified fattening period expires or the shipment determined by the predictive model that machine-learned the normal pig weight gain trend pattern in advance to reach the optimal shipment weight. Information about the timing may be included.

On the other hand, in the case of pigs determined to be celiac fat-forming pigs by the predictive model, the pig shipment report includes the pigs at the time of shipment determined as the time when the pigs reach the pre-specified minimum shipping weight based on the weight gain trend pattern of celiac fat-forming pigs. information may be included.

At this time, the time at which the minimum shipment weight is reached is limited to be set prior to the 26th week when the specified fattening period expires. Proposals to stop pigs may be included in the pig release report.

In addition, the report generating unit 115 provides breeding environment information between a pig room in which a relatively large amount of abdominal fat-forming pigs are generated among a plurality of pig rooms in which pigs are bred in the fattening period, and a pig room in which a relatively small amount of abdominal fat-forming pig is generated. A pig release report can also be generated to further include comparative information on the differences. Here, the breeding environment information may include, for example, temperature, humidity, an operating state of a ventilation fan, breeding density, feed amount and number of times of feeding.

The storage unit 117 stores the pig photo information and weight measurement information corresponding to the entity recognition information recognized by the entity recognition unit 131 . In addition, in the storage unit 117, the received breeding environment information (eg, temperature, humidity, the number of breeding heads per pig farm, feed amount, etc.) sensed in each pig room 210 corresponds to each pig room 210 . saved to do

The controller 119 controls operations of the prediction unit 113 and the report generation unit 115 included in the shipment timing determining device 110 .

In addition, the shipment timing determining apparatus 110 may further include a web crawler 121 and an analysis unit 123 .

The web crawler 121 collects pig industry forecast information including a pig pig shipment forecast by using the Internet communication network.

The web crawler 121 automatically searches and indexes, for example, trend forecast information issued by the Ministry of Agriculture, Food and Rural Affairs, the Livestock and Livestock Association, etc., and the pig industry outlook information including newspaper articles, etc. using the Internet communication network in relation to the pig shipment forecast. can be configured.

The analysis unit 123 extracts forecast information on pig shipments by day from the pig industry forecast information collected by the web crawler 121, and selects days with relatively small pig shipments compared to the previous day or weekly average as recommended shipment dates, respectively. can be selected

The above-mentioned report generator 115 predicts the time when the pig reaches a predetermined minimum shipment weight based on the weight gain trend pattern of the abdominal fat forming pig in the pig shipment report with respect to the pig determined to be a celiac fat forming pig In addition to information on one shipment time, information on a recommended shipment date adjacent to the corresponding shipment time among the recommended shipment dates selected by the analysis unit 123 may be further included.

5 is a flowchart illustrating a method for determining a time of shipment of pigs according to an embodiment of the present invention.

Referring to FIG. 5 , in step 510 , the pig shipping timing determining device 110 machine-learns a predictive model using breeding history data designated as learning data.

Here, the breeding history data may be photographic information, ultrasound data, and weight measurement information accumulated in the existing breeding process for pigs corresponding to the fattening period in the corresponding livestock farm. In addition, the learning data may include breeding history data, breeding history data generated by other livestock farms, breeding history data that can be collected through the Internet, and the like.

As the predictive model is machine-learned using learning data, as will be described later, photo information and It can be used to determine whether a pig is a normal pig or a fat-forming pig by using the weight measurement information, and to generate optimal shipping time information according to whether it is a normal pig or a fat-forming pig.

In step 520, the pig shipment time determination device 110 receives the photo information and weight measurement information for each of the pigs raised in the pig room 210 corresponding to the fattening period. An object recognition unit 131 , a camera unit 133 , and a weight measurement unit 135 may be provided in the pig room 210 so that photo information and weight measurement information for each of the pigs can be received.

In step 530 , the pig shipping time determining device 110 determines whether or not the body shape information value interpreted from the photo information for each pig is greater than or equal to a predetermined threshold using the pre-machine-learned prediction model.

For the body shape information value, for example, after analyzing the height value of the pig and the height value of the pig's torso using a predetermined image analysis technique from the upper surface photograph information taken by the camera unit 133, and analyzing the pig's trunk width value from the side photograph information, , can be calculated as a calculated value obtained by dividing the circumference of the body estimated by the body height value and the body width value by the height value of the pig.

As a result of the determination in step 530, if there is no pig having a body shape information value equal to or greater than the threshold value, it is determined that all pigs are normal pigs, and the process proceeds to step 520 again.

However, as a result of the determination in step 520, if there is a pig having a body shape information value greater than or equal to the threshold value, in step 530, the pig shipment timing determining device 110 selects the pig as a candidate pig, and a predictive model that has been machine-learned in advance. is used to determine whether the weight change trend measured so far for the candidate pig matches the weight gain trend pattern of the abdominal fat-forming pig derived by machine learning in advance.

If the weight gain trend pattern of the abdominal fat-forming pig is not met, it is determined that the candidate pig is a normal pig and the process proceeds to step 520 again.

However, if it conforms to the weight gain trend pattern of the abdominal fat forming pig, in step 550, the pig shipment time determining device 110 determines that the candidate pig is the abdominal fat forming pig, and uses the machine-learned prediction model in advance. , based on the weight gain trend pattern, generates predictive information on when the corresponding abdominal fat forming pig will gain weight up to a pre-specified minimum shipment weight.

For example, if the normal fattening period when shipping pigs is set from 22 to 26 weeks of age, the release time according to the prediction information may be limited in advance to predict a time earlier than the time when the normal fattening period expires. .

