Chicken ingestion behavior identification method, system and equipment for flat-fed chickens and storage medium

Abstract

The invention discloses a chicken ingestion behavior identification method and system for flat-fed chickens, terminal equipment and a computer readable storage medium. The method comprises the following steps: acquiring a chicken image in an ingestion area in real time and carrying out preprocessing operation; inputting the preprocessed chicken image into a foraging area chicken identification model, and outputting target frame information of the chicken; calculating the number of chickens according to the target frame information; if the number of the chickens in the same chicken image exceeds one, the chickens in the corresponding ingestion area are expelled to leave the ingestion area so as to ensure that only one chicken exists in the ingestion area; under the condition that only one chicken exists in the ingestion area, a video of the chickens in the ingestion trough area is obtained in real time, and a single-frame chicken image is processed; performing frame difference method image processing on the processed continuous multi-frame image, and if no change exists, indicating that the chicken does not peck; and if the non-pecking time exceeds the preset time, expelling the chickens. According to the invention, the number of chickens and behaviors of the chickens can be monitored in real time, and the accuracy and efficiency of automatic measurement of the intelligent tester for the poultry feed conversion rate are effectively improved.

Description

Technical Field

[0001] The present invention relates to the technical field of intelligent poultry breeding of livestock and poultry, and in particular to a chicken feeding behavior recognition method, system, terminal equipment and computer-readable storage medium for flat-raising chickens. Background Art

[0002] As consumers' requirements for the quality and safety of poultry products continue to increase, the breeding industry faces the dual pressure of improving breeding efficiency and product quality. Farmers need to adopt efficient feeding strategies to ensure the high quality of poultry growth while reducing feeding costs. One way to reduce breeding costs is to improve feed conversion rate. In terms of poultry feed conversion rate intelligent measuring instruments, by monitoring the feeding behavior of chickens, their feed conversion rate and subsequent improvement directions can be analyzed. With the rapid development of science and technology in my country, breeding farms and entrepreneurs have increasing requirements for the accuracy and intelligence of poultry feed conversion rate intelligent measuring instruments. How to reduce the burden of monitoring chicken feeding behavior and improve its recognition accuracy has become a crucial issue. Summary of the invention

[0003] In order to solve the deficiencies of the above-mentioned prior art, the present invention provides a chicken feeding behavior recognition method, system, terminal device and computer-readable storage medium for flat-raised chickens, which can monitor the number of chickens and the behavior of chickens in real time, effectively improve the accuracy of detection, and at the same time improve the accuracy and efficiency of automatic measurement of poultry feed conversion rate intelligent measuring instrument.

[0004] The first object of the present invention is to provide a method for identifying the feeding behavior of chickens in flat-raised chickens.

[0005] The second object of the present invention is to provide a chicken feeding behavior recognition system for flat-raised chickens.

[0006] The third object of the present invention is to provide a terminal device.

[0007] A fourth object of the present invention is to provide a computer-readable storage medium.

[0008] The first object of the present invention can be achieved by adopting the following technical solutions: A method for identifying chicken feeding behavior in flat-raising chickens, the method comprising: Acquire images of chickens in the feeding area in real time and perform pre-processing operations; The preprocessed chicken image is input into the chicken recognition model in the feeding area, and the target frame information of the chicken is output; Calculate the number of chickens based on the target frame information; if there is more than one chicken in the same chicken image, expel the chicken in the corresponding feeding area to ensure that there is only one chicken in the feeding area; When there is only one chicken in the feeding area, the video of the chicken in the feeding trough area is obtained in real time, and a single frame of the chicken image is processed; The processed continuous multi-frame images are processed by frame difference method. If there is no change, it means that the chickens are not pecking. If the chicken does not peck for longer than the preset time, it will be expelled.

[0009] Furthermore, the pre-processed chicken image is input into the chicken recognition model in the feeding area to output the target frame information of the chicken, including: The preprocessed chicken images are input into a chicken recognition model in the feeding area to extract feature maps; The feature map is processed by the detection head, and the detection head uses anchor points to predict the position of the chicken; A non-maximum suppression method was used for each predicted bird to ensure that each bird was detected only once; Output the target frame information of the chicken.

[0010] Furthermore, the chicken identification model in the feeding area is obtained by optimizing and compressing the trained improved YOLOvX target detection model using ONNX technology, and redesigning the algorithm for ONNX Runtime combined with post-processing steps; The improved YOLOvX target detection model includes adding an EMA attention mechanism, replacing the backbone network with HGNetV2, and adopting a WIoU loss function; the value of X is any integer greater than 4 and less than 11.

