KR20240045700A - Livestock shed target management system through the AI analysis - Google Patents

Abstract

The present invention relates to a livestock target management system through AI analysis, and more specifically, to a biometric information unit for checking the biometric information of each livestock object, an imaging unit for taking images of the livestock house, and the imaging unit and each biometric information unit. A control unit for receiving a signal and detecting a specific action through AI analysis, wherein the control unit includes an image detection unit for detecting a specific action of a preset object in the received image of the image unit, the image detection unit A biometric information verification unit that analyzes the signal of the biometric information unit for the object in the detected image, and a loading unit that extracts and uploads a clip video matching the specific activity when the biometric information verification unit confirms a specific action of the object. , an output unit that outputs the image uploaded by the loading unit, a target display unit that displays the corresponding object on the output unit, and an alarm unit that transmits an alarm to a preset terminal when the corresponding image is uploaded by the loading unit. , includes.
According to the present invention as described above, the biometric information of livestock is checked in real time, and when a preset specific action is detected, the corresponding image is stored and a notification is sent, and the subject is actually displayed through a terminal connected through short-distance communication or long-distance communication. By being able to confirm whether a specific action is being performed, reliability and management efficiency can be improved.

Description

Livestock shed target management system through the AI analysis}

This patent application is for the "Smart Farm Multi-Ministry Package Innovation Technology Development Project (Project No. 421000-00)" of the Ministry of Agriculture, Food and Rural Affairs, the Ministry of Science, ICT and Future Planning, and the Korea Institute of Agriculture, Food and Rural Development, and the Smart Farm Research and Development Foundation, funded by the Rural Development Administration. It was carried out with research funding support from the Smart Farm Multi-Ministry Package Innovation Technology Development Project.

The present invention relates to a livestock management system, and more specifically, to a terminal connected through short-distance communication or long-distance communication to check the biometric information of livestock in real time and, when a preset specific action is detected, to store the video and receive a notification. It is about a livestock target management system through AI analysis that can improve reliability and management efficiency by confirming whether the entity is actually performing a specific action.

In general, for all cattle farms, especially farms raising dairy cattle, the milk produced by cows that have given birth to calves or the calves produced are the main source of income, so infertile cows that are unable to conceive are sold, which has a negative impact on farm management. It goes crazy.

Therefore, in order to maintain milk productivity, artificial insemination is performed at the right time when the cow ovulates during the estrus period to enable the cow to become pregnant, and after giving birth to a calf, milk is produced to generate income for the farm.

Therefore, inducing normal estrus in cows and increasing the conception rate through appropriate artificial insemination is the most important and eco-friendly breeding management related to farm income.

Currently, the signs to recognize whether a cow is in estrus include observation of unstable movement, sniffing, increased number of steps, changes in vocalization, monk behavior, and monk acceptance behavior.

Among the signs of estrus, the best way to check them with the naked eye is estrus behavior, allowing estrus, and unstable behavior, but these are difficult to observe for 24 hours by a person, so livestock farms are complaining of difficulties in detecting estrus.

In particular, some farms install CCTV cameras to continuously observe the recorded images after recording, but this is cumbersome and time-consuming, so the development of a device that can observe signs of estrus is still awaited.

Currently, estrus observation aids and systems using activity level, number of steps, electrical conductivity of vaginal mucus, and body temperature changes have been developed overseas, but it is difficult for small farms to purchase expensive equipment with low accuracy.

Conventional technology includes Korean Patent Publication No. 2010-0012614, which is a estrus detection system using image recognition.

Accordingly, the present invention was created to solve the above problems, and includes a biometric information unit for checking the biometric information of each livestock object, an imaging unit for taking images of the livestock house, and signals from the imaging unit and each biometric information unit. It includes a control unit for receiving and detecting a specific action through AI analysis, wherein the control unit includes an image detection unit for detecting a specific action of a preset object in the received image of the image unit, A biometric information verification unit that analyzes the signal of the biometric information unit for the corresponding object in the video, a loading unit that extracts and uploads a clip video matching the specific activity when the biometric information verification unit confirms a specific action of the object, the above An output unit that outputs the image uploaded by the loading unit, a target display unit that displays the corresponding object on the output unit, and an alarm unit that sends an alarm to a preset terminal when the corresponding image is uploaded by the loading unit. Including real-time verification of livestock's biometric information, and when a preset specific action is detected, the video is saved and a notification is sent to indicate that the individual is actually performing a specific action through a terminal connected through short-distance communication or long-distance communication. The purpose is to provide a livestock target management system through AI analysis that can confirm and improve reliability and management efficiency.

