WO2020119184A1

WO2020119184A1 - Intelligent feeding system and method for livestock

Info

Publication number: WO2020119184A1
Application number: PCT/CN2019/103284
Authority: WO
Inventors: 罗扬; 李磊鑫
Original assignee: 京东数字科技控股有限公司
Priority date: 2018-12-12
Filing date: 2019-08-29
Publication date: 2020-06-18
Also published as: CN109618961B; CN109618961A

Abstract

An intelligent feeding system and method for livestock. The system comprises: a storage module (5), configured to store a feeding rule of livestock; a camera (1), mounted above a trough to collect a face image of the livestock closest to the trough after a photographing instruction is received; an identification module (3), configured to determine an identifier of the livestock according to the face image; an electronic feeder (4), mounted on the trough to add feed into the trough according to a received discharging instruction; and a controller (2), configured to send the photographing instruction to the camera (1) according to the feeding rule of the storage module (5), receive the face image and a feed image from the camera (1) and forward same to the identification module (3), receive the identifier of the livestock and the feed surplus amount of the trough from the identification module (3), and send the discharging instruction to the electronic feeder (4) according to the identifier of the livestock in combination with the feeding rule. According to the system, all the livestock are distinguished by using a face identification technology of artificial intelligence, the livestock individuals are remotely and precisely fed and controlled, and the feeding rule is conveniently made and adjusted.

Description

Intelligent feeding system and method for domestic animals

This application requires the priority of the Chinese patent application submitted to the Chinese Patent Office on December 12, 2018, with the application number 201811515215.1 and the invention titled "An Intelligent Feeding System and Method for Livestock", the entire content of which is incorporated by reference in In this application.

Technical field

This application relates to the field of livestock breeding, in particular to an intelligent feeding system and method for livestock.

Background of the invention

As a big country of traditional agriculture and animal husbandry, the pig breeding industry has always occupied a dominant position in my country.

At present, my country still follows the more traditional way of feeding pigs, mostly based on the experience summarized in production. Especially for sows, different feeding methods, including feed type, feeding amount and feeding time, are formulated in different growth cycles of pigs, in order to adapt to the growth rules of pigs to the greatest extent and meet the nutritional needs of pigs.

Summary of the invention

In view of this, the present application provides an intelligent feeding system and method for livestock, so as to realize an independent feeding process for each livestock, and realize fine management of livestock breeding.

The technical solution of this application is implemented as follows:

An intelligent feeding system for domestic animals includes a camera, a controller, an identification module, an electronic feeder and a storage module, wherein,

The storage module is used to store the feeding rules of domestic animals;

The camera is installed above the food trough to collect the facial images of the livestock closest to the food trough when receiving the photographing instruction;

The identification module is used to determine the identification of the domestic animal based on the facial image;

The electronic feeder is installed in the trough to feed the trough according to the received cutting instruction;

A controller, the controller is electrically connected to the storage module, the camera, the identification module and the electronic feeder to send the photograph to the camera according to the feeding rules stored by the storage module Instruction to receive the facial image from the camera and forward it to the recognition module, receive the identification of the livestock from the identification module, and feed the electronic feed according to the identification of the livestock and the feeding rules The feeder sends the cutting instruction.

Further, the feeding rule includes for each livestock in the enclosure where the trough is located:

Identification, feeding time period, single feed input, total feed input.

Further, the camera is also used to collect feed images in the trough;

The identification module is also used to determine the remaining amount of feed in the trough according to the feed image;

The controller is also used to judge whether the total feed amount of the current livestock is reached according to the amount of feed already fed and the remaining amount of feed in the trough during the current feeding process of the livestock, and if so, stop Send the cutting instruction, otherwise continue to send the cutting instruction according to the feeding rules.

Further, the intelligent feeding system for livestock also includes:

Statistics module, the statistics module is electrically connected to the identification module and the storage module to obtain the identification of the livestock, and determine the identification location according to the backfat data and weight data of the livestock corresponding to the identification The feeding rules of the corresponding domestic animals under the current backfat and body weight conditions, and update the determined feeding rules to the storage module.

