A chicken farm intelligent audio and video monitoring method and system

Abstract

The present invention relates to the field of smart breeding technology, and specifically discloses an intelligent audio-visual supervision method and system for a chicken farm, the method comprising dividing the breeding area into a plurality of sub-areas, selecting sub-areas according to the danger level, inserting audio-visual points, and installing audio-visual equipment at the audio-visual points; real-time statistics of audio clips of rooster crows and audio clips of natural enemies, building an audio library, regularly selecting audio from the audio library, and sending it to an audio player for playback; acquiring and recording chicken movement information through a camera containing a thermometer, establishing a mapping relationship between audio and chicken movement information; receiving the required movement information sent by the breeder, selecting audio based on the mapping relationship, and sending it to the audio player. The present invention constructs a mapping relationship between audio and chicken movement information, and based on the mapping relationship, the chicken movement information desired by the breeder can be converted into audio and sent to the audio player, so that the movement state of the chickens can be changed, providing a remote management architecture.

Description

Technical Field

[0001] The present invention relates to the field of smart breeding technology, and in particular to an intelligent audio-visual monitoring method and system for a chicken farm. Background Art

[0002] The existing breeding process has gradually changed from free-range breeding to centralized breeding. As a common meat poultry, centralized breeding is more common. However, chickens themselves are very fragile creatures and can be easily affected by the outside world, causing certain losses to the breeders. Therefore, supervision is needed.

[0003] The previous supervision method was that staff members would conduct regular inspections. With the popularization of imaging equipment, cameras are now basically installed at the breeding sites for remote supervision by staff members. However, supervision can only obtain information about the chickens and cannot affect the movement status of the chickens. Although a loudspeaker can be added in the prior art, the staff can have a certain influence on the chickens by shouting, but the effect is not good. In real life, influencing the status of the chickens requires the staff members' language and body movements. How to provide a management architecture that can remotely influence the chickens is a technical problem that the technical solution of the present invention wants to solve. Summary of the invention

[0004] The purpose of the present invention is to provide an intelligent audio and video monitoring method and system for a chicken farm to solve the problems raised in the above-mentioned background technology.

[0005] To achieve the above object, the present invention provides the following technical solutions:

[0006] A chicken farm intelligent audio and video monitoring method, the method comprising:

[0007] Obtaining a floor plan of a chicken farm and decoration information of the chicken farm, and dividing sub-areas containing danger levels in the floor plan according to the decoration information;

[0008] Analyze the sub-areas in turn, select the sub-areas according to the danger level, insert audio and video points, and install audio and video equipment at the audio and video points; the audio and video equipment includes an audio player and a camera with a temperature detector;

[0009] Real-time statistics of audio clips of rooster crows and natural enemy audio clips are collected to build an audio library. Audio clips are selected from the audio library at regular intervals and sent to the audio player for playback.

[0010] The movement information of the chicken flock is obtained and recorded through a camera with a temperature detector, and a mapping relationship between the audio and the movement information of the chicken flock is established;

[0011] The required movement information sent by the breeder is received, audio is selected based on the mapping relationship, and is sent to an audio player.

[0012] As a further solution of the present invention: the step of obtaining the floor plan of the chicken farm and the decoration information of the chicken farm, and dividing the sub-areas containing the danger level in the floor plan according to the decoration information comprises:

[0013] Obtain a floor plan of a chicken farm, and mark an interactive channel in the floor plan; the interactive channel is a connection node between the chicken farm and the outside world, including doors and windows;

[0014] Get the decoration information of the chicken farm, and determine the material and height of the nearest object directly above each position in the floor plan based on the decoration information;

[0015] Calculate a hazard score for each location based on the material and height of the nearest object in the interaction corridor and directly above;

[0016] Calculate the danger score of each location and divide it into sub-areas with danger levels;

[0017] The calculation process of the risk score is as follows:

[0018]

[0019] Where F(x,y) is the risk score at the position (x,y), C 1 and C 2 is the preset coefficient, N is the total number of interactive channels, (x i ,y i ) is the coordinate of the ith interaction channel, H is the height of the nearest object directly above, and Q is the connection strength determined by the material of the nearest object directly above. The connection strength is used to characterize the degree of stability. The lower the connection strength, the higher the probability of falling.

[0020] As a further solution of the present invention: the step of calculating the risk score of each location and dividing the sub-areas containing the risk level includes:

[0021] The hazard score of each location is counted, and the score difference between two adjacent locations is calculated; whether two locations are adjacent is determined by the four-neighborhood judgment rule;

[0022] When the score difference is less than a preset difference threshold, the two positions are classified into one category;

[0023] After all positions have been subjected to the score difference calculation process, the classification results are read to obtain the sub-areas;

[0024] For any sub-area, the average score of the sub-area is calculated, and the danger level is determined according to the magnitude of the average score; one magnitude corresponds to one danger level.

