Pig farm feed putting system based on image recognition technology

Abstract

The invention relates to the technical field of pig farm feed delivery, in particular to a pig farm feed delivery system based on an image recognition technology, which is provided with a pigsty information module, a pigsty inspection module, a data processing module, an intelligent feeding module, an alarm module and a data display module, and can automatically track influence factors influencing pig appetite. The feed putting amount of live pigs in the pigsty is calculated in real time, the feed formula is automatically adjusted, the feed putting amount and the feed formula are controlled, excessive ingestion of the live pigs is prevented, inappetence of the live pigs is prevented, problems are found in time, healthy growth of the live pigs is guaranteed by means of food non-eating-up detection, temperature and humidity detection and the like, and meanwhile the feed putting amount and the feed formula are automatically adjusted. By comparing the back fat value and the fat degree of the live pigs in the pigsty with the back fat value and the fat degree of the live pigs of the corresponding ear numbers of the pigsty, the situation that the information of the live pigs corresponding to the pigsty is forgotten to be replaced due to pigsty replacement of the live pigs can be found in time, and the problem that the feeding amount of the live pigs in the pigsty is unreasonable is avoided.

Description

Technical Field

[0001] The invention relates to the technical field of pig farm feed placement, and in particular to a pig farm feed placement system based on image recognition technology. Background Art

[0002] In modern pig farms, in order to effectively manage the health status, feed requirements and other information of each pig, the pigs are usually tagged or marked with ears to record their identity information. Correspondingly, a file card containing a QR code and a digital ear number is hung in front of the pig pen. Traditional pig farm management methods rely on manual record keeping and file management, which is time-consuming and prone to errors. At the same time, the amount of feed given to pigs depends more on the experience of the breeder, which is highly subjective. Summary of the invention

[0003] In view of this, the purpose of the present invention is to provide a pig farm feed delivery system based on image recognition technology, which can automatically calculate the feed delivery amount and feed formula, promptly discover problems in the breeding process, and automatically check the situation where the pigs forget to change the corresponding pig information when changing the pig pens, thereby improving the level of automated management of the pig farm.

[0004] To achieve the above object, the present invention provides the following technical solution: a pig farm feed delivery system based on image recognition technology, comprising:

[0005] A pig pen information module, which includes at least a pig pen file card and a pig file card, wherein the pig file card corresponds to the ear number of the pigs fed in the pig pen, and the pig pen file card and the pig file card are displayed in the form of a QR code;

[0006] The pig pen inspection module is equipped with a visible light camera to collect images of the pig pen file card and the pig file card; a depth camera to collect depth images of the pig backs of the pigs in the pig pen; and a temperature and humidity sensor to collect the pig pen environmental temperature t0 and humidity h0;

[0007] The data processing module receives the pig pen file card and the pig file card images collected by the pig pen inspection module, obtains the pig pen information and the pig information fed in the pig pen by scanning the pig pen file card and the pig file card QR code, wherein the pig pen information at least includes: the pig pen number idh, the pig information at least includes: the pig ear number idp, the pig growth time gt and the pig target fatness ht0; receives the pig back depth image, calculates the current pig back fat value df and fatness through the pig back fat calculation model and the pig fatness calculation model ht; receiving the data sent back by the temperature and humidity sensor, recording the pig pen ambient temperature t0 and humidity h0, and further calculating the current pig feed input fq in the pig pen, where fq = f(gt, df, ht, ht0, t0, h0), indicating that the pig feed input fq is a function of the pig growth time gt, pig back fat value df, pig fatness ht, pig target fatness ht0, pig pen temperature t0, and humidity h0; the relevant data processing results are sent to the intelligent feeding module, and the sent data at least includes: pig pen number idh and feed input fq;

[0008] The intelligent feeding module receives the data sent by the data processing module and interacts with the intelligent feeder, matches the corresponding pig pen according to the pig pen number idh, and realizes accurate feed delivery according to the feed delivery amount fq.

[0009] In particular, the data processing module also queries the database according to the pig ear number idp to obtain the historical feed formula of the pig. When the feeding time of the previous formula feed exceeds the set threshold, the data processing module recalculates the feeding formula ff, where ff=g(t0,h0), indicating that the feed formula ff is a function of the pig pen temperature t0 and humidity h0.

[0010] In particular, the data processing module also receives the visible light image data in the pigpen collected by the pigpen inspection module, inspects the garbage in the pigpen through the target detection algorithm, and evaluates the hygiene conditions of the pigpen.