Further, the storage module is a software-as-a-service SaaS module.

Further, the recognition module recognizes the facial image using artificial intelligence AI method to determine the identity of the domestic animal.

An intelligent feeding method for domestic animals, including:

Collect facial images of livestock closest to the trough;

Determine the identity of the livestock based on the facial image;

The feed is put into the trough according to the identification of the livestock and the feeding rules.

Identification, feeding time period, single feed input, total feed input.

Further, the intelligent feeding method of domestic animals further includes:

Collecting feed images in the trough;

Determine the remaining amount of feed in the trough according to the feed image;

In the current feeding process of livestock, it is judged whether the total feed amount of the current livestock is reached according to the amount of feed already fed and the remaining amount of feed in the trough, if so, stop feeding into the trough Feed, otherwise continue to feed into the trough according to the feeding rules.

Further, the intelligent feeding method of domestic animals further includes:

Obtain the identification of the livestock, and according to the backfat data and weight data of the livestock corresponding to the identification, determine the feeding rules of the livestock corresponding to the identification under the current backfat and body weight conditions To update.

Further, an artificial intelligence AI method is used to identify the facial image to determine the identity of the domestic animal.

A storage medium, the storage medium stores instructions, and when the instructions are executed by a processor, the above-mentioned intelligent feeding method of livestock is realized.

It can be seen from the above solution that the intelligent feeding system and method of domestic animals of the present application uses artificial intelligence facial recognition technology to distinguish each animal, and then uses the feeding rules for each animal to realize the targeting of each animal Sexual feed is fed to achieve precise feeding for individual livestock. At the same time, in the intelligent feeding system of livestock, in addition to the camera and the electronic feeder, other components can be built remotely and connected to the camera and the electronic feeder by the network, so as to realize the remote control and feeding of livestock Centralized management, and conducive to the formulation and adjustment of feeding rules. The collected feeding data, accumulated over time, can form valuable big data, which provides a reference for studying the growth of livestock.

Brief description of the drawings

1 is a schematic structural diagram of an intelligent feeding system for livestock according to an embodiment of the present application;

2 is a flowchart of a method for intelligently feeding livestock according to an embodiment of the present application;

FIG. 3 is a schematic flowchart of a pig feeding process in the embodiment of the present method;

FIG. 4 is a schematic diagram of the process of monitoring the remaining amount of feed in a trough and updating feeding rules in an embodiment of the present method.

Implementation

In order to make the purpose, technical solutions and advantages of the present application clearer, the following describes the present application in further detail with reference to the accompanying drawings and embodiments.

In an embodiment of the present application, feeding the sow includes the following steps:

(1) Identify sows by ear tagging them. According to the production cycle, the sows are divided into different batches for management. The sows of the same batch are fed in accordance with the same feeding rules, so as to save time and feed costs to the greatest extent, and maximize the production capacity of sows. .

(2) In large-scale farms, the feed is fed through the feed line, and each field can be artificially set the feeding time and feeding amount to reach the preset time and feed uniformly. If you want to adjust the feeding amount, the breeder needs to adjust the feeding amount of the barrel manually.

(3) According to the division of the type of the column, it can be divided into four categories: large column, nursery column, positioning column and delivery bed. There will be multiple pigs in the big pen and nursery pen, and there is only one trough in each pen, so there will be multiple pigs eating in one trough.

The above-mentioned pig feeding process has the following disadvantages:

(1) The pig farm marks the pigs by ear tagging them, which is not the optimal solution. This method will cause some damage to the pig's ears. At the same time, fighting and biting will often occur between pigs and the ear tags may fall off, causing the pig to lose its own identification information and unable to find the pig's growth records. . Sometimes in order to avoid this situation, pigs have to be labeled with multiple ear tags, which will cause more harm to the pigs. In addition, when the breeder inquires about the ear tag, he needs to grasp the pig ear to read the number on the ear tag, which is extremely inconvenient to operate.