[0025] As a further solution of the present invention, the steps of real-time counting of audio clips of rooster crows and audio clips of natural enemies, building an audio library, regularly selecting audio from the audio library, and sending it to an audio player for playing include:

[0026] Record audio clips of chicken crows and natural enemies to build an audio library;

[0027] Perform data processing on the audio in the audio library, create similar audio, and expand the audio library; the data processing method includes frame extraction and fitting;

[0028] Determine a maximum number, randomly select audio in the audio frequency and combine them, and expand the audio library; wherein the number of the selected audio does not exceed the maximum number;

[0029] Select audio from the audio library at a fixed time, select an audio player, and send the audio to the audio player for playback.

[0030] As a further solution of the present invention: the step of acquiring and recording the movement information of the chicken flock by using a camera having a temperature detector and establishing a mapping relationship between the audio and the movement information of the chicken flock includes:

[0031] The chicken movement video within a preset time after the audio is played is obtained through the camera, and the temperature information is collected in real time through the thermometer as the temperature layer of the chicken movement video;

[0032] Locate the chickens in the chicken movement video based on the temperature layer, connect the chicken positions according to the time sequence, and generate the chicken's movement trajectory;

[0033] Determine the directional relationship between the movement trajectory of each chicken and the audio viewpoint corresponding to the audio player playing the audio, and determine the movement information of the chicken group;

[0034] The played audio and chicken movement information are used as samples to train the neural network model, and the trained neural network model is used as the mapping relationship between the audio and the chicken movement information.

[0035] As a further solution of the present invention: the step of determining the directional relationship between the movement trajectory of each chicken and the audio viewpoint corresponding to the audio player playing the audio, and determining the movement information of the chicken group includes:

[0036] Read the movement trajectory of each chicken in turn, connect the starting point and the audio-visual point corresponding to the audio player playing the audio, and obtain the theoretical line segment;

[0037] Calculate the projection of the chicken's motion trajectory on the theoretical line segment, and create the chicken's motion vector according to the projection; when the motion trajectory points to the sound viewpoint, the direction of the motion vector is positive, and when the motion trajectory is away from the sound viewpoint, the direction of the motion vector is negative;

[0038] Accumulate the motion vectors of all chickens as the chicken group motion information.

[0039] The technical solution of the present invention also provides an intelligent audio and video monitoring system for a chicken farm, the system comprising:

[0040] The area segmentation module is used to obtain the floor plan of the chicken farm and the decoration information of the chicken farm, and segment the sub-areas containing the danger level in the floor plan according to the decoration information;

[0041] An audio-visual equipment installation module is used to analyze the sub-areas in turn, select the sub-areas according to the danger level, insert audio-visual points, and install audio-visual equipment at the audio-visual points; the audio-visual equipment includes an audio player and a camera with a temperature detector;

[0042] The audio playback module is used to count the audio clips of rooster crows and natural enemy audio clips in real time, build an audio library, select audio from the audio library at regular intervals, and send it to the audio player for playback;

[0043] A mapping relationship establishment module is used to obtain and record the movement information of the chicken flock through a camera with a temperature detector, and establish a mapping relationship between the audio and the movement information of the chicken flock;

[0044] The mapping relationship application module is used to receive the required movement information sent by the breeding party, select audio based on the mapping relationship, and send it to the audio player.

[0045] As a further solution of the present invention: the region segmentation module includes:

[0046] An interactive channel marking unit is used to obtain a floor plan of a chicken farm and mark an interactive channel in the floor plan; the interactive channel is a connection node between the chicken farm and the outside world, including doors and windows;

[0047] The object query unit is used to obtain the decoration information of the chicken farm and determine the material and height of the nearest object directly above each position in the floor plan according to the decoration information;

[0048] A danger score calculation unit, for calculating a danger score for each location based on the material and height of the interaction channel and the nearest object directly above;

[0049] The segmentation execution unit is used to calculate the risk score of each location and segment the sub-areas containing the risk level;

[0050] The calculation process of the risk score is as follows:

[0051]

[0052] Where F(x,y) is the risk score at the position (x,y), C 1 and C 2is the preset coefficient, N is the total number of interactive channels, (x i ,y i ) is the coordinate of the ith interaction channel, H is the height of the nearest object directly above, and Q is the connection strength determined by the material of the nearest object directly above. The connection strength is used to characterize the degree of stability. The lower the connection strength, the higher the probability of falling.