The present invention to achieve the above object includes a biometric information unit for checking the biometric information of each livestock object, an imaging unit for taking images of the livestock house, and a specific device through AI analysis by receiving signals from the imaging unit and each biometric information unit. Includes a control unit for detecting behavior.

Preferably, the control unit includes an image detection unit that detects a specific action of a preset object in the received image of the image unit, and a biometric information confirmation unit that analyzes a signal of the biometric information unit for the corresponding object in the image detected by the image detection unit. A loading unit that extracts and uploads a clip video matching the specific activity when a specific action of the object is confirmed in the biometric information verification unit, an output unit that outputs the video uploaded by the loading unit, the output unit It includes a target display unit that displays the corresponding object, and an alarm unit that transmits an alarm to a preset terminal when the corresponding image is uploaded by the loading unit.

And the video detection unit detects a specific action by dividing the received video of the video unit into a certain time unit, and a slow detection unit that learns sparse key frames so that they are spaced at regular intervals, and the interval is smaller than the slow detection unit. It further includes a fast detection unit that performs narrow keyframe learning, and the slow detection unit and the fast detection unit are merged in parallel.

Additionally, the slow detection unit categorizes the color, texture, and brightness structure of the image, and the fast detection unit recognizes rapidly changing motion information.

And the above-mentioned specific action is one or more of lying down, standing up, eating food, and performing a monk's prayer.

In addition, it further includes a video storage unit for storing the video of the video unit and the video uploaded by the loading unit.

In addition, the biometric information unit is filled with a housing body having a storage space portion open on one side inside, a housing cover for closing the storage space portion, a fastening portion for fastening the housing cover to the housing body, and the storage space portion. It includes a solidification unit for fixing the housing cover by solidifying the fastening unit, and a sensor unit stored in the storage space unit to measure biometric information of the object.

And the other end of the housing body is formed in a hemispherical shape, and one end of the housing cover is formed in a hemispherical shape.

In addition, the housing body further includes a seating portion for preventing fluctuation of the sensor portion stored in the storage space portion.

And the seating portion includes a guide jaw protruding from the inner surface of the storage space portion.

In addition, the housing cover has a cover space part that is open to the other side inside so as to communicate with the storage space part, and a plurality of cover ribs formed along the inner surface of the cover space part to solidify the filled solidification portion and improve the fixing force. Includes.

And the fastening part includes a fastening body extending from the other end of the housing cover so as to be inserted into the storage space of the housing body, and a third protruding body along the outer peripheral surface of the other end of the fastening body to be pressed into the inner peripheral surface of the storage space. It includes a first fastening jaw, a second fastening jaw protruding along the outer peripheral surface of the middle portion of the fastening body so as to fit tightly into the inner peripheral surface of the storage space portion, and a solidification groove portion formed to allow the solidified portion filled in the storage space portion to flow in. .

In addition, the solidification groove portion includes a solidification main groove formed on the outer peripheral surface of the fastening body, and a solidification inlet groove formed on the outer peripheral surface of the other end of the fastening body to allow the solidification to flow into the solidification main groove.

In addition, the left and right widths of the solidification main groove are formed to be larger than the left and right widths of the solidification inlet groove to prevent the inflowed solidification from escaping into the solidification inlet groove when solidified.

In addition, it further includes a second fastening part for improving the coupling force between the housing cover and the solidified part.

And the second fastening part is a second fastening body that has a certain thickness and is formed to protrude to the other side of the cover space of the housing cover, and a second fastening body formed in the second fastening body to solidify in a state where the solidification part penetrates. Includes 2 fastening holes.

Additionally, the solidified portion is filled with epoxy in a molten state.

And the sensor unit is one or more of a temperature sensor, a position sensor, and a gyro sensor.

According to the livestock target management system through AI analysis according to the present invention, the biometric information of livestock is checked in real time, and when a preset specific action is detected, the video is stored and connected through short-distance communication or long-distance communication according to the notification. It is a very useful and effective invention that can improve reliability and management efficiency by confirming whether the relevant entity is actually performing a specific action through the terminal.

1 is a diagram showing a livestock target management system according to the present invention,
Figure 2 is a diagram showing a control unit according to the present invention,
Figure 3 is a diagram showing a biometric information unit according to the present invention;
Figure 4 is a diagram showing the solidification groove according to the present invention,
Figure 5 is a diagram showing a second fastening part according to the present invention.