(2) The health status of animals is easily affected by external environmental factors. Pigs are particularly sensitive to the external environment (such as bacteria and viruses), and the bacteria carried on humans are easily transmitted to pigs, causing pigs to become sick and even an epidemic. Regular farms will undergo strict disinfection when entering the production workshop, and limit the number of people entering the pig farm, which is intended to minimize this risk. The above-mentioned feeding methods inevitably require a lot of manual setting of the feeding amount of each pen, which increases the risk of infection virtually.

(3) Batch management of sow production can save the cost of breeding and increase productivity, but it is far from achieving refined breeding. Each pig has its own habits, and feeding should vary from pig to pig. Even in the same batch, the feeding amount and feeding time of each pig need to be personalized. In addition, in the case of multiple pigs in a pen, such as large pens and nursery pens, there will be predation. Some pigs will overfeed, resulting in a reduction in feed conversion rate, and some pigs will not be able to grab food, resulting in malnutrition. affect production.

(4) The current pig feeding method can only be based on "feel" and "experience", lack of accurate data support, such as how much feed each pig should eat, how much feed actually eaten, the above breeding methods are not capable of Collect such data in daily production.

As shown in FIG. 1, the intelligent feeding system for livestock according to the embodiment of the present application includes a camera 1, a controller 2, an identification module 3, an electronic feeder 4 and a storage module 5. The storage module 5 is used to store the feeding rules of domestic animals. The camera 1 is installed above the food trough to collect the facial image of the livestock closest to the food trough when receiving the photographing instruction. The recognition module 3 is used to determine the identity of the domestic animal based on the facial image. The electronic feeder 4 is installed in the food trough to feed the food trough according to the received feeding instruction. The controller 2 is electrically connected to the storage module 5, the camera 1, the identification module 3, and the electronic feeder 4 to send a photographing instruction to the camera 1 according to the feeding rules stored in the storage module 5 and receive a face from the camera 1 The image is forwarded to the identification module 3, the identification of the livestock is received from the identification module 3, and the feed instruction is sent to the electronic feeder 4 according to the identification of the livestock and the feeding rules.

In a specific embodiment, the storage module 5 adopts a SaaS (Software as Service) software. Among them, SaaS is an innovative software application model in Internet applications, which realizes the storage of cloud data. In the embodiment of the present application, the use of the SaaS module can reduce the hardware cost of the intelligent feeding system for livestock, on the other hand, it also provides a basis for cloud computing and statistics of intelligent feeding of livestock.

SaaS is a hardware and rule management system. In the embodiment of the present application, SaaS can bind the relationship between the camera 1, the controller 2, and the livestock breeding farm. On the other hand, it is responsible for setting and storing the feeding rules for each animal, including the identification of the animal (equivalent to the ear tag in the prior art), feeding time period, single feed input, total feed input, etc. content. Each time feeding, the feeding rules are read from SaaS, and then the corresponding action is triggered by the controller 2 to make the whole process run normally.

In a specific embodiment, the camera 1 is also used to collect feed images in the trough. The identification module 3 is also used to determine the remaining amount of feed in the trough based on the feed image. The controller 2 is also used for judging whether the total feed amount of the current livestock has been reached according to the amount of feed already fed and the remaining amount of feed in the trough during the current feeding process of the livestock, if it is, it stops sending the feed instruction, Otherwise, continue to send cutting instructions according to the feeding rules.

The camera 1 is an image acquisition device in an embodiment of the present application, and is responsible for taking pictures and recording videos of the faces and food troughs of domestic animals. For the camera 1, in one embodiment, the resolution is at least 720P. The installation position of the camera 1 needs to be able to clearly capture the face and trough of livestock, provide clear and stable data for the statistics module 6, and avoid being touched by livestock, so the best installation location of the camera 1 is the railing directly in front of the fence Between and directly above the trough.