[0011] Furthermore, before the preprocessed chicken images are input into the chicken recognition model in the feeding area, the improved YOLOvX target detection model is trained, including: Acquire and pre-process images of chickens in the feeding area, and annotate the pre-processed images of chickens; The improved YOLOvX target detection model is trained using the labeled chicken images to obtain a trained improved YOLOvX target detection model.

[0012] Furthermore, during the training process, the improved YOLOvX target detection model is used to learn the labeled chicken image features, which can accurately distinguish the chickens from the background.

[0013] Furthermore, after obtaining the image of the chickens in the feeding area in real time, the chickens in the feeding area and their target frame information are displayed in real time; After obtaining the video of the chickens in the feeding trough area in real time, the text corresponding to the identified feeding behavior of the chickens in the feeding trough area is displayed in real time.

[0014] Furthermore, the preprocessing operation includes cropping the chicken image to a fixed size, and performing adaptive histogram equalization and high-pass filtering on the cropped chicken image.

[0015] Furthermore, the processing operation includes cropping a single frame chicken image into a fixed size, and performing Gaussian filtering, grayscale binarization and Otsu threshold segmentation on the cropped chicken image.

[0016] The second object of the present invention can be achieved by adopting the following technical solutions: An intelligent monitoring system for determining poultry feed conversion rate for flat chicken farming, the system comprising: The first acquisition module is used to acquire images of chickens in the feeding area in real time and perform preprocessing operations; An output module is used to input the preprocessed chicken image into the chicken recognition model in the feeding area and output the target frame information of the chicken; A calculation module is used to calculate the number of chickens according to the target frame information; if there are more than one chicken in the same chicken image, the chicken in the corresponding feeding area is driven out of the feeding area to ensure that there is only one chicken in the feeding area; The second acquisition module is used to acquire the video of the chickens in the feeding trough area in real time when there is only one chicken in the feeding area, and to process the single-frame chicken image; The recognition module is used to perform frame difference image processing on the processed continuous multiple frames of images. If there is no change, it means that the chickens are not pecking; if the non-pecking time exceeds a preset time, the chickens are expelled.

[0017] The third object of the present invention can be achieved by adopting the following technical solutions: A terminal device includes a processor and a memory for storing a program executable by the processor. When the processor executes the program stored in the memory, the above-mentioned chicken feeding behavior recognition method for flat-raising chickens is implemented.

[0018] The fourth object of the present invention can be achieved by adopting the following technical solutions: A computer-readable storage medium stores a program, and when the program is executed by a processor, the above-mentioned chicken feeding behavior recognition method for flat-raising chickens is implemented.

[0019] The present invention has the following beneficial effects compared with the prior art: Compared with traditional manual methods, the present invention can automatically identify and count the number of chickens in the feeding area, and automatically detect the feeding behavior of chickens, which significantly reduces labor intensity and improves the accuracy of data acquisition and the speed of data processing; through optimized image processing and data analysis, the reliability and accuracy of the analysis results are ensured, thereby helping breeders to better understand the feeding status of chickens, thereby improving livestock and poultry farming practices and increasing production efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0020] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on the structures shown in these drawings without paying creative work.

[0021] Figure 1 This is a flow chart of a method for identifying chicken feeding behavior in flat-raised chickens according to Embodiment 1 of the present invention; Figure 2 This is a display diagram of the chicken identification result in the feeding area of ​​Example 1 of the present invention; Figure 3 This is a structural block diagram of a chicken feeding behavior recognition system for flat-raising chickens according to Embodiment 2 of the present invention; Figure 4 This is a structural block diagram of the terminal device of embodiment 3 of the present invention. DETAILED DESCRIPTION

[0022] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments in the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention. It should be understood that the specific embodiments described are only used to explain the present application and are not used to limit the present application.

[0023] Embodiment 1: The chicken feeding behavior identification method for flat-raised chickens provided in this embodiment first identifies and counts the chickens in the feeding area. If there is more than one chicken, the chickens are driven away to ensure that there is only one chicken in the feeding area. When there is only one chicken in the feeding area, the chicken's feeding behavior is automatically detected. If the chicken is not in a state of feeding, the chicken is driven away, thereby ensuring the accuracy of the data of the feed conversion rate meter.

[0024] It should be noted that the feeding area in this embodiment is the feeding area in the feed conversion rate machine. In order to clearly calculate the feed conversion rate of each chicken, it is required that only one chicken is in the feeding area during feeding.

[0025] like Figure 1 As shown, this embodiment provides a method for identifying the feeding behavior of chickens in flat-raised chickens. Compared