Hereinafter, preferred embodiments according to the present invention will be described in detail with reference to the attached drawings. The detailed description set forth below in conjunction with the accompanying drawings is intended to illustrate exemplary embodiments of the invention and is not intended to represent the only embodiments in which the invention may be practiced. The following detailed description includes specific details to provide a thorough understanding of the invention. However, those skilled in the art will recognize that the present invention may be practiced without these specific details.

In some cases, in order to avoid ambiguity of the concept of the present invention, well-known structures and devices may be omitted or may be shown in block diagram form focusing on the core functions of each structure and device.

Throughout the specification, when a part is said to “comprise or include” a certain element, this means that it does not exclude other elements but may further include other elements, unless specifically stated to the contrary. do. Additionally, the term “… unit” used in the specification refers to a unit that processes at least one function or operation. In addition, the terms "a or an", "one", "the", and similar related terms are used in the context of describing the invention (particularly in the context of the claims below) as used herein. It may be used in both singular and plural terms, unless indicated otherwise or clearly contradicted by context.

In describing embodiments of the present invention, if a detailed description of a known function or configuration is judged to unnecessarily obscure the gist of the present invention, the detailed description will be omitted. The terms described below are terms defined in consideration of functions in the embodiments of the present invention, and may vary depending on the intention or custom of the user or operator. Therefore, the definition should be made based on the contents throughout this specification.

Hereinafter, preferred embodiments of the present invention will be described with reference to the attached drawings.

Figure 1 is a diagram showing a livestock target management system according to the present invention, Figure 2 is a diagram showing a control unit according to the present invention, Figure 3 is a diagram showing a biometric information unit according to the present invention, and Figure 4 is a diagram showing the present invention. It is a drawing showing a solidification groove according to the invention, and Figure 5 is a drawing showing a second fastening part according to the present invention.

As shown in the drawing, the livestock target management system through AI analysis includes a biometric information unit 10, an imaging unit 20, and a control unit 30.

The biometric information unit 10 is installed on each livestock object to check the biometric information of each object, and is fed, hung, or fixed.

Since this biometric information unit 10 includes each unique identification number, each object can be identified.

And the imaging unit 20 is provided to capture images of the livestock house.

The control unit 30 is provided to receive signals from the imaging unit 20 and each biometric information unit 10 and detect specific actions through AI analysis.

According to this livestock management system through AI analysis, the biometric information of each object can be checked in real time by the biometric information unit 10, and the situation of the livestock house is checked in real time by the imaging unit 20, and preset specific actions are detected. If so, the object's biometric information and image are stored and an alarm is sent to the preset terminal.

Accordingly, it is possible to quickly confirm whether a specific action has actually occurred, which not only leads to a quick situational judgment, but also improves management efficiency by solving the problems of conventional workers checking with the naked eye and checking the entire video overnight. there is.

For this purpose, the control unit 30 includes an image detection unit 32, a biometric information confirmation unit 33, a loading unit 34, an output unit 35, a target display unit 36, and an alarm unit 37.

The image detection unit 32 detects a specific action of a preset object in the received image of the image unit 20.

And the biometric information verification unit 33 analyzes the signal of the biometric information unit 10 for the corresponding object in the image detected by the image detection unit 32.

When the biometric information verification unit 33 confirms a specific behavior of the object, the loading unit 34 extracts and uploads a clip video matching the specific behavior.

Additionally, the output unit 35 is provided to output the image uploaded by the loading unit 34, and in one embodiment, a monitor is used.

The target display unit 36 is provided to display the corresponding object on the output unit 35, and displays the object performing a specific action and the object that is the target, respectively, and preferably uses different colors to distinguish them.

As the objects move, the worker can easily check their location upon arrival at the site, leading to a quick response.

In particular, a more rapid response can be induced by linking with the unique identification number of the biometric information unit 10.

And when the corresponding video is uploaded by the loading unit 34, the alarm unit 37 transmits an alarm to the preset terminal.

Of course, the preset terminal can access the information of the control unit 30 through the application.

Here, the image detection unit 32 receives, analyzes and reads signals from the sensor unit 600 of each biometric information unit 10, and the baseline is behavior analysis based on moving average, standard deviation, and k-means.

This baseline constructs learning data by dividing cow behavior into subclasses for a single object in order to transfer learn yolo5.