[0053] As a further solution of the present invention: the audio playback module includes:

[0054] A recording unit, used to record audio clips of chicken calls and natural enemies, and build an audio library;

[0055] The first expansion unit is used to perform data processing on the audio in the audio library, create similar audio, and expand the audio library; wherein the data processing method includes frame extraction and fitting;

[0056] A second expansion unit is used to determine a maximum number, randomly select audios in the audio frequency and combine them to expand the audio library; wherein the number of the selected audios does not exceed the maximum number;

[0057] The selection and playback unit is used to select audio from the audio library at a fixed time, select an audio player, and send the audio to the audio player for playback.

[0058] As a further solution of the present invention: the mapping relationship establishment module includes:

[0059] A video acquisition unit is used to obtain a chicken movement video within a preset time period after the audio is played through a camera, and a temperature layer generation unit is used to collect temperature information in real time through a temperature meter as a temperature layer of the chicken movement video;

[0060] A positioning unit, used to locate the chickens in the chicken movement video based on the temperature layer, connect the chickens' positions according to the time sequence, and generate the chickens' movement trajectory;

[0061] A motion information generating unit, used to determine the directional relationship between the motion trajectory of each chicken and the audio viewpoint corresponding to the audio player playing the audio, and determine the motion information of the chicken group;

[0062] The model training unit is used to use the played audio and the chicken movement information as samples to train the neural network model, and use the trained neural network model as the mapping relationship between the audio and the chicken movement information.

[0063] Compared with the prior art, the beneficial effects of the present invention are: the present invention sets a camera with an audio player, selects pre-recorded audio for playback, and records the movement information of the chicken flock at the same time, and constructs a mapping relationship between the audio and the movement information of the chicken flock. In practical applications, based on the mapping relationship, the chicken flock movement information desired by the breeder can be converted into audio and sent to the audio player, thereby changing the movement state of the chicken flock, and providing a remote management architecture. BRIEF DESCRIPTION OF THE DRAWINGS

[0064] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only some embodiments of the present invention.

[0065] Figure 1 This is a flowchart of the intelligent audio and video monitoring method for chicken farms.

[0066] Figure 2 This is the first sub-process flowchart of the intelligent audio and video monitoring method for chicken farms.

[0067] Figure 3 This is the second sub-process flowchart of the intelligent audio and video monitoring method for chicken farms.

[0068] Figure 4 This is the third sub-process flowchart of the intelligent audio and video monitoring method for chicken farms.

[0069] Figure 5 This is the structural diagram of the intelligent audio and video monitoring system for a chicken farm. DETAILED DESCRIPTION

[0070] In order to make the technical problems, technical solutions and beneficial effects to be solved by the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not used to limit the present invention.

[0071] Figure 1 The flowchart of the intelligent audio and video monitoring method of a chicken farm is shown in the following figure. In an embodiment of the present invention, an intelligent audio and video monitoring method of a chicken farm is provided, and the method includes:

[0072] Step S100: obtaining a floor plan of a chicken farm and decoration information of the chicken farm, and dividing sub-areas containing danger levels in the floor plan according to the decoration information;

[0073] The application scenario of the technical solution of the present invention is a chicken farm with smart devices installed. The floor plan of the chicken farm is obtained as a map of the chicken farm, and the decoration information of the chicken farm is obtained, wherein the decoration information is which additional equipment or buildings are installed in the chicken farm, among which the roof of the chicken farm is regarded as a building; the decoration information is analyzed based on the floor plan, the floor plan can be divided into multiple areas, and the danger level of each area can be determined simultaneously; generally speaking, for a certain location, the closer to the outside world, the more dangerous it is (the easier it is to come into contact with natural enemies outside), and the lower the installation density of the equipment or building above (the easier it is to fall), the more dangerous it is.

[0074] Step S200: Analyze the sub-areas in turn, select the sub-areas according to the danger level, insert audio and video points, and install audio and video equipment at the audio and video points; the audio and video equipment includes an audio player and a camera with a temperature detector;

[0075] Each divided area is analyzed in turn to query the danger level of the area. When the danger level is high enough, an audio-visual point is inserted in the corresponding area, and audio-visual equipment is installed at the audio-visual point. The audio-visual equipment in this application can be a camera with audio interaction function and temperature monitoring function.

[0076] It is worth mentioning that the audio and video point selected in the area is generally the center point of the area.

[0077] Step S300: Count the audio clips of rooster crows and natural enemy audio clips in real time, build an audio library, select audio from the audio library at regular intervals, and send it to an audio player for playback;

[0078] Count the audio clips of crowing chickens and natural enemies, both of which can affect the movement of the chickens. The natural enemy audio clips in the present application include not only animal calls (such as dog barking), but also natural sounds (such as thunder), sounds that can be captured by chickens and produce emotions, and even include crowing chickens during fighting. These are collectively referred to as natural enemy audio clips; further, the crowing chicken audio clips are mainly young chicken audio, which can attract the chickens.

