Chicken house farming safety prevention and control method and system based on self-learning

Through self-learning algorithms, the intelligent management of safety prevention and control of chicken house breeding is realized, and the problems of low efficiency and high error rate of traditional methods are solved, ensuring safe production of chicken houses.

CN115410304BActive Publication Date: 2024-11-26SHANDONG DAJIA MASCH CO LTD +1
View PDF 0 Cites 0 Cited by

Patent Information

Application Number
CN202211032405.4
Authority / Receiving Office
CN · China
Patent Type
Patents(China)
Current Assignee / Owner
Filing Date
2022-08-26
Publication Date
2024-11-26
Estimated Expiration
2042-08-26

AI Technical Summary

Technical Problem

Traditional chicken house safety prevention and control methods are inefficient and have high error rates, and they cannot effectively identify and record the abnormal behavior of breeders, resulting in the safety production of chicken houses being affected.

Method used

The safety prevention and control methods and systems of chicken house breeding based on self-learning are adopted, and the access control control is optimized through face recognition and behavior monitoring, combined with self-learning algorithms, and the identity and behavior of people entering the chicken house are identified and recorded to achieve automated security prevention and control.

Benefits of technology

It improves the safety prevention and control efficiency of chicken house breeding, reduces the error rate caused by human monitoring, and realizes intelligent and efficient safety management, preventing the harm of outsiders and abnormal behaviors to chicken house production.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN115410304B_ABST
    Figure CN115410304B_ABST
Patent Text Reader

Abstract

The present invention discloses a chicken house farming safety prevention and control method and system based on self-learning, the method comprising: collecting facial images of personnel and performing facial recognition and matching, judging whether the personnel to be entered and exited are wearing masks, if so, collecting the fingerprints or ID card information of the personnel to be entered and exited, and controlling the access control working state with this recognition result, otherwise directly controlling the access control working state according to the facial recognition and matching results, and updating the facial information of the entering personnel through a self-learning algorithm if the match is successful; collecting the behavior images of the personnel entering the chicken house in real time, obtaining the behavior data of the entering personnel, and judging whether the behavior of the entering personnel is unknown behavior, and if so, updating the behavior information through a self-learning algorithm, otherwise judging whether the known behavior is abnormal behavior. The present invention optimizes facial recognition access control and abnormal behavior recognition by autonomous learning, and solves the problem of low efficiency and high error rate of chicken house farming safety prevention and control work caused by human monitoring.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present invention relates to the technical field of safety control in poultry and livestock farms, and in particular to a chicken house farming safety control method and system based on self-learning. Background Art

[0002] Since the scale of the chicken farming technology industry, scientific and effective feeding and management methods have been continuously sought to promote standardized livestock and poultry farming and build a modern farming system. Farming safety prevention and control plays a pivotal role in livestock and poultry farming, especially in chicken house farming. Preventing outsiders from wandering around the chicken house and affecting the safe production of the chicken house, and identifying the unnecessary losses caused by abnormal behavior of farmers to the production of the chicken house, has become the focus of chicken house management.

[0003] Traditional chicken house safety control methods and systems usually use the method of locking the chicken house door or installing a password lock or fingerprint lock for safety control. However, this solution cannot achieve barrier-free unlocking, and the identification of abnormal behavior of farmers requires chicken house managers to keep a close eye on the chicken house monitoring at all times. While wasting manpower and material resources, manual monitoring leads to low efficiency in chicken house farming safety control and a high error rate in monitoring. Summary of the invention

[0004] In order to solve the shortcomings of the above-mentioned prior art, the present invention provides a chicken house farming safety control method and system based on self-learning, which optimizes face recognition access control through autonomous learning, identifies and records abnormal behaviors of farmers entering the chicken house, and further optimizes abnormal behavior recognition through self-learning, thereby solving the problem of low work efficiency and high error rate of chicken house farming safety control due to personnel restrictions, and realizing intelligent, efficient and safe control of chicken house farming.

[0005] In a first aspect, the present disclosure provides a chicken house farming safety control method based on self-learning, comprising:

[0006] Collect facial images of people waiting to enter and exit the chicken house and perform facial recognition and matching to determine whether the people waiting to enter and exit are wearing masks. If so, collect the fingerprint or ID card information of the people waiting to enter and exit, and control the access control working status according to the fingerprint or ID card information recognition results. Otherwise, directly control the access control working status according to the facial recognition and matching results. If the match is successful, update the facial information of the people entering and exiting through the self-learning algorithm;

[0007] The behavioral images of people entering the chicken house are collected in real time, and the behavioral data of the people entering are obtained based on the behavioral images to determine whether the behavior of the people entering is unknown behavior. If so, the behavior information is updated through the self-learning algorithm. Otherwise, it is determined whether the known behavior is abnormal behavior. If so, an alarm is issued.