[0011] Furthermore, the pig farm feed delivery system also includes an alarm module. When the pig pen temperature t0 and humidity h0 exceed the set temperature and humidity thresholds, the alarm module issues an environmental warning; when the target detection detects the presence of garbage in the pig pen, the alarm module issues a hygiene warning; when the difference Δdf and Δht between the back fat value df and the fatness ht of the pig obtained based on the depth image and the most recent back fat value df_last and fatness ht_last of the pig queried in the database through the pig ear number idp exceeds the set threshold, the system alarm module issues a pig mismatch warning, where Δdf = |df-df_last|, Δht = |ht-ht_last|.

[0012] In particular, the pig pen inspection module is also equipped with an infrared camera to collect infrared data of pigs. The data processing module receives the infrared data of pigs and calculates the body temperature tp of the live pigs in the pig pen. When tp exceeds the set threshold, the system alarm module issues an abnormal body temperature warning for the live pigs.

[0013] Furthermore, when the intelligent feeding module completes feeding at the set time, the data processing module also processes the visible light image data in the pig pen, captures the image of the pig trough in the pig pen, and determines whether there is any feed remaining in the trough through a binary image classification model. When there is feed remaining, the system alarm module issues a feed residue warning.

[0014] Preferably, the breeder can scan the QR code in the pig pen information module to obtain static and dynamic information of the pig pen and the pigs in the pen, wherein the static information includes at least: the pig pen number idh, the shape and size of the pig pen, the pig pen type, the pig ear number idp, the target pig fatness ht0, and the pig breed; the dynamic information includes at least: the ambient temperature t0 and humidity h0 in the pig pen, the growth time gt of the pigs in the pig pen, the growth stage of the pigs, the current back fat value df of the pigs, the current fatness ht of the pigs, the change curve of the back fat value df and fatness ht of the pigs with the growth time, the current pig feed formula, and the current pig feed input.

[0015] Preferably, the pig farm feed delivery system also includes a data display module. In addition to displaying the static and dynamic information of each pig pen and the pigs in the pig pen, the data display module also counts at least the number of all pig pens in the current pig farm, the number of occupied pig pens, the total number of pigs, the number of pigs in each stage, the daily feed ration, pig farm alarm information and the inspection video data of the pig pen inspection module.

[0016] Preferably, the data processing module executes the data processing function at a set frequency and a set time.

[0017] Compared with the prior art, the present invention has the following beneficial effects:

[0018] 1. The present invention can automatically track the factors that affect pig appetite, calculate the feed input amount of pigs in the pig pen in real time and automatically adjust the feed formula, so as to control the feed input amount and feed formula to prevent the pigs from overeating, and prevent the pigs from losing their appetite through food unfinished detection, temperature and humidity detection and other means, discover problems in time and ensure the healthy growth of pigs.

[0019] 2. The present invention compares the back fat value and fatness of the pigs in the pig pen with the back fat value and fatness of the pigs with the corresponding ear number in the pig pen, so as to timely discover the situation where the corresponding pig information of the pig pen is forgotten to be changed due to the pig pen change, thereby avoiding the problem of unreasonable feeding amount of the pigs in the pen. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] Figure 1 It is a schematic diagram of the system module of the present invention;

[0021] Figure 2 It is a schematic diagram of the linkage between the pig pen inspection module and the data processing module of the present invention. DETAILED DESCRIPTION

[0022] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0023] like Figure 1 , 2 As shown, the present invention provides a pig farm feed delivery system based on image recognition technology, and the system is mainly divided into 6 major parts:

[0024] 1. Pig pen information module. This module mainly provides a QR code, which includes the QR code of the pig pen and the QR code of the live pig in the pig pen (mainly for positioning pen, that is, only one pig is fed in a pig pen). Among them, the QR code of the pig pen is generally fixed, which records the files about the pig pen, and the QR code of the live pig is generally replaceable. When the live pig is transferred or new live pigs are added, the corresponding QR code generally needs to be replaced (when new live pigs are added, the back fat, fatness, etc. of the pig will be measured immediately as initial data to improve the live pig information and generate an access QR code). The carrier of the QR code can be an electronic screen, paper or a tablet made of other materials. When the QR code is displayed through an electronic screen, at this time, the electronic screen can generally be interconnected with the system. When the pig pen is changed for the live pig, it is only necessary to replace the live pig QR code corresponding to the pig pen on the system. By scanning the QR code, breeders or other personnel can obtain timely and detailed information about the pig pen and pigs. The information generally includes static information and dynamic information. Static information refers to information that does not change over time. This information can also be directly printed next to the pig pen and pig QR code information card, including: pig pen number idh, pig pen shape and size, pig pen type, pig ear number idp, pig target fatness ht0, pig breed, etc. Through static information, breeders can quickly understand the basic information of the pig pen and the pigs being fed. ; Dynamic information is generally information that changes over time. This type of information needs to be measured and stored in the database in a timely manner in order to obtain effective data in a timely manner, including: ambient temperature t0 and humidity h0 in the pig pen, growth time gt of pigs in the pig pen, growth stage of pigs, current back fat value df of pigs, current fatness or thinness of pigs ht, current pig feed formula, current feed input amount of pigs, etc. In particular, in order to facilitate people to understand the information of pigs more quickly, scanning the QR code will also display the change curve of the back fat value df and fatness or thinness ht of pigs with growth time.