[0079] In fact, the crowing audio clip and the natural enemy audio clip in this application are audios with two functions. The former is the audio to attract chickens, and the latter is the audio to drive away chickens.

[0080] Step S400: Acquire and record the movement information of the chickens through a camera with a temperature detector, and establish a mapping relationship between the audio and the movement information of the chickens;

[0081] The camera is used to obtain real-time video of the chicken farm. Since the camera contains a thermometer, the chicken positioning process is very easy. By identifying the video with the help of temperature information, the chicken movement information can be obtained. The chicken movement information is generated by the audio played in step S300. Therefore, the audio is regarded as input and the chicken movement information is regarded as output, and a mapping relationship from audio to chicken movement information can be constructed.

[0082] Step S500: receiving the required movement information sent by the breeder, selecting audio based on the mapping relationship, and sending it to the audio player;

[0083] In the actual application stage, the breeder will input the required movement information. The required movement information is the required chicken movement information. Its specific format is not limited. The simplest way is to use the same data structure as the chicken movement information. A more complicated way is to simplify the operation of the breeder. The breeder can input a destination based on the house plan, and then the execution subject of this method generates the required movement information pointing to the destination.

[0084] Match the required movement information with the chicken movement information and query the corresponding audio. This is the inverse application of the mapping relationship. After the corresponding audio is queried, it can be sent to the audio player.

[0085] It should be noted that when sending to an audio player, generally only one audio player will be selected. If it is audio to attract chickens, the audio player closest to the destination (the end point of the required movement information) will be selected. If it is audio to drive away chickens, the audio player farthest from the destination will be selected.

[0086] Figure 2 This is a first sub-process flowchart of the intelligent audio-visual supervision method for a chicken farm. The steps of obtaining the floor plan and the decoration information of the chicken farm, and dividing the sub-areas containing danger levels in the floor plan according to the decoration information include:

[0087] Step S101: Obtain a floor plan of a chicken farm, and mark an interactive channel in the floor plan; the interactive channel is a connection node between the chicken farm and the outside world, including doors and windows;

[0088] Step S102: Obtaining decoration information of the chicken farm, and determining the material and height of the nearest object directly above each position in the floor plan according to the decoration information;

[0089] Step S103: Calculate the danger score of each location based on the material and height of the interactive channel and the nearest object directly above;

[0090] Step S104: Count the risk scores of each location and segment out sub-areas containing risk levels.

[0091] The above content makes specific restrictions on the sub-area segmentation process, obtains the floor plan of the chicken farm, and marks the interactive channels in the floor plan. The interactive channels are mainly doors and windows, and natural enemies may reach the inside from the outside. It is worth mentioning that part of the chicken farm is actually open-air. In this case, the interactive channel is the door. Only in a closed space will the interactive channel include doors and windows.

[0092] The decoration information of the chicken farm is obtained. The decoration information is a general concept, which indicates the equipment and buildings installed in the chicken farm, such as lamps and ceilings.

[0093] For any location in the chicken farm, the distance between the location and all interactive channels is obtained, and then the equipment and buildings above the location are queried, and the two are combined to determine the risk score of each location. Among them, the distance between adjacent locations in the chicken farm is a preset value. The smaller the distance, the more locations, the finer the area division, and correspondingly, the more resources are invested.

[0094] The calculation process of the risk score is as follows:

[0095]

[0096] Where F(x,y) is the risk score at the position (x,y), C 1 and C 2 is the preset coefficient, N is the total number of interactive channels, (x i ,y i ) is the coordinate of the ith interaction channel, H is the height of the nearest object directly above, and Q is the connection strength determined by the material of the nearest object directly above. The connection strength is used to characterize the degree of stability. The lower the connection strength, the higher the probability of falling.

[0097] The calculation principle of the risk score is that, for a certain location, the closer it is to the interaction channel, the greater the inverse of the distance. The sum of the inverses of the distances to all interaction channels is calculated, and the resulting value is used to reflect the degree of risk caused by the interaction channel. The height and connection strength of the nearest object directly above are obtained. The greater the height, the higher the risk, and the greater the connection strength, the lower the risk.

[0098] In summary, the first item in the calculation process reflects the risk brought by the interaction channel, and the second item reflects the risk brought by the building.