[0008] A further technical solution is to perform face recognition based on the collected face images of the people to be in and out of the chicken house, extract the facial feature data of the people, and perform face matching based on the extracted facial feature data;

[0009] If a normal match is detected, face matching is performed based on the extracted facial feature data and the facial feature data of all breeding personnel pre-stored in the personnel information repository;

[0010] If the match is successful, the door will be controlled to open and the behavior of the person entering will be identified. At the same time, the facial feature data of the person pre-stored in the personnel information repository will be updated through the self-learning algorithm;

[0011] If the match fails, the access control will still be closed, and the facial feature data of the person extracted will be stored as facial feature data of a suspicious person through a self-learning algorithm.

[0012] In a further technical solution, the updating of the facial feature data pre-stored in the personnel information repository by the self-learning algorithm specifically refers to:

[0013] If the number of facial feature data of the person stored in the personnel information repository is less than the set number, the facial feature data extracted this time is placed at the end of the queue of facial feature data stored for the person;

[0014] If the facial feature data of the person stored in the personnel information repository is sufficient to meet the set number, the deviation values ​​of the facial feature data originally stored for the person and the facial feature data extracted this time are calculated respectively according to the facial feature deviation model, and the facial feature data with the largest deviation value is deleted. At the same time, the facial feature data extracted this time is placed at the end of the queue of the facial feature data stored for the person, thereby completing the self-learning update.

[0015] A further technical solution is that if it is detected that a normal match cannot be made, the fingerprint or ID card information of the person to be entered or exited is collected, the fingerprint feature data or the ID card feature data is extracted, and data matching is performed based on the extracted fingerprint feature data or the ID card feature data and the fingerprint feature data or the ID card feature data of all the breeding personnel pre-stored in the personnel information repository;

[0016] If the match is successful, the access control will be opened, and the person’s ID information and entry and exit time will be recorded, and the person entering will be tracked and detected.

[0017] A further technical solution is that the tracking detection refers to matching the skin color characteristics and body shape characteristics of the person entering, specifically:

[0018] Collecting behavioral images of people entering the chicken house in real time, obtaining images containing the people entering, identifying the people entering the image, extracting the body feature data and skin color feature data of the people, and performing portrait matching based on the extracted body feature data and skin color feature data and the body feature data and skin color feature data of the matching people pre-stored in the personnel information repository;

[0019] If the match is successful, the behavior of the person entering is identified; if the match fails, an alarm is issued.

[0020] A further technical solution, the behavior recognition of the entrants refers to: monitoring the behavior of the entrants, frame-processing the behavior images of the entrants, obtaining a set of frame images, obtaining the motion behavior data of the entrants based on the obtained image set, and judging whether the behavior of the entrants is unknown behavior and whether it is abnormal behavior based on the motion behavior data.

[0021] A further technical solution is to determine whether the behavior of the person entering is unknown behavior or abnormal behavior based on the motion behavior data, specifically including:

[0022] Performing behavior matching between the acquired movement behavior data of the entering person and the normal movement behavior data and abnormal movement behavior data of the matching person pre-stored in the abnormal behavior storage library;

[0023] If the match fails, the behavior image of the person entering is stored as unknown behavior. After being marked by the administrator, the behavior information in the abnormal behavior repository is updated through the self-learning algorithm;

[0024] If the match is successful, it is determined again whether the behavior of the entrant is abnormal. If it matches the abnormal behavior data, it is determined that the behavior of the entrant is abnormal, and the information of the entrant and the abnormal behavior image are sent to the user terminal; if it matches the normal behavior data, it is determined that the behavior of the entrant is normal.

[0025] In a second aspect, the present disclosure provides a chicken house farming safety control system based on self-learning, including a face recognition access control module, a personnel abnormal behavior recognition module and a control module;

[0026] The face recognition access control module is used to collect personnel information, including personnel face images, fingerprint information and ID card information, extract characteristic data of personnel information, send the characteristic data of personnel information to the control module, and control the access control working state according to the access control command issued by the control module;

[0027] The personnel abnormal behavior recognition module is used to collect personnel behavior images, obtain personnel movement behavior data based on the behavior images, send the personnel movement behavior data to the control module, and perform tracking detection according to the personnel ID and tracking detection instructions issued by the control module;

[0028] The control module is used to receive the personnel information feature data sent by the face recognition access control module, and perform personnel identification and matching in combination with the personnel information repository; receive the motion behavior data sent by the personnel abnormal behavior recognition module, and perform abnormal behavior matching in combination with the abnormal behavior repository; and simultaneously perform self-learning and updating of personnel facial features and abnormal behavior features.