[0025] 2. Pig pen inspection module, this module is mainly used to inspect the entire pig farm regularly. It is usually installed on the track robot of the pig farm (other inspection equipment is also acceptable). The inspection module is equipped with a visible light camera, a depth camera, an infrared camera, and a temperature and humidity sensor. Of course, other equipment can also be installed when more functions are needed, such as a multi-spectral camera, an air sensor, etc. The data collected by the inspection module is generally sent to the back-end server for unified processing (the next module function). Of course, when the computing power resources are sufficient, the collected data can also be directly processed locally, and then the processing results can be directly sent to the server for unified storage and display.

[0026] 3. Data processing module: This module is the core function calculation module, which is generally in the server (for unified processing of all inspection module data) and is also directly integrated in the carrier device of the inspection module. It is mainly responsible for counting and calculating the feed delivery amount fq, feed formula ff and alarm information required by the subsequent intelligent feeding module, and mainly includes:

[0027] (1) Receive image data collected by the visible light camera, which includes two aspects: a) the relevant QR code image of the pig pen information module; b) the scene picture of the entire pig house. In the pig pen information QR code part, the data processing module can also obtain the same information as the person scanning the QR code by scanning the QR code, but in the data processing, the module only needs to obtain the pig pen number idh (to determine the subsequent feed delivery object), the pig ear number idp (to determine which pig is to be calculated later), the pig growth time gt and the pig target fatness ht0. All these data are recorded and wait for the combination of other data to jointly calculate the feed delivery amount fq, feed formula ff and alarm information; in the scene picture part of the pig house, the data processing module uses the target detection algorithm to detect whether there is garbage (such as feces, pig food residues, etc.) in the pig house to ensure the hygiene of the pig house, and on the other hand, detect the pig house. The trough position is determined and the trough image is captured (the target detection neural network can be used, or the trough position data of each pig pen can be built-in in advance, and the corresponding position image can be directly obtained when using it). A simple binary classification network is used to directly determine whether there is any pig food left in the trough (the remaining standard is determined by the annotation standard during the binary classification network training). When the pigs have been fed for a certain period of time (or a period of time before the next meal needs to be fed), if there is still uneaten pig food in the trough, there may be some problems at this time, or the temperature in the pig pen is too high, or the pig feed is too single, or the pig is sick and has a poor appetite. Therefore, a warning message needs to be given to the breeder. Please ask the breeder to clarify the specific reasons based on other feedback data.

[0028] (2) Receive the depth image of the pig's back captured by the depth camera, and calculate the current pig's back fat value df and fatness ht through the pig back fat calculation model and the pig fatness calculation model. All these data are recorded and wait for the feed input fq, feed formula ff and alarm information to be calculated in combination with other data.

[0029] (3) Receive data sent back by the temperature and humidity sensor, record the pig pen ambient temperature t0 and humidity h0, and wait for the comprehensive calculation of feed input amount fq, feed formula ff and alarm information.

[0030] (4) Receive infrared camera data, calculate and record the body temperature tp of the pigs in the pig pen, and then calculate the alarm information.

[0031] (5) Based on the above data, the feed input amount fq, feed formula ff and alarm information are calculated. The feed input amount fq is related to the pig growth time gt (pigs have different food requirements at different stages), pig back fat value df, pig fatness ht and pig target fatness ht0 (the closer the current pig fatness is to the target fatness, the more the required feed amount can be controlled, and the back fat value df can also reflect the fatness of the pig), pig pen temperature t0, humidity h0 (if the temperature and humidity are not suitable, the pig's appetite will be affected). Therefore, an equation for the pig feed input amount fq and these influencing factors can be established: fq = f(gt, df, ht, ht 0, t0, h0), the equation can quickly calculate the amount of feed for the pigs in the current pig pen, so that it is reasonable, scientific and reasonable, to prevent the pigs from overeating and too much food in the pig stomach that has not been digested; in terms of feed formula ff, the present invention mainly considers two aspects: 1) Long-term single feed formula may cause pigs to have a poor appetite; 2) Inappropriate temperature and humidity in the pen will cause a decrease in the feed intake of fattening pigs. Therefore, the timing of changing the feed formula of the present invention is selected when the feeding time of the previous formula feed exceeds the set threshold, and the feed formula is calculated based on the two influencing factors of the pig pen temperature t0 and humidity h0, that is, ff=g(t0,h0). In actual applications, it can be set that when the temperature and humidity are within a certain range, a certain formula of feed is selected, or the proportional relationship of feed ingredients is established according to the temperature t0 and humidity h0, and finally the feed formula is directly calculated; in terms of alarm information, the data processing module calculates the alarm type and sends an alarm instruction to the alarm module of the system. The corresponding alarm judgment criteria are shown in the following table:

[0032]

[0033] It can be seen from the above table that the present invention can not only detect the relevant factors that affect the food intake of pigs, but also find out the situation where the corresponding pig information of the pig pen is forgotten to be changed due to the pig pen change, thereby avoiding the problem of unreasonable feeding amount of pigs in the pen.

[0034] It should be noted that the inspection frequency of the pig pen inspection module is not necessarily consistent with the data processing frequency of the data processing module (that is, not every inspection will run all the functions in the data processing). For example, the remaining food in the data processing is only tested after a period of time after the last feeding (it can also be tested before the current feeding). The back fat, fatness, and feed input and feed formula calculations are all tested once before the feeding is required. This is mainly because operations such as back fat calculations are more computationally resource-intensive and should not be performed frequently. For other calculations, considering the real-time nature of information, the data can be updated every time the inspection is carried out to ensure that the breeders or managers can obtain the latest and valid data and update the management strategy in a timely manner. Of course, in actual applications, the time for feeding will also be preset according to the situation (generally, the time for feeding will be selected when the temperature is suitable) or calculated and inferred based on the ambient temperature and humidity.

[0035] 4. Intelligent feeding module. The intelligent feeding module will communicate with the data processing module and work together with the intelligent feeder to accurately deliver feed on time, according to quantity, formula and pig pen number. It can control the feed delivery amount and feed formula to prevent excessive feeding of pigs, and prevent loss of appetite of pigs through food unfinished detection, temperature and humidity detection and other means, detect problems in time and ensure the healthy growth of pigs.

[0036] 5. Alarm module. The alarm module receives the warning information sent by the data processing module and issues an alarm. The content of the alarm mainly includes: the type of warning, the key indicators of the warning, and the alarm can be a more common alarm mode, such as a warning light, a warning whistle, the system interface (data display module) is highlighted and marked, and a warning message is sent to the mobile phone. Through the alarm information, the management personnel are reminded to solve the problem in time. For example, if the temperature in the pig pen is too high, the fan ventilation needs to be turned on (the fan can be associated with the system to achieve automatic control or remote control); if there is garbage in the pig pen, it needs to be cleaned (it can be cleaned manually or by driving the corresponding cleaning equipment); if there is feed surplus in the pig pen trough, it is necessary to further analyze the cause of the surplus pig food based on other alarm information.

[0037] 6. Data display module. The data display module mainly displays all the collected information of the pig farm in real time. On the one hand, it provides real-time supervision for management personnel. On the other hand, it provides decision-making data for pig farming through data aggregation and analysis. It mainly displays the overall information of the pig farm and the detailed information of each pig pen. The overall information analyzes the overall trend and assists in decision-making analysis. The detailed information focuses on special cases and solves specific problems. Among them, the detailed information mainly includes: the static and dynamic information of the pig pen and the pigs in the pig pen, that is, the scanning information of the QR code of the pig pen information module. The overall information mainly includes: the number of all pig pens in the current pig farm, the number of occupied pig pens, the total number of pigs, the number of pigs in each stage, the daily feed ration, and the pig farm alarm information. In addition, as a direct display module of the system of the present invention, the data display module also supports real-time inspection video display and data playback, etc., to ensure timely discovery of problems and problem tracing.