[0099] Specifically, the step of calculating the danger score of each location and dividing the sub-areas containing danger levels includes:

[0100] The hazard score of each location is counted, and the score difference between two adjacent locations is calculated; whether two locations are adjacent is determined by the four-neighborhood judgment rule;

[0101] When the score difference is less than a preset difference threshold, the two positions are classified into one category;

[0102] After all positions have been subjected to the score difference calculation process, the classification results are read to obtain the sub-areas;

[0103] For any sub-area, the average score of the sub-area is calculated, and the danger level is determined according to the magnitude of the average score; one magnitude corresponds to one danger level.

[0104] The above content provides a specific solution for generating regions based on positions. This process can be compared to the image contour recognition process. Each position is regarded as a pixel point, and its risk score is regarded as a grayscale value. The risk scores of adjacent positions are compared and the score difference is calculated. If the score difference is small, it means that the two positions are similar enough and they are classified into one category; if the score difference is large, it means that the two positions belong to two categories.

[0105] After all locations have been analyzed, multiple sub-regions can be obtained.

[0106] Among them, whether the positions are adjacent adopts the four-neighborhood judgment rule, that is, the four positions above, below, left and right of a position are regarded as its adjacent positions.

[0107] Figure 3 This is a second sub-flow diagram of the intelligent audio and video supervision method for a chicken farm. The steps of real-time statistics of chicken crowing audio clips and natural enemy audio clips, building an audio library, regularly selecting audio from the audio library, and sending it to an audio player for playback include:

[0108] Step S301: Recording audio clips of rooster crowing and natural enemy audio clips to build an audio library;

[0109] Step S302: performing data processing on the audio in the audio library, creating similar audio, and expanding the audio library; wherein the data processing method includes frame extraction and fitting;

[0110] Step S303: determining a maximum number, randomly selecting and combining audios in the audio frequency, and expanding the audio library; wherein the number of the selected audios does not exceed the maximum number;

[0111] Step S304: regularly selecting audio from the audio library, selecting an audio player, and sending the audio to the audio player for playing.

[0112] In an example of the technical solution of the present invention, the construction and application stages of the audio library are limited; some audio clips of rooster crows and natural enemies are recorded to obtain an initial audio library, and the number of audios in the initial audio library is limited and related to the environment during recording; the audios in the initial audio library are processed, such as frame extraction (deleting certain frames) and fitting (adjusting each frame data according to the mean to make it closer to the mean), and a variety of new audios based on the initial audio can be obtained, thereby expanding the audio library.

[0113] In addition, there may be a matching relationship between different audios. The breeder may pre-set a quantity range, randomly determine a quantity within this quantity range, select the determined number of audios for combination, obtain the combined audio, and further expand the audio library.

[0114] Select audio from the expanded audio library, then select one from multiple audio players (only one can be selected, otherwise interference will occur), and send the audio to the audio player for playback.

[0115] It is worth mentioning that the similar audio generation process in step S302 can also adopt other audio conversion schemes, which is equivalent to a detail processing process on the original audio. Preprocessing processes such as denoising can be used as one of the processing schemes.

[0116] Figure 4 This is a flowchart of the third sub-process of the intelligent audio-visual monitoring method for a chicken farm. The steps of acquiring and recording the movement information of the chicken flock through a camera with a temperature detector and establishing a mapping relationship between the audio and the movement information of the chicken flock include:

[0117] Step S401: obtaining a chicken movement video within a preset time after the audio is played through a camera, and collecting temperature information in real time through a thermometer as a temperature layer of the chicken movement video;

[0118] Step S402: locating the positions of the chickens in the chicken movement video based on the temperature layer, connecting the positions of the chickens according to the time sequence, and generating the movement trajectory of the chickens;

[0119] Step S403: determining the directional relationship between the movement trajectory of each chicken and the audio viewpoint corresponding to the audio player playing the audio, and determining the movement information of the chicken group;

[0120] Step S404: using the played audio and the chicken movement information as samples to train a neural network model, and using the trained neural network model as a mapping relationship between the audio and the chicken movement information.

[0121] After playing the audio, the camera is used to obtain a video of the chickens' movement within a preset time period after the audio is played. The preset time period can be half a minute, one minute or longer. At the same time, the temperature information is collected in real time through the thermometer as the temperature layer of the chickens' movement video. The temperature is used to locate the chickens, making the video recognition process easier. By identifying the video, the movement trajectory of each chicken can be obtained.

[0122] The movement trajectory of each chicken is analyzed to obtain the movement information of the flock.

[0123] The played audio and chicken movement information are used as samples to train a neural network model, and the trained neural network model is used as a mapping relationship between audio and chicken movement information. In the technical solution of the present application, the process of using the played audio and chicken movement information as samples actually includes two situations. One is to use audio as input and chicken movement information as output, and the other is to use chicken movement information as input and audio as output. There are two types of trained neural network models. In the application stage, it is necessary to query audio from chicken movement information. If it is a neural network model from audio to chicken movement information, it is necessary to test multiple audios. When the output obtained is similar enough to the required movement information, the audio is used as the corresponding audio. If it is a neural network model from chicken movement information to audio, the required movement information can be directly input to obtain the audio.