At this time, the subclass includes specific actions such as lying down, standing up, eating food, and performing monastic activities.

And the signs of specific actions are performed through DBSCAN analysis, taking into account the fact that they are far from the frequently occurring main cluster distributions.

In addition, it is desirable to minimize errors by analyzing specific behavior signs using GAN, which has been actively researched recently.

Here, the specific behavior is one or more of the following: instability of movement, sniffing, increased number of steps, change in vocalization, behavior of monks, behavior of allowing monks, lying down, standing up, and food intake. Even if females and males are separated separately, the breeding season is prolonged. When this happens, a specific action occurs, and this specific action causes stress to all individuals in the livestock farm, so a prompt response is required.

And, as shown in FIG. 2, the image detection unit 32 divides the received image of the image unit 20 into certain time units to detect a specific action, and optimizes the image in two ways in parallel.

This image detection unit 32 further includes a slow detection unit 32a and a fast detection unit 32b.

In other words, model learning that is optimized by simultaneously learning sparse keyframes spaced at regular intervals using the slow/fast theory and learning dense keyframes with narrower intervals than sparse keyframe learning is added. do.

The slow detection unit 32a, which learns sparse keyframes, categorizes the structure of the image, such as color, texture, and brightness, and configures the amount of information and calculation to be high.

And the fast detection unit 32b, which performs detailed keyframe learning, recognizes rapidly changing motion information.

Looking at this in more detail, by adding learning using keyframes and optical flow for a total of 10 seconds before and after each action, the work of the slow area, which is sparse keyframe learning, and the work of the fast area, which is dense keyframe learning, are performed in parallel. By combining them, specific actions can be read precisely.

And, as shown in FIG. 3, the biometric information unit 10 includes a housing body 100, a housing cover 200, a fastening unit 300, a coagulation unit 400, and a sensor unit 600.

The housing body 100 has a storage space portion open on one side formed therein.

And the housing cover 200 is provided to close the storage space portion of the housing body 100.

The fastening part 300 is provided to fasten the housing cover 200 to the housing body 100.

In addition, the solidification portion 400 is filled in the storage space of the housing body 100 and is provided to solidify the fastening portion 300 to fix the housing cover 200.

The sensor unit 600 is stored in the storage space of the housing body 100 and is provided to measure biometric information of the object.

Here, the sensor unit 600 has an identification ID and is one or more of a temperature sensor, a momentum sensor, a temperature and humidity sensor, and a gyro sensor.

Accordingly, it is possible to check the livestock's body temperature, momentum due to location changes, and behavior in real time.

And the other end of the housing body 100 is formed in a hemispherical shape, and one end of the housing cover 200 is formed in a hemispherical shape.

This biometric information unit 10 is preferably used by hanging it around the neck of livestock using a wire or belt.

Of course, the shape in which the housing cover 200 is fastened to the housing body 100 can be formed into a capsule shape for humans to take, so livestock can also take it.

And the housing body 100 further includes a seating portion 110 to prevent fluctuation of the storage body stored in the storage space.

This seating portion 110 includes a guide jaw 112 protruding from the inner surface of the storage space.

Accordingly, the storage body can be guided to the storage space without tilting and guided to an accurate position.

And the housing cover 200 has a cover space part that is open to the other side inside so as to communicate with the storage space part, and a plurality of cover ribs formed along the inner surface of the cover space part to improve the fixing force when the filled solidification part 400 is solidified. Includes (210).

Of course, a plurality of body ribs 110 may be formed along the storage space portion of the housing body 100.

The cover rib 210 not only maintains the shape, but also improves the bonding force with the solidified portion 400 to be filled.

Additionally, the fastening part 300 includes a fastening body 310, a first fastening jaw 320, a second fastening jaw 330, and a solidification groove 340.

The fastening body 310 extends from the other end of the housing cover 200 to be inserted into the storage space of the housing body 100.

And the first fastening jaw 320 is formed to protrude along the outer peripheral surface of the other end of the fastening body 310 so as to fit snugly into the inner peripheral surface of the storage space.

The second fastening jaw 330 is formed to protrude along the outer peripheral surface of the middle portion of the fastening body 310 so as to fit snugly into the inner peripheral surface of the storage space.

In addition, the solidification groove portion 340 is formed to allow the solidification portion 400 filled in the storage space to flow into the solidification groove portion 340.