[0029] In a third aspect, the present disclosure further provides an electronic device, comprising a memory and a processor, and computer instructions stored in the memory and executed on the processor, wherein when the computer instructions are executed by the processor, the steps of the method described in the first aspect are completed.

[0030] In a fourth aspect, the present disclosure further provides a computer-readable storage medium for storing computer instructions, which, when executed by a processor, complete the steps of the method described in the first aspect.

[0031] One or more of the above technical solutions have the following beneficial effects:

[0032] 1. The present invention provides a chicken house farming safety control method and system based on self-learning. It optimizes face recognition access control through autonomous learning, identifies and records abnormal behaviors of farmers entering the chicken house, and further optimizes abnormal behavior recognition through self-learning, thereby solving the problem of low efficiency and high error rate of chicken house farming safety control caused by manual monitoring, and realizing intelligent, efficient and safe control of chicken house farming.

[0033] 2. The present invention collects facial information, fingerprint information or ID card information through face recognition access control, and allows entry after successful matching, so as to prevent outsiders from wandering into the chicken house and causing harm to the safe production of the chicken house. By identifying the abnormal behavior of the breeders, the unnecessary losses caused by the abnormal behavior of the breeders to the chicken house production are reduced, and the problem of low work efficiency and high error rate caused by the limited personnel in the safety control of the chicken house is solved. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0035] Figure 1 This is a flow chart of access control in the chicken house farming safety prevention and control method based on self-learning according to the first embodiment of the present invention;

[0036] Figure 2 This is a flow chart of behavior recognition control in the chicken house farming safety prevention and control method based on self-learning according to the first embodiment of the present invention;

[0037] Figure 3 This is a flowchart of a facial feature data self-learning algorithm in the method of Embodiment 1 of the present invention;

[0038] Figure 4 A flowchart of a motion behavior feature self-learning algorithm in the method of Embodiment 1 of the present invention;

[0039] Figure 5 This is a schematic diagram of the structure of a chicken house farming safety control system based on self-learning according to the second embodiment of the present invention;

[0040] Figure 6 This is a distribution diagram of corresponding devices of the system described in Embodiment 2 of the present invention;

[0041] Figure 7 This is a front view of the corresponding equipment of the face recognition access control module in the system described in Example 2 of the present invention. DETAILED DESCRIPTION

[0042] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0043] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0044] Embodiment 1

[0045] This embodiment provides a chicken house farming safety control method based on self-learning, such as Figure 1 and Figure 2 As shown, including:

[0046] Collect facial images of people waiting to enter and exit the chicken house and perform facial recognition and matching to determine whether the people waiting to enter and exit are wearing masks. If so, collect the fingerprint or ID card information of the people waiting to enter and exit, and control the access control working status according to the fingerprint or ID card information recognition results. Otherwise, directly control the access control working status according to the facial recognition and matching results. If the match is successful, update the facial information of the people entering and exiting through the self-learning algorithm;

[0047] The behavioral images of people entering the chicken house are collected in real time, and the behavioral data of the people entering are obtained based on the behavioral images to determine whether the behavior of the people entering is unknown behavior. If so, the behavior information is updated through the self-learning algorithm. Otherwise, it is determined whether the known behavior is abnormal behavior. If so, an alarm is issued.

[0048] In this embodiment, facial images of people who are about to enter and exit the chicken house are collected, and face recognition is performed on the collected facial images. First, the position of the face in the image is identified, and face alignment is performed after identification, that is, preset facial recognition key points are found, including the corners of the eyes, the corners of the mouth, the tip of the nose, etc., and a series of image processing transformations such as light, rotation, cutting, filtering, noise reduction, enlargement, and reduction are performed using the corresponding key points to transform the face image into a standard face image; then, the Histogram of Oriented Gradient (HOG) facial feature extraction is performed on the transformed standard face image, and face matching is performed based on the extracted facial feature data.

[0049] If a normal match is detected, face matching is performed based on the extracted facial feature data and the facial feature data of all breeding personnel pre-stored in the personnel information repository. If the match is successful, the access control is opened and the behavior of the entering person is identified. At the same time, the facial feature data of the person pre-stored in the personnel information repository is updated through the self-learning algorithm. If the match fails, the access control is still closed. At the same time, the extracted facial feature data of the person is stored as facial feature data of a suspicious person through the self-learning algorithm.

[0050] The above self-learning algorithm updates the facial feature data pre-stored in the personnel information repository, such as Figure 3 As shown, specifically, it means: adopting the maximum deviation replacement idea, each time the facial feature data is successfully matched, if the facial feature data of the person stored in the personnel information repository is less than the set number, which is set to 7 in this embodiment, then the facial feature data extracted this time is placed at the end of the queue of the facial feature data stored by the person; if the facial feature data of the person stored in the personnel information repository is sufficient to the set number, that is, 7, then the deviation values ​​of the 7 facial feature data originally stored by the person and the facial feature data extracted this time are calculated respectively according to the facial feature deviation model, and the facial feature data with the largest deviation value is deleted from the queue of the facial feature data stored by the person, and the facial feature data extracted this time is placed at the end of the queue of the facial feature data stored by the person, thereby completing the self-learning update.

[0051] The expression of the above facial feature deviation model is:

[0052]

[0053] Among them, Dj (j=1,2,3,4,5,6,7) is the deviation value of the 7 facial feature data of the person, D i is the deviation value of each feature detail, W i is the detail weight of each feature.

[0054] If it is detected that the match cannot be done normally, it is considered that the person to be entered or exited is wearing a mask. At this time, the verification method is changed, that is, the fingerprint or ID card information of the person to be entered or exited is collected, and the access control working state is controlled according to the fingerprint or ID card information recognition result. Specifically, the fingerprint or ID card information of the person to be entered or exited is collected, the fingerprint feature data or ID card feature data is extracted, and data matching is performed based on the extracted fingerprint feature data or ID card feature data and the fingerprint feature data or ID card feature data of all the breeding personnel pre-stored in the personnel information repository. If the match is successful, the access control is controlled to be opened, and the ID information and entry and exit time of the person are recorded at the same time, and the person entering is tracked and detected; if the match fails, the access control is still controlled to be closed.

[0055] The above-mentioned tracking detection refers to matching the skin color characteristics and body shape characteristics of the entrants to avoid the problem of other malicious persons stealing the fingerprint information or ID card information of the breeders to enter the breeding chicken house. Specifically, the behavior images of the entrants are collected in real time, and the images containing the entrants are obtained based on the behavior images, and the entrants in the images are identified. Specifically, the position of the entrants in the image is identified, and the portrait is aligned after identification, and the preset body shape recognition key points, including the head, elbow joints, etc., are found. The corresponding key points are used to perform a series of processing transformations such as light, rotation, cutting, filtering, noise reduction, magnification, and reduction to transform the portrait into a standard portrait; then, based on the standard portrait, the body shape feature data of the entrants is obtained according to the magnification of the standard portrait, and the skin color feature data of the entrants is obtained according to the skin color detection algorithm. Based on the extracted body shape feature data and skin color feature data, and the body shape feature data and skin color feature data of the matching person pre-stored in the personnel information repository, the portrait is matched. If the match is successful, the behavior of the entrant is identified. If the match fails, an alarm is issued, such as the warning light is on, the warning speaker is long beeping, etc.

[0056] As another real-time method, the above tracking detection can also be used to match the facial features of the entrants. Specifically, the behavioral images of the entrants are collected in real time, and the images containing the entrants are obtained based on the behavioral images, and the faces of the entrants in the images are recognized. When the entrants take off their masks, the images at this time are extracted, the position of the entrants' faces in the images is identified, and the face images are transformed into standard face images through face alignment. Then, the directional gradient histogram HOG facial features are extracted from the transformed standard face images, and face matching is performed based on the extracted facial feature data. If the match is successful, the entrants' behavior recognition is performed. If the match fails, an alarm is issued, such as the warning light is turned on, the warning speaker is beeped, etc.

[0057] The above-mentioned personnel behavior recognition refers to monitoring the behavior of people entering the chicken house to prevent them from engaging in behaviors that are harmful to chicken house farming. Specifically, the behavior images of people entering the chicken house are collected in real time, and the behavior images of people entering the chicken house are processed by frame division to obtain a set of frame-by-frame images. Based on the obtained image set, the behavior data of people entering the chicken house is obtained.

[0058] Specifically, based on the acquired image set, the background difference method is used to obtain the moving target. The video data is processed by frame division to obtain a set of frame-by-frame images, and the moving target is obtained by the background difference method. In this embodiment, a single Gaussian background modeling is used, and the modeling process is as follows:

[0059] A video is a sequence of multiple frames. For a certain pixel in the sequence, its change on the time axis is a discrete distribution state. The pixel value change process in the sequence is expressed as follows:

[0060] {X1,X2,.....,X t}={I i (x,y):1≤i≤t}

[0061] Among them, I i (x,y) represents the pixel value at position (x,y) in the i-th frame image;

[0062] According to the Gaussian modeling principle, each pixel point conforms to the Gaussian distribution, and the following background model is established:

[0063]

[0064] Among them, μ and σ represent the mean and standard deviation of the Gaussian distribution at time t, respectively, and X t is the pixel value at time t, and the mean and variance calculation formulas are as follows:

[0065]

[0066]

[0067] Where N is the length of the time window for modeling, and μ0 and Then, the background model is built for the new frame image, and the mean and variance of the model are updated. The update formula is as follows:

[0068] μ t =(1-ρ)μ t-1 +ρX t

[0069]

[0070] Wherein, ρ is the recognition rate, and the value in this embodiment is 0.01; the calculation formula for determining whether the pixel point is a background point or a foreground moving target point is as follows:

[0071]

[0072] Among them, c is a constant with a value of 2.5, D t (x,y)=0 represents background, D t (x,y)=1 indicates a foreground moving target.