[0124] Specifically, the step of determining the directional relationship between the movement trajectory of each chicken and the audio viewpoint corresponding to the audio player playing the audio, and determining the movement information of the chicken group includes:

[0125] Read the movement trajectory of each chicken in turn, connect the starting point and the audio-visual point corresponding to the audio player playing the audio, and obtain the theoretical line segment;

[0126] Calculate the projection of the chicken's motion trajectory on the theoretical line segment, and create the chicken's motion vector according to the projection; when the motion trajectory points to the sound viewpoint, the direction of the motion vector is positive, and when the motion trajectory is away from the sound viewpoint, the direction of the motion vector is negative;

[0127] Accumulate the motion vectors of all chickens as the chicken group motion information.

[0128] The above content specifically refines the concept of chicken flock motion information and converts the chicken flock motion information into a vector. The generation process is described as follows:

[0129] For each chicken, there is a movement trajectory, which may point to the audio player or move away from the audio player. By connecting the starting point of the movement trajectory and the audio-visual point corresponding to the audio player, a line segment can be obtained. Then, the projection of the movement trajectory on the line segment is calculated. The length of the projection can reflect the impact of the audio on the chicken. The longer the length, the greater the impact. The actual meaning of the projection length is how much the chicken is close to (or away from) the audio-visual point under the influence of the audio within the preset time length. The greater the impact, the greater the distance.

[0130] In the process of calculating the projection, it is determined whether the motion trajectory points to the audio-visual point or away from the audio-visual point, thereby adding a direction to the projection to obtain a vector. Finally, the motion trajectory of each chicken is converted into a vector, and the chicken flock movement information is converted into a vector accumulation result.

[0131] Figure 5 1 is a structural diagram of an intelligent audio-visual monitoring system for a chicken farm. In an embodiment of the present invention, an intelligent audio-visual monitoring system for a chicken farm is provided. The system 10 includes:

[0132] The area segmentation module 11 is used to obtain the floor plan of the chicken farm and the decoration information of the chicken farm, and segment the sub-areas containing the danger level in the floor plan according to the decoration information;

[0133] The audio-visual equipment installation module 12 is used to analyze the sub-areas in turn, select the sub-areas according to the danger level, insert the audio-visual points, and install the audio-visual equipment at the audio-visual points; the audio-visual equipment includes an audio player and a camera with a temperature detector;

[0134] The audio playback module 13 is used to count the audio clips of chicken crows and natural enemy audio clips in real time, build an audio library, select audio from the audio library at regular intervals, and send it to the audio player for playback;

[0135] A mapping relationship establishing module 14 is used to obtain and record the movement information of the chicken flock through a camera with a temperature detector, and establish a mapping relationship between the audio and the movement information of the chicken flock;

[0136] The mapping relationship application module 15 is used to receive the required movement information sent by the breeder, select audio based on the mapping relationship, and send it to the audio player.

[0137] Furthermore, the region segmentation module 11 includes:

[0138] An interactive channel marking unit is used to obtain a floor plan of a chicken farm and mark an interactive channel in the floor plan; the interactive channel is a connection node between the chicken farm and the outside world, including doors and windows;

[0139] The object query unit is used to obtain the decoration information of the chicken farm and determine the material and height of the nearest object directly above each position in the floor plan according to the decoration information;

[0140] A danger score calculation unit, for calculating a danger score for each location based on the material and height of the interaction channel and the nearest object directly above;

[0141] The segmentation execution unit is used to calculate the risk score of each location and segment the sub-areas containing the risk level;

[0142] The calculation process of the risk score is as follows:

[0143]

[0144] Where F(x,y) is the risk score at the position (x,y), C 1 and C 2 is the preset coefficient, N is the total number of interactive channels, (x i ,y i ) is the coordinate of the ith interaction channel, H is the height of the nearest object directly above, and Q is the connection strength determined by the material of the nearest object directly above. The connection strength is used to characterize the degree of stability. The lower the connection strength, the higher the probability of falling.

[0145] Specifically, the audio playback module 13 includes:

[0146] A recording unit, used to record audio clips of chicken calls and natural enemies, and build an audio library;

[0147] The first expansion unit is used to perform data processing on the audio in the audio library, create similar audio, and expand the audio library; wherein the data processing method includes frame extraction and fitting;

[0148] A second expansion unit is used to determine a maximum number, randomly select audios in the audio frequency and combine them to expand the audio library; wherein the number of the selected audios does not exceed the maximum number;

[0149] The selection and playback unit is used to select audio from the audio library at a fixed time, select an audio player, and send the audio to the audio player for playback.