In the embodiment of the present application, the intelligent feeding system for domestic animals further includes a statistical module 6, the statistical module 6 is electrically connected to the identification module 3 and the storage module 5 to obtain the identification of the domestic animal, and according to the backfat data of the domestic animal corresponding to the identification and Weight data, determine the feeding rules of the livestock corresponding to the identification under the current backfat and body weight conditions, and update the determined feeding rules to the storage module 5.

The controller 2 is the core part in the embodiment of the present application, and is responsible for connecting the other components in the intelligent feeding system of the entire livestock in series, sending instructions to the various components, and triggering the related actions of each component. In a specific embodiment, the controller 2 is mainly composed of a development board, supports OpenGL ES1.1/2.0/3.0/3.1, OpenVG1.1, OpenCL, DX11, AFBC (frame buffer compression), and supports PCIe x2, SATA, USB 2.0, USB 3.0, HSIC, SSIC, Audio, UIM, I2C multiple interfaces, can connect Wi-Fi, Bluetooth, Gigabit Ethernet. In one embodiment, the controller 2 is remotely connected to the camera 1 and the electronic feeder 4 through Wi-Fi, Gigabit Ethernet, for example, to realize remote control of intelligent feeding of livestock.

In the embodiment of the present application, the electronic feeder 4 can add an intelligent control module on the basis of the traditional feeder, and be remotely controlled by the controller 2 through the network, and trigger the cutting action according to the instruction of the controller 2. In the embodiment of the present application, the electronic feeder 4 controls the accuracy of the amount of feed fed through the installed sensor, which can be accurate to the gram, ensuring that the amount of livestock feed is strictly implemented according to the established plan.

In the embodiment of the present application, the recognition module 3 uses the AI (Artificial Intelligence) method to recognize the facial image to determine the identity of the domestic animal.

Among them, the AI algorithm involved in the embodiments of the present application includes a livestock face recognition algorithm and a trough feed remaining amount recognition algorithm.

Among them, domestic animal face recognition algorithms mainly include MTCNN algorithm, YOLO algorithm and MobileNet algorithm.

MTCNN proposes a Multi-task (multi-task) face detection framework, which simultaneously performs face detection and face feature point detection. In the embodiment of the present application, the algorithm is transplanted into the process of livestock identification, which can also be applied to the application scenarios of livestock feeding.

The YOLO algorithm is an object detection algorithm, which processes the problem into a regression problem, and uses a convolutional neural network structure to directly predict the bounding box and category probability from the input image.

Google's MobileNet further researched the use of depthwise separable convolutions (depth separable convolution structure) and designed the MobileNet algorithm. The essence of depthwiseseparable convolutions is a sparse expression with less redundant information. On this basis, two options for efficient model design are given: width factor (width multiplier) and resolution factor (resolution multiplier); by weighing the size, delay time and accuracy, to build a smaller and faster MobileNet algorithm . The Google team has demonstrated the effectiveness of the MobileNet algorithm as an efficient basic network through a variety of experiments.

OpenCV algorithm for identifying the remaining amount of trough feed. OpenCV is a cross-platform computer vision library released under the BSD (Berkeley Software Distribution) license (open source) license (open source), which can run on Linux, Windows, Android and Mac OS operating systems. OpenCV is lightweight and efficient. OpenCV consists of a series of C functions and a small number of C++ classes. It also provides interfaces for languages such as Python, Ruby, and MATLAB, and implements many general algorithms in image processing and computer vision.

FIG. 2 shows a flow of a method for intelligently feeding livestock according to an embodiment of the present application. As shown in FIG. 2, the method for intelligently feeding livestock according to an embodiment of the present application includes:

Step 1. Collect the facial images of the livestock closest to the trough;

Step 2. Determine the identity of the domestic animal based on the facial image;

Step 3. Put feed into the trough according to the identification of livestock and the feeding rules.

Among them, the feeding rules include the identification of each animal in the field where the trough is located, the feeding time period, the amount of single feed, and the total feed.