At this time, if it is predicted that the minimum shipment weight will not be reached until the normal fattening period expires based on the weight gain trend pattern of the abdominal fat-forming pig, the pig shipment timing determining device 110 in step 540 determines the corresponding abdominal fat formation Pigs can also generate suggestion information to stop breeding at the current point in time.

In step 560, the pig shipment timing determining device 110 extracts forecast information on the daily pig shipment volume from the pig industry forecast information collected using the web crawler function, and the day when the pig shipment volume is relatively small compared to the previous day or the weekly average value. to select recommended shipping dates.

In step 570, the pig shipment time determining device 110 generates a pig shipment report including information on the optimal shipment time of pigs corresponding to the fattening period.

In the pig shipment report, in the case of pigs determined to be normal pigs, the optimal shipping time is the 26th week when the pre-specified fattening period expires, or the time when the predictive model that machine-learned the normal pig weight gain trend pattern in advance reaches the optimal shipping weight. Information on the time of shipment may be included.

On the other hand, in the pig shipment report, in the case of pigs determined to be celiac fat forming pigs, information on the time of shipment determined as the time when the pig reached the predetermined minimum shipping weight based on the weight gain trend pattern of celiac fat forming pigs is not available. may be included. In addition, among pigs determined to be celiac fat forming pigs, pigs that are predicted not to reach the minimum shipping weight until the pre-specified fattening period expires may include a proposal to stop breeding at the current time.

In addition, the pig shipment report describes the difference in breeding environment information between a pig house in which a relatively large amount of abdominal fat-forming pigs were generated and a pig house in which a relatively small amount of abdominal fat-forming pigs among a plurality of pigs in the finishing period were raised. Comparison information may be further included.

It goes without saying that the above-described method for determining the time of shipment of pigs may be implemented as a software program, an application, etc. embedded in a digital processing device, and may be performed as an automated procedure according to a time-series sequence. Codes and code segments constituting the program or the like can be easily inferred by a computer programmer in the art. In addition, the program is stored in a computer readable media, read and executed by the computer to implement the method.

Although the above has been described with reference to the embodiments of the present invention, those of ordinary skill in the art can variously modify the present invention within the scope without departing from the spirit and scope of the present invention described in the claims below. and may be changed.

110: shipment time determining device 111: receiving unit
113: forecasting unit 115: report generating unit
117: storage 119: controller
121: web crawler 123: analysis unit
131: object recognition unit 133, 133a, 133b: camera unit
135: weight measuring unit 137: feeding amount measuring unit
210: money room 215: partition
220: feeding table

Claims (1)

A computer program stored on a computer-readable medium for performing a method for determining when to release pigs, the computer program causing the computer to perform the following steps, the steps comprising:
(a) Presence of abdominal fat-forming pigs among pigs breeding in the pig room by applying the photo information received for each of the pigs raised in the pre-specified pig room corresponding to the pre-specified fattening period to the machine-learned prediction model in advance determining whether there is a candidate pig for determining whether or not;
(b) If the candidate pig exists, the change in weight information measured and collected for the candidate pig is applied to the predictive model in which the weight gain trend pattern of the abdominal fat-forming pig is machine-learned in advance, so that the candidate pig is determining whether the pig is celiac fat forming;
(c) when the candidate pig is a celiac fat forming pig, the predictive model is machine-learned in advance, the weight gain trend pattern of the celiac fat forming pig, and the change trend of the weight information measured and collected for the candidate pig so far , generating shipping time information by predicting when the candidate pigs, which are abdominal fat forming pigs, will gain up to a predetermined minimum shipping weight; and
(d) generating a pig shipping report on the shipping time for each of the pigs raised in the pig room corresponding to the fattening period,
The minimum shipment weight is a relatively small weight compared to the optimal shipment weight, which is a target weight for normal pigs to gain at the end of the fattening period, and the shipment date information of the abdominal fat-forming pigs is a date before expiration after the start of the fattening period. is limited in advance to be specified as
The normal pigs are the remaining pigs that are not judged to be abdominal fat forming pigs among the pigs raised in the pig room,
In step (a), the prediction model is
(a-1) After analyzing the height value and the trunk height value of the pig in the side photo information among the received photo information, and the trunk width value of the pig in the top photo information, the trunk height value and the trunk width value calculating a body shape information value that is a calculated value obtained by dividing the estimated pig's trunk circumference by the pig's height value; and
(a-2) performing the step of determining that the calculated body type information value is greater than or equal to a predetermined threshold as the candidate pig for determining the presence of abdominal fat forming pigs,
In step (b),
The predictive model determines that the candidate pig is an abdominal fat forming pig if the change in weight information measured and collected so far for the candidate pig corresponds to the weight gain trend pattern of the abdominal fat forming pig previously machine-learned,
In the pig shipment report generated in step (d) above,
i) In the case of normal pigs, the time when the end of the fattening period or the time when the optimal shipping weight is reached by the predictive model that machine-learned the normal pig weight gain trend pattern in advance is included as information on the shipping time,
ii) In the case of celiac fat-forming pigs, the time of gaining up to the minimum shipping weight is included as information on the shipping time, but based on the weight gain trend pattern of celiac fat-forming pigs, up to the minimum shipping weight until the expiration of the fattening period If it is determined that the increase is not possible, information that suggests stopping the breeding of the candidate pigs at the current time is included,
In step (d), forecast information on pig shipments by day is extracted from the pig industry forecast information collected using the web crawler function, and days with relatively small pig shipments compared to the previous day or weekly average are selected. Information is further included in the swine shipment report, wherein the recommended shipment dates included in the swine shipment report are limited to days falling within the range from the time when the candidate pig has gained to the minimum shipment weight to the expiration date of the finishing period A computer program stored on a computer-readable medium, characterized in that it becomes.

Images