[0150] Furthermore, the mapping relationship establishing module 14 includes:

[0151] A video acquisition unit is used to obtain a chicken movement video within a preset time period after the audio is played through a camera, and a temperature layer generation unit is used to collect temperature information in real time through a temperature meter as a temperature layer of the chicken movement video;

[0152] A positioning unit, used to locate the chickens in the chicken movement video based on the temperature layer, connect the chickens' positions according to the time sequence, and generate the chickens' movement trajectory;

[0153] A motion information generating unit, used to determine the directional relationship between the motion trajectory of each chicken and the audio viewpoint corresponding to the audio player playing the audio, and determine the motion information of the chicken group;

[0154] The model training unit is used to use the played audio and the chicken movement information as samples to train the neural network model, and use the trained neural network model as the mapping relationship between the audio and the chicken movement information.

[0155] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions and improvements made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An intelligent audio and video monitoring method for a chicken farm, characterized in that: The method comprises: Obtaining a floor plan of a chicken farm and decoration information of the chicken farm, and dividing sub-areas containing danger levels in the floor plan according to the decoration information; Analyze the sub-areas in turn, select the sub-areas according to the danger level, insert audio and video points, and install audio and video equipment at the audio and video points; the audio and video equipment includes an audio player and a camera with a temperature detector; Real-time statistics of audio clips of rooster crows and natural enemy audio clips are collected to build an audio library. Audio clips are selected from the audio library at regular intervals and sent to the audio player for playback. The movement information of the chicken flock is obtained and recorded through a camera with a temperature detector, and a mapping relationship between the audio and the movement information of the chicken flock is established; receiving the required movement information sent by the breeder, selecting the audio based on the mapping relationship, and sending it to the audio player; The step of obtaining the floor plan of the chicken farm and the decoration information of the chicken farm, and dividing the sub-areas containing the danger level in the floor plan according to the decoration information comprises: Obtain a floor plan of a chicken farm, and mark an interactive channel in the floor plan; the interactive channel is a connection node between the chicken farm and the outside world, including doors and windows; Get the decoration information of the chicken farm, and determine the material and height of the nearest object directly above each position in the floor plan based on the decoration information; Calculate a hazard score for each location based on the material and height of the nearest object in the interaction corridor and directly above; Calculate the danger score of each location and divide it into sub-areas with danger levels; The calculation process of the risk score is as follows: Where F(x,y) is the risk score at the position (x,y), C1 and C2 are the preset coefficients, N is the total number of interaction channels, (x i ,y i ) is the coordinate of the ith interaction channel, H is the height of the nearest object directly above, and Q is the connection strength determined by the material of the nearest object directly above. The connection strength is used to characterize the degree of stability. The lower the connection strength, the higher the probability of falling.

2. The intelligent audio-visual monitoring method for a chicken farm according to claim 1 is characterized in that: The step of calculating the danger score of each location and dividing the sub-areas containing danger levels includes: The hazard score of each location is counted, and the score difference between two adjacent locations is calculated; whether two locations are adjacent is determined by the four-neighborhood judgment rule; When the score difference is less than a preset difference threshold, the two positions are classified into one category; After all positions have been subjected to the score difference calculation process, the classification results are read to obtain the sub-areas; For any sub-area, the average score of the sub-area is calculated, and the danger level is determined according to the magnitude of the average score; one magnitude corresponds to one danger level.

3. The intelligent audio-visual monitoring method for a chicken farm according to claim 1 is characterized in that: The steps of real-time counting of audio clips of rooster crows and audio clips of natural enemies, building an audio library, regularly selecting audio from the audio library, and sending the audio to an audio player for playing include: Record audio clips of chicken crows and natural enemies to build an audio library; Perform data processing on the audio in the audio library, create similar audio, and expand the audio library; the data processing method includes frame extraction and fitting; Determine a maximum number, randomly select audio in the audio frequency and combine them, and expand the audio library; wherein the number of the selected audio does not exceed the maximum number; Select audio from the audio library at a fixed time, select an audio player, and send the audio to the audio player for playback.