In the embodiment of the present application, the intelligent feeding method for domestic animals further includes:

Step 4. Collect feed images in the trough;

Step 5. Determine the remaining amount of feed in the trough based on the feed image;

Step 6. In the current feeding process of livestock, it is judged whether the total feed amount of the current livestock is reached according to the amount of feed already fed and the remaining amount of feed in the trough, if so, stop feeding the feed into the trough, otherwise Continue to feed into the trough according to the feeding rules.

The intelligent feeding method for livestock according to the embodiment of the present application further includes a method for updating the feeding rules. The method specifically further includes:

Obtain the identification of the livestock, and determine the feeding rules of the livestock corresponding to the identification under the current conditions of backfat and body weight according to the backfat data and weight data of the corresponding livestock, so as to update the feeding rules.

The intelligent feeding method of domestic animals in the embodiment of the present application adopts artificial intelligence AI method to recognize facial images to determine the identity of domestic animals.

The following is a supplementary description of the intelligent feeding system and method of the livestock of the present application in combination with the feeding process of pigs.

Fig. 3 is a flowchart of the feeding process of pigs in the embodiment of the present method. The process includes the following steps.

A. The controller monitors the feeding rules in the storage module and enters B;

B. The storage module judges whether it is time to feed, if it is, then enters C, otherwise returns to A;

C. The storage module feeds back a start signal to the controller, and then enters D;

D. The camera starts continuous image collection, such as shooting 2 frames per second, or continuous video recording, and then enters E;

E. The camera sends the image to the recognition module through the controller, and then enters F;

F. The recognition module recognizes the pig's face to determine which pig (sign), and then enters G;

G. The storage module queries the feeding rules of the pig according to the identification result (identification) of the identification module (for example: the feeding time period is from 11:00 to 12:00; the single feed feeding amount is 10 grams; the total feed feeding amount is 500 grams ), then enter H;

H. The storage module feeds back the feeding rules to the controller, and then enters I;

I. Determine whether the feed amount of the pig reaches the total feed amount, if it is, then end the feeding process of the pig, otherwise enter J;

J. The controller issues instructions to the electronic feeder according to the feeding rules, and the electronic feeder feeds the feed and proceeds to step D.

Fig. 4 is a schematic diagram of a process of monitoring the remaining amount of feed in a trough and updating feeding rules in an embodiment of the present method. As shown in FIG. 4, the process of monitoring the remaining amount of feed in a trough and updating feeding rules in the embodiment of the present application includes the following steps.

A’, the controller judges whether the electronic feeder starts to feed, when the electronic feeder starts to feed, it enters B’;

B’, the controller sends instructions to the camera to continue image acquisition, and then enters C’;

C’, the camera continuously collects the image data of the feed in the trough, and then enters D’;

D’. The camera sends the image data of the feed in the trough to the recognition module through the controller, and then enters E’;

E’. Determine whether the feed in the trough is finished, if it is finished, go to F’;

F’. Count the actual feeding amount of each pig, which can be obtained by accumulating historical data of multiple feedings, and then enter G’;

G’, combined with parameters such as back fat, body weight, and age, summarize the feeding rules of each pig in the future for a period of time according to the pig feeding business model, and then enter H’;

H’. Update the feeding rules to the storage module, and then end.

An embodiment of the present application further provides a storage medium that stores instructions, and when the instructions are executed by a processor (such as a CPU), the above-mentioned intelligent feeding method for livestock is implemented. For the specific implementation process of the method, please refer to the above embodiment, and no more details are provided here.

The intelligent feeding system, method and storage medium for domestic animals provided by the embodiments of the present application use artificial intelligence facial recognition technology to distinguish each animal, and then use the feeding rules for each animal to achieve the aim of each animal Feed feeding to achieve precise feeding for individual livestock. At the same time, in the intelligent feeding system of livestock, in addition to the camera and the electronic feeder, other components can be built remotely and connected to the camera and the electronic feeder by the network, so as to realize the remote control and feeding of livestock Centralized management, and conducive to the formulation and adjustment of feeding rules. The collected feeding data, accumulated over time, can form valuable big data, which provides a reference for studying the growth of livestock.