As shown in FIG. 4, the solidification groove portion 340 includes a solidification main groove 342 and a solidification inlet groove 344.

The solidification main groove 342 is formed on the outer peripheral surface of the fastening body 310.

And the solidification inflow groove 344 is formed on the outer peripheral surface of the other end of the fastening body 310 to allow the solidification portion 400 to flow into the solidification main groove 342.

Here, the left and right widths of the solidification main groove 342 are formed to be larger than the left and right widths of the solidification inlet groove 344, so that when the inflowed solidification portion 400 is solidified, the solidification inlet groove 344 has a width smaller than that of the solidification main groove 342. ), the bonding force can be maintained by preventing separation.

Meanwhile, a plurality of through holes 346 are further formed on the bottom of the solidification main groove 342.

Each of these through-holes 346 communicates with the solidification main groove 342, the storage space, and the cover space, and when the filled solidification portion 400 is introduced, the solidification main groove 342 and the solidification inlet groove 344 The air is easily discharged, leading to complete filling.

Each of these through holes 346 is formed so that the cross-sectional area gradually decreases as it goes inward, thereby discharging air from the solidification main groove 342 into the storage space and the cover space, and preventing and minimizing backflow.

And as shown in FIG. 5, it further includes a second fastening portion 500 to improve the bonding force between the housing cover 200 and the solidified portion 400.

This second fastening portion 500 includes a second fastening body 510 and a second fastening hole 520.

The second fastening body 510 has a certain thickness and is formed to protrude to the other side of the cover space of the housing cover 200.

And the second fastening hole 520 is formed in the second fastening body 510 so that the solidified portion 400 is solidified in a penetrating state.

Accordingly, the coupling strength between the housing cover 200 and the solidified portion 400 can be improved and maintained, and thus the coupling strength with the housing body 100 can be improved and maintained.

Here, the solidified portion 400 is filled with epoxy in a molten state.

Of course, it is natural that any material similar to epoxy can be used.

When filling the solidified portion 400 in a molten state, it solidifies as it penetrates the second fastening hole 520 of the second fastening portion 500 and flows into the solidification groove 340 of the fastening portion 300. As it solidifies, the bonding strength between the housing body 100 and the housing cover 200 can be improved.

10: Biometric information unit 20: Imaging unit
30: control unit 32: image detection unit
34: Biometric information confirmation unit 36: Loading unit
38: Alarm unit

Claims (6)

A biometric information unit to check the biometric information of each livestock object;
An imaging unit for taking images of the barn; and
It includes a control unit for receiving signals from the imaging unit and each biometric information unit and detecting specific actions through AI analysis,
The control unit,
An image detection unit that detects a specific action of a preset object in the received image of the image unit;
a biometric information confirmation unit that analyzes signals from the biometric information unit for the corresponding object in the image detected by the image detection unit;
A loading unit that extracts and uploads a clip video matching the specific activity when the biometric information verification unit confirms a specific activity of the object;
an output unit that outputs the video uploaded by the loading unit;
a target display unit that displays the corresponding object on the output unit; and
A livestock target management system through AI analysis that includes an alarm unit that transmits an alarm to a preset terminal when the video is uploaded by the loading unit. The method of claim 1, wherein the image detection unit,
A slow detection unit that detects a specific action by dividing the received video of the video unit into certain time units and performs sparse keyframe learning so that they are spaced at regular intervals; and
It further includes a fast detection unit that learns tight keyframes with narrower intervals than the slow detection unit,
A livestock target management system through AI analysis, characterized in that the slow detection unit and the fast detection unit are combined in parallel. According to paragraph 2,
The slow detection unit categorizes the color, texture, and brightness structure of the image,
The fast detection unit is a livestock target management system through AI analysis, characterized in that it recognizes rapidly changing motion information. According to claim 1, the specific act is,
Livestock target management system through AI analysis characterized by one or more of the following: lying down, standing up, feeding, and monking. According to paragraph 1,
A livestock target management system through AI analysis further comprising a video storage unit for storing the video of the video unit and the video uploaded by the loading unit. The method of claim 1, wherein the biometric information unit,
A housing body having a storage space portion open on one side therein;
a housing cover to close the storage space;
A fastening part for fastening the housing cover to the housing body;
A solidification portion filled in the storage space and solidifying the fastening portion to secure the housing cover; and
A livestock target management system through AI analysis that includes a sensor unit stored in the storage space and measuring biometric information of the object.

Images