4. The intelligent audio-visual monitoring method for a chicken farm according to claim 1 is characterized in that: The step of obtaining and recording the movement information of the chicken flock by using a camera having a temperature detector and establishing a mapping relationship between the audio and the movement information of the chicken flock comprises: The chicken movement video within a preset time after the audio is played is obtained through the camera, and the temperature information is collected in real time through the thermometer as the temperature layer of the chicken movement video; Locate the chickens in the chicken movement video based on the temperature layer, connect the chicken positions according to the time sequence, and generate the chicken's movement trajectory; Determine the directional relationship between the movement trajectory of each chicken and the audio viewpoint corresponding to the audio player playing the audio, and determine the movement information of the chicken group; The played audio and chicken movement information are used as samples to train the neural network model, and the trained neural network model is used as the mapping relationship between the audio and the chicken movement information.

5. The intelligent audio-visual monitoring method for a chicken farm according to claim 4 is characterized in that: The step of determining the directional relationship between the movement trajectory of each chicken and the audio viewpoint corresponding to the audio player playing the audio, and determining the movement information of the chicken group includes: Read the movement trajectory of each chicken in turn, connect the starting point and the audio-visual point corresponding to the audio player playing the audio, and obtain the theoretical line segment; Calculate the projection of the chicken's motion trajectory on the theoretical line segment, and create the chicken's motion vector according to the projection; when the motion trajectory points to the sound viewpoint, the direction of the motion vector is positive, and when the motion trajectory is away from the sound viewpoint, the direction of the motion vector is negative; Accumulate the motion vectors of all chickens as the chicken group motion information.

6. An intelligent audio and video monitoring system for chicken farms, characterized in that: The system comprises: The area segmentation module is used to obtain the floor plan of the chicken farm and the decoration information of the chicken farm, and segment the sub-areas containing the danger level in the floor plan according to the decoration information; An audio-visual equipment installation module is used to analyze the sub-areas in turn, select the sub-areas according to the danger level, insert audio-visual points, and install audio-visual equipment at the audio-visual points; the audio-visual equipment includes an audio player and a camera with a temperature detector; The audio playback module is used to count the audio clips of rooster crows and natural enemy audio clips in real time, build an audio library, select audio from the audio library at regular intervals, and send it to the audio player for playback; A mapping relationship establishment module is used to obtain and record the movement information of the chicken flock through a camera with a temperature detector, and establish a mapping relationship between the audio and the movement information of the chicken flock; A mapping relationship application module, used for receiving the required movement information sent by the breeder, selecting audio based on the mapping relationship, and sending it to the audio player; The region segmentation module includes: An interactive channel marking unit is used to obtain a floor plan of a chicken farm and mark an interactive channel in the floor plan; the interactive channel is a connection node between the chicken farm and the outside world, including doors and windows; The object query unit is used to obtain the decoration information of the chicken farm and determine the material and height of the nearest object directly above each position in the floor plan according to the decoration information; A danger score calculation unit, for calculating a danger score for each location based on the material and height of the interaction channel and the nearest object directly above; The segmentation execution unit is used to calculate the risk score of each location and segment the sub-areas containing the risk level; The calculation process of the risk score is as follows: Where F(x,y) is the risk score at the position (x,y), C1 and C2 are the preset coefficients, N is the total number of interaction channels, (x i ,y i ) is the coordinate of the ith interaction channel, H is the height of the nearest object directly above, and Q is the connection strength determined by the material of the nearest object directly above. The connection strength is used to characterize the degree of stability. The lower the connection strength, the higher the probability of falling.

7. The intelligent audio and video monitoring system for chicken farms according to claim 6 is characterized in that: The audio playback module includes: A recording unit, used to record audio clips of chicken calls and natural enemies, and build an audio library; The first expansion unit is used to perform data processing on the audio in the audio library, create similar audio, and expand the audio library; wherein the data processing method includes frame extraction and fitting; A second expansion unit is used to determine a maximum number, randomly select audios in the audio frequency and combine them to expand the audio library; wherein the number of the selected audios does not exceed the maximum number; The selection and playback unit is used to select audio from the audio library at a fixed time, select an audio player, and send the audio to the audio player for playback.

8. The intelligent audio and video monitoring system for chicken farms according to claim 6 is characterized in that: The mapping relationship establishment module includes: A video acquisition unit is used to obtain a flock of chickens motion video within a preset time period after the audio is played through a camera, and a temperature layer generation unit is used to collect temperature information in real time through a temperature detector as a temperature layer of the flock of chickens motion video; A positioning unit, used to locate the chickens in the chicken movement video based on the temperature layer, connect the chickens' positions according to the time sequence, and generate the chickens' movement trajectory; A motion information generating unit, used to determine the directional relationship between the motion trajectory of each chicken and the audio viewpoint corresponding to the audio player playing the audio, and determine the motion information of the chicken group; The model training unit is used to use the played audio and the chicken movement information as samples to train the neural network model, and use the trained neural network model as the mapping relationship between the audio and the chicken movement information.

Images