The intelligent feeding system and method of domestic animals according to the embodiments of the present application are not only suitable for feeding pigs, but also suitable for feeding large animals such as cattle and sheep.

The above are only the preferred embodiments of this application and are not intended to limit this application. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of this application should be included in this application Within the scope of protection.

Claims

An intelligent feeding system for domestic animals, characterized by:

Including camera, controller, identification module, electronic feeder and storage module, where,

The storage module is used to store the feeding rules of domestic animals;

The camera is installed above the food trough to collect the facial images of the livestock closest to the food trough when receiving the photographing instruction;

The identification module is used to determine the identification of the domestic animal based on the facial image;

The electronic feeder is installed in the trough to feed the trough according to the received cutting instruction;

A controller, the controller is electrically connected to the storage module, the camera, the identification module and the electronic feeder to send the photograph to the camera according to the feeding rules stored by the storage module Instruction to receive the facial image from the camera and forward it to the recognition module, receive the identification of the livestock from the identification module, and feed the electronic feed according to the identification of the livestock and the feeding rules The feeder sends the cutting instruction.
The intelligent feeding system for livestock according to claim 1, wherein the feeding rule includes: for each livestock in the pen where the trough is located:

Identification, feeding time period, single feed input, total feed input.
The intelligent feeding system for livestock according to claim 1, characterized in that:

The camera is also used to collect feed images in the trough;

The identification module is also used to determine the remaining amount of feed in the trough according to the feed image;

The controller is also used to judge whether the total feed amount of the current livestock is reached according to the amount of feed already fed and the remaining amount of feed in the trough during the current feeding process of the livestock, and if so, stop Send the cutting instruction, otherwise continue to send the cutting instruction according to the feeding rules.
The intelligent feeding system for livestock according to claim 1, wherein the intelligent feeding system for livestock further comprises:

Statistics module, the statistics module is electrically connected to the identification module and the storage module to obtain the identification of the livestock, and determine the identification location according to the backfat data and weight data of the livestock corresponding to the identification The feeding rules of the corresponding domestic animals under the current backfat and body weight conditions, and update the determined feeding rules to the storage module.
The intelligent feeding system for livestock according to claim 1, characterized in that:

The storage module is a software-as-a-service SaaS module.
The intelligent feeding system for livestock according to any one of claims 1 to 5, characterized in that:

The recognition module recognizes the facial image using artificial intelligence AI method to determine the identity of the domestic animal.
An intelligent feeding method for domestic animals, which is characterized by including:

Collect facial images of livestock closest to the trough;

Determine the identity of the livestock based on the facial image;

The feed is put into the trough according to the identification of the livestock and the feeding rules.
The intelligent feeding method for livestock according to claim 7, characterized in that the feeding rule includes: for each livestock in the field where the trough is located:

Identification, feeding time period, single feed input, total feed input.
The intelligent feeding method for livestock according to claim 7, wherein the intelligent feeding method for livestock further comprises:

Collecting feed images in the trough;

Determine the remaining amount of feed in the trough according to the feed image;

In the current feeding process of livestock, it is judged whether the total feed amount of the current livestock is reached according to the amount of feed already fed and the remaining amount of feed in the trough, if so, stop feeding into the trough Feed, otherwise continue to feed into the trough according to the feeding rules.
The intelligent feeding method for livestock according to claim 7, wherein the intelligent feeding method for livestock further comprises:

Obtain the identification of the livestock, and according to the backfat data and weight data of the livestock corresponding to the identification, determine the feeding rules of the livestock corresponding to the identification under the current backfat and body weight conditions To update.
The intelligent feeding method for livestock according to claim 7, characterized in that:

An artificial intelligence AI method is used to identify the facial image to determine the identity of the domestic animal.
A storage medium characterized in that the storage medium stores instructions, and when the instructions are executed by a processor, an intelligent feeding method for livestock according to any one of claims 7-11 is realized.