[0073] Through the above scheme, the moving target is obtained. Then, the behavior set of the moving target is extracted from the frame-by-frame image, the behavior feature of each behavior is extracted from the behavior set, and the motion behavior data of the moving target is calculated according to the behavior feature. Based on the motion behavior data, it is judged whether the behavior of the entrant is an unknown behavior. Specifically, the acquired motion behavior data of the entrant is matched with the normal motion behavior data and abnormal motion behavior data of the matching person pre-stored in the abnormal behavior storage library. If the match fails, it is considered that the behavior of the entrant is an unknown behavior. At this time, the behavior image and operation behavior data of the entrant are stored. After being marked by the chicken house manager, the behavior information in the abnormal behavior storage library is updated through the self-learning algorithm.

[0074] Specifically, Figure 4 As shown, when the behavior of the entering person is judged to be unknown behavior, the behavior image and operation behavior data of the person are sent to the user terminal. The chicken house manager determines whether it is an abnormal behavior of the entering person based on the position behavior image at the user terminal. If so, based on the job type of the entering person, the acquired motion behavior data is stored in the job type corresponding to the abnormal behavior database, and the abnormal motion behavior data of the job type is updated.

[0075] If the match is successful, the behavior of the entrant is considered to be a known behavior. At this time, the behavior of the entrant is judged again based on the matching result to see if it is an abnormal behavior. If the match is successful with the abnormal behavior data, the behavior of the entrant is considered to be abnormal, and the information of the entrant and the abnormal behavior image are sent to the client terminal; if the match is successful with the normal behavior data, the behavior of the entrant is considered to be normal.

[0076] In fact, in this embodiment, the above-mentioned personnel information repository pre-stores the facial feature data, fingerprint feature data, ID card feature data, body feature data and skin color feature data of all breeding personnel, wherein the ID card feature data includes the ID and job type of different types of workers, and the pre-stored facial feature data, fingerprint feature data, body feature data and skin color feature data correspond to the ID card feature data one by one. The above-mentioned job types include chicken house managers, breeders, veterinarians and cleaners. The staff of each job type has different permissions. When the staff acts beyond their permissions, it is identified as abnormal behavior.

[0077] The abnormal behavior storage library pre-stores normal movement behavior data and abnormal movement behavior data of all breeding personnel, and the normal movement behavior data and abnormal movement behavior data correspond to the pre-stored feature data in the personnel information storage library, and different breeding personnel have different movement behavior data.

[0078] In addition, when the chickens are raised in the chicken coop and are ready for slaughter, the farmers need to catch the chickens and need to frequently enter and exit the chicken coop. Access control and abnormal behavior recognition will affect work efficiency. The mode can be set to normal mode and open mode. The chicken coop administrator can set the open mode on the user terminal to keep the access control in the open state and turn off abnormal behavior recognition.

[0079] The above scheme of this embodiment optimizes the facial recognition access control through autonomous learning, identifies and records the abnormal behavior of farmers entering the chicken house, and further optimizes the abnormal behavior recognition through self-learning, so as to solve the problem of low work efficiency and high error rate of chicken house farming safety control due to personnel restrictions, and realize intelligent, efficient and safe control of chicken house farming.

[0080] The above scheme of this embodiment collects facial information, fingerprint information or ID card information through face recognition access control, and allows entry after successful matching, so as to prevent outsiders from wandering into the chicken house and causing harm to the safe production of the chicken house. By identifying the abnormal behavior of the breeders, the unnecessary losses caused by the abnormal behavior of the breeders to the chicken house production are reduced, and the problem of low work efficiency and high error rate due to personnel restrictions in the safety control of the chicken house is solved.

[0081] Embodiment 2

[0082] This embodiment provides a chicken house farming safety control system based on self-learning, such as Figure 5 As shown, it includes a face recognition access control module, a personnel abnormal behavior recognition module and a control module.

[0083] Among them, the face recognition access control module is used to collect personnel information, including personnel facial images, fingerprint information and ID card information, extract characteristic data of personnel information, send the personnel information characteristic data to the control module, and control the access control working status according to the access control instructions issued by the control module.

[0084] The face recognition access control module includes a collection unit, a face recognition unit, a control unit and a warning unit. The collection unit is used to collect face images through a face collection device, and send the collected face images to the face recognition unit, as well as collect personnel fingerprint information through a fingerprint collection device, collect personnel ID card information through a card reader, and send the collected fingerprint information and ID card information to the control module; the face recognition unit is used to receive the face image sent by the collection unit, perform face recognition on the collected image, extract facial feature data, and send the facial feature data to the control module; the control unit is used to receive the access control opening or closing instruction sent by the control module, and change the access control working state according to the corresponding instruction; the warning unit is used to receive the warning instruction sent by the control module, and control the warning light installed on the access control to light up and the warning speaker to sound according to the warning instruction, and when the termination warning instruction sent by the control module is directly received, the warning light installed on the access control is turned off and the warning speaker stops sounding.

[0085] In this embodiment, the face recognition access control module is based on a face recognition access control device, such as Figure 7 As shown, it includes a web camera, a face image display screen, a fingerprint collector, an ID card reader, a warning light and a warning speaker.

[0086] The personnel abnormal behavior recognition module is used to collect personnel behavior images, obtain personnel movement behavior data based on the behavior images, send the personnel movement behavior data to the control module, and perform tracking detection according to the personnel ID and tracking detection instructions issued by the control.

[0087] The abnormal behavior identification module includes a video acquisition unit, an image processing unit and a data conversion unit. The video acquisition unit is used to acquire the behavior image of the person through a video acquisition device, and send the behavior image to the image processing unit; the image processing unit is used to receive the behavior image sent by the video acquisition unit, obtain a collection of frame-by-frame images based on the behavior image, perform portrait recognition on the image collection, transform the recognized portrait into a standard portrait, and send the standard portrait to the data conversion unit; the data conversion unit is used to receive the standard portrait, obtain the body shape feature data and skin color feature data of the person based on the standard portrait, and send the acquired feature data to the control module.

[0088] In this embodiment, the video acquisition unit is based on a video acquisition device, such as Figure 6 As shown, it includes multiple webcams covering the entire chicken house.

[0089] The control module is used to receive the personnel information feature data sent by the face recognition access control module, and perform personnel identification and matching in combination with the personnel information repository, receive the motion behavior data sent by the personnel abnormal behavior recognition module, and perform abnormal behavior matching in combination with the abnormal behavior repository, and at the same time perform self-learning and updating of personnel facial features and abnormal behavior features.

[0090] The control module includes a judgment unit, a self-learning unit, a communication unit, a personnel information storage library and an abnormal behavior storage library. The personnel information storage library pre-stores the facial feature data, fingerprint feature data, ID card feature data, body feature data and skin color feature data of all farmers; the abnormal behavior storage library pre-stores the normal movement behavior data and abnormal movement behavior data of all farmers.

[0091] The above-mentioned judgment unit is used to receive the personnel information feature data sent by the face recognition access control module, and match the received feature data in the personnel information storage library. If the match is successful, a door opening instruction is sent to the control unit of the face recognition access control module, and the entry and exit time of the person is recorded, and the person ID is sent to the personnel abnormal behavior identification module. If the match fails, a verification failure instruction is sent to the control unit of the face recognition access control module.

[0092] As another embodiment, the above-mentioned judgment unit is also used to receive motion behavior data sent by the personnel abnormal behavior identification module, and match the received motion behavior data in the abnormal behavior storage library. If the match is abnormal behavior, the communication unit sends the behavior subject's ID and behavior image to all user terminals.

[0093] The self-learning unit is used to update the facial feature data of the person in the personnel information storage library through the self-learning algorithm when the judgment unit successfully matches the facial feature data, and to store the unknown facial feature data as abnormal person facial feature data when the facial feature data matching fails.

[0094] As another embodiment, the self-learning unit is used to send the unknown behavior image to the user terminal through the communication module when the judgment unit matches the motion behavior data as unknown behavior. After being marked by the administrator, the behavior information in the abnormal behavior repository is updated through the self-learning algorithm.

[0095] The communication unit is used to send personnel information and behavior images to user terminals, which include mobile terminals and PC terminals.

[0096] Embodiment 3

[0097] This embodiment provides an electronic device, including a memory and a processor, and computer instructions stored in the memory and running on the processor. When the computer instructions are executed by the processor, the steps in the chicken house farming safety prevention and control method based on self-learning as described above are completed.

[0098] Embodiment 4

[0099] This embodiment also provides a computer-readable storage medium for storing computer instructions. When the computer instructions are executed by a processor, the steps in the chicken house farming safety prevention and control method based on self-learning as described above are completed.

[0100] The steps involved in the above embodiments 2 to 4 correspond to the method embodiment 1, and the specific implementation methods can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0101] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0102] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0103] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A chicken house farming safety prevention and control method based on self-learning, characterized in that: include: Collect facial images of people waiting to enter and exit the chicken house and perform facial recognition and matching to determine whether the people waiting to enter and exit are wearing masks. If so, collect the fingerprint or ID card information of the people waiting to enter and exit, and control the access control working status according to the fingerprint or ID card information recognition results. Otherwise, directly control the access control working status according to the facial recognition and matching results. If the match is successful, update the facial information of the people entering and exiting through the self-learning algorithm; Collect the behavior images of people entering the chicken house in real time, and obtain the behavior data of the people entering the chicken house based on the behavior images to determine whether the behavior of the people entering is unknown. If so, update the behavior information through the self-learning algorithm. Otherwise, determine whether the known behavior is abnormal. If so, issue an alarm. Performing face recognition based on the collected face images of the people waiting to enter and exit the chicken house, extracting the facial feature data of the people, and performing face matching based on the extracted facial feature data; If a normal match is detected, face matching is performed based on the extracted facial feature data and the facial feature data of all breeding personnel pre-stored in the personnel information repository; If the match is successful, the door will be controlled to open and the behavior of the person entering will be identified. At the same time, the facial feature data of the person pre-stored in the personnel information repository will be updated through the self-learning algorithm; The behavior recognition of the entering person refers to: monitoring the behavior of the entering person, performing frame processing on the behavior image of the entering person, obtaining a set of frame-by-frame images, obtaining the motion behavior data of the entering person based on the obtained image set, and judging whether the behavior of the entering person is an unknown behavior or an abnormal behavior based on the motion behavior data; The method of acquiring the moving target based on the acquired image set using the background difference method is as follows: A video is a sequence of multiple frames of images. The pixel value change process in the sequence is expressed as: {X1,X2,……,X t }={I i (x,y):1≤i≤t} Among them, I i (x,y) represents the pixel value at position (x,y) in the i-th frame image; According to the Gaussian modeling principle, each pixel point conforms to the Gaussian distribution, and the following background model is established: Among them, μ and σ represent the mean and standard deviation of the Gaussian distribution at time t, respectively, and X t is the pixel value at time t; Perform background modeling on the new frame image and update the mean and variance of the model. The update formula is as follows: m t =(1-ρ)μ t-1 +ρX t Among them, ρ is the recognition rate; the calculation formula for judging whether the pixel is a background point or a foreground moving target point is as follows: Where c is a constant, D t (x,y)=0 represents background, D t (x,y) = 1 indicates a foreground moving target; The updating of the facial feature data of the person pre-stored in the person information repository by the self-learning algorithm specifically refers to: If the number of facial feature data of the person stored in the personnel information repository is less than the set number, the facial feature data extracted this time is placed at the end of the queue of facial feature data stored for the person; If the number of facial feature data of the person stored in the personnel information repository is sufficient to be set, the deviation values ​​of the facial feature data originally stored for the person and the facial feature data extracted this time are calculated respectively according to the facial feature deviation model, and the facial feature data with the largest deviation value is deleted. At the same time, the facial feature data extracted this time is put into the end of the queue of the facial feature data stored for the person, so as to complete the self-learning update; The expression of the facial feature deviation model is: Among them, D j (j=1,2,3,4,5,6,7) is the deviation value of the 7 facial feature data of the person, D i is the deviation value of each feature detail, W i is the detail weight of each feature.

2. The chicken house farming safety control method based on self-learning as claimed in claim 1 is characterized in that: If the match fails, the access control will still be closed, and the facial feature data of the person extracted will be stored as facial feature data of a suspicious person through a self-learning algorithm.

3. The chicken house farming safety control method based on self-learning as claimed in claim 1 is characterized in that: If it is detected that a normal match cannot be made, the fingerprint or ID card information of the person to be entered or exited is collected, the fingerprint feature data or ID card feature data is extracted, and data matching is performed based on the extracted fingerprint feature data or ID card feature data and the fingerprint feature data or ID card feature data of all the breeding personnel pre-stored in the personnel information repository; If the match is successful, the access control will be opened, and the person’s ID information and entry and exit time will be recorded, and the person entering will be tracked and detected.

4. The chicken house farming safety control method based on self-learning as claimed in claim 3 is characterized in that: The tracking detection refers to matching the skin color and body shape characteristics of the person entering. Specifically: Collecting behavioral images of people entering the chicken house in real time, obtaining images containing the people entering, identifying the people entering the image, extracting the body feature data and skin color feature data of the people, and performing portrait matching based on the extracted body feature data and skin color feature data and the body feature data and skin color feature data of the matching people pre-stored in the personnel information repository; If the match is successful, the behavior of the person entering is identified; if the match fails, an alarm is issued.

5. The chicken house farming safety control method based on self-learning as claimed in claim 1 is characterized in that: Based on the motion behavior data, determine whether the behavior of the person entering is unknown behavior or abnormal behavior, specifically including: Performing behavior matching between the acquired movement behavior data of the entering person and the normal movement behavior data and abnormal movement behavior data of the matching person pre-stored in the abnormal behavior storage library; If the match fails, the behavior image of the person entering is stored as unknown behavior. After being marked by the administrator, the behavior information in the abnormal behavior repository is updated through the self-learning algorithm; If the match is successful, it is determined again whether the behavior of the entrant is abnormal. If it matches the abnormal behavior data, it is determined that the behavior of the entrant is abnormal, and the information of the entrant and the abnormal behavior image are sent to the user terminal; if it matches the normal behavior data, it is determined that the behavior of the entrant is normal.

6. A chicken house breeding safety control system based on self-learning, characterized in that: It includes face recognition access control module, personnel abnormal behavior recognition module and control module; The face recognition access control module is used to collect personnel information, including personnel face images, fingerprint information and ID card information, extract characteristic data of personnel information, send the characteristic data of personnel information to the control module, and control the access control working state according to the access control command issued by the control module; The personnel abnormal behavior recognition module is used to collect personnel behavior images, obtain personnel movement behavior data based on the behavior images, send the personnel movement behavior data to the control module, and perform tracking detection according to the personnel ID and tracking detection instructions issued by the control module; The control module is used to receive the personnel information feature data sent by the face recognition access control module, and perform personnel identification and matching in combination with the personnel information repository; receive the motion behavior data sent by the personnel abnormal behavior recognition module, and perform abnormal behavior matching in combination with the abnormal behavior repository; and simultaneously perform self-learning and updating of personnel facial features and abnormal behavior features; Performing face recognition based on the collected face images of the people waiting to enter and exit the chicken house, extracting the facial feature data of the people, and performing face matching based on the extracted facial feature data; If a normal match is detected, face matching is performed based on the extracted facial feature data and the facial feature data of all breeding personnel pre-stored in the personnel information repository; If the match is successful, the door will be controlled to open and the behavior of the person entering will be identified. At the same time, the facial feature data of the person pre-stored in the personnel information repository will be updated through the self-learning algorithm; The behavior recognition of the entering person refers to: monitoring the behavior of the entering person, performing frame processing on the behavior image of the entering person, obtaining a set of frame-by-frame images, obtaining the motion behavior data of the entering person based on the obtained image set, and judging whether the behavior of the entering person is an unknown behavior or an abnormal behavior based on the motion behavior data; The method of acquiring the moving target based on the acquired image set using the background difference method is as follows: A video is a sequence of multiple frames of images. The pixel value change process in the sequence is expressed as: {X1,X2,……,X t }={I i (x,y):1≤i≤t} Among them, I i (x,y) represents the pixel value at position (x,y) in the i-th frame image; According to the Gaussian modeling principle, each pixel point conforms to the Gaussian distribution, and the following background model is established: Among them, μ and σ represent the mean and standard deviation of the Gaussian distribution at time t, respectively, and X t is the pixel value at time t; Perform background modeling on the new frame image and update the mean and variance of the model. The update formula is as follows: m t =(1-ρ)μ t-1 +ρX t Among them, ρ is the recognition rate; the calculation formula for judging whether the pixel is a background point or a foreground moving target point is as follows: Where c is a constant, D t (x,y)=0 represents background, D t (x,y) = 1 indicates a foreground moving target; The updating of the facial feature data of the person pre-stored in the person information repository by the self-learning algorithm specifically refers to: If the number of facial feature data of the person stored in the personnel information repository is less than the set number, the facial feature data extracted this time is placed at the end of the queue of facial feature data stored for the person; If the number of facial feature data of the person stored in the personnel information repository is sufficient to be set, the deviation values ​​of the facial feature data originally stored for the person and the facial feature data extracted this time are calculated respectively according to the facial feature deviation model, and the facial feature data with the largest deviation value is deleted. At the same time, the facial feature data extracted this time is put into the end of the queue of the facial feature data stored for the person, so as to complete the self-learning update; The expression of the facial feature deviation model is: Among them, D j (j=1,2,3,4,5,6,7) is the deviation value of the 7 facial feature data of the person, D i is the deviation value of each feature detail, W i is the detail weight of each feature.

7. An electronic device, characterized in that: It includes a memory and a processor, and computer instructions stored in the memory and running on the processor. When the computer instructions are run by the processor, the steps of a chicken house farming safety prevention and control method based on self-learning as described in any one of claims 1-5 are completed.

8. A computer-readable storage medium, characterized in that: Used to store computer instructions, which, when executed by a processor, complete the steps of a chicken house farming safety prevention and control method based on self-learning as described in any one of claims 1-5.