[0038] The above description of the disclosed embodiments enables one skilled in the art to implement or use the present invention. Various modifications to these embodiments will be apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present invention. Therefore, the present invention will not be limited to the embodiments shown herein, but rather to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A pig farm feed delivery system based on image recognition technology, characterized in that the system include: A pig pen information module, which includes at least a pig pen file card and a pig file card, wherein the pig file card corresponds to the ear number of the pigs fed in the pig pen, and the pig pen file card and the pig file card are displayed in the form of a QR code; The pig pen inspection module is equipped with a visible light camera to collect images of the pig pen file card and the pig file card; a depth camera to collect depth images of the pig backs of the pigs in the pig pen; and a temperature and humidity sensor to collect the pig pen environmental temperature t0 and humidity h0; The data processing module receives the pig pen file card and the pig file card images collected by the pig pen inspection module, obtains the pig pen information and the pig information fed in the pig pen by scanning the pig pen file card and the pig file card QR code, wherein the pig pen information at least includes: the pig pen number idh, the pig information at least includes: the pig ear number idp, the pig growth time gt and the pig target fatness ht0; receives the pig back depth image, calculates the current pig back fat value df and fatness through the pig back fat calculation model and the pig fatness calculation model ht; receiving the data sent back by the temperature and humidity sensor, recording the pig pen ambient temperature t0 and humidity h0, and further calculating the current pig feed input fq in the pig pen, where fq = f(gt, df, ht, ht0, t0, h0), indicating that the pig feed input fq is a function of the pig growth time gt, pig back fat value df, pig fatness ht, pig target fatness ht0, pig pen temperature t0, and humidity h0; the relevant data processing results are sent to the intelligent feeding module, and the sent data at least includes: pig pen number idh and feed input fq; The intelligent feeding module receives the data sent by the data processing module and works in conjunction with the intelligent feeder, matches the corresponding pig pen according to the pig pen number idh, and realizes accurate feed delivery according to the feed delivery amount fq.

2. The pig farm feed delivery system based on image recognition technology according to claim 1 is characterized in that: The data processing module also queries the database according to the pig ear number idp to obtain the historical feed formula of the pig. When the feeding time of the previous formula feed exceeds the set threshold, the data processing module recalculates the feeding formula ff, where ff=g(t0,h0), indicating that the feed formula ff is a function of the pig pen temperature t0 and humidity h0.

3. The pig farm feed delivery system based on image recognition technology according to claim 2 is characterized in that: The data processing module also receives the visible light image data inside the pigpen collected by the pigpen inspection module, inspects the garbage in the pigpen through the target detection algorithm, and evaluates the hygiene conditions of the pigpen.

4. The pig farm feed delivery system based on image recognition technology according to claim 3 is characterized in that: The pig farm feed delivery system also includes an alarm module. When the pig pen temperature t0 and humidity h0 exceed the set temperature and humidity thresholds, the alarm module issues an environmental warning; when the target detection detects the presence of garbage in the pig pen, the alarm module issues a hygiene warning; when the difference Δdf and Δht between the back fat value df and the fatness ht of the pig obtained based on the depth image and the most recent back fat value df_last and fatness ht_last of the pig queried in the database through the pig ear number idp exceeds the set threshold, the system alarm module issues a pig mismatch warning, where Δdf = |df-df_last|, Δht = |ht-ht_last|.

5. The pig farm feed delivery system based on image recognition technology according to claim 4 is characterized in that: The pig pen inspection module is also equipped with an infrared camera to collect infrared data of pigs. The data processing module receives the infrared data of pigs and calculates the body temperature tp of the live pigs in the pig pen. When tp exceeds the set threshold, the system alarm module issues an abnormal body temperature warning for the live pigs.

6. The pig farm feed delivery system based on image recognition technology according to claim 5 is characterized in that: When the intelligent feeding module completes feeding at the set time, the data processing module also processes the visible light image data in the pig pen, captures the image of the pig trough in the pig pen, and determines whether there is feed remaining in the trough through a binary image classification model. When there is feed remaining, the system alarm module issues a feed residue warning.

7. The pig farm feed delivery system based on image recognition technology according to claim 6 is characterized in that: The breeder can scan the QR code in the pig pen information module to obtain static and dynamic information of the pig pen and the pigs in the pen. The static information includes at least: pig pen number idh, pig pen shape and size, pig pen type, pig ear number idp, pig target fatness ht0, and pig breed. The dynamic information includes at least: ambient temperature t0 and humidity h0 in the pig pen, growth time gt of the pigs in the pig pen, growth stage of the pigs, current back fat value df of the pigs, current fatness ht of the pigs, change curve of back fat value df and fatness ht of the pigs with growth time, current pig feed formula, and current pig feed input.

8. The pig farm feed delivery system based on image recognition technology according to claim 7 is characterized in that: The pig farm feed delivery system also includes a data display module. In addition to displaying the static and dynamic information of each pig pen and the pigs in the pig pen, the data display module also at least counts the number of all pig pens in the current pig farm, the number of occupied pig pens, the total number of pigs, the number of pigs in each stage, the daily feed ration, pig farm alarm information and the inspection video data of the pig pen inspection module.

9. The pig farm feed delivery system based on image recognition technology according to claim 8, characterized in that: The data processing module executes the data processing function according to the set frequency and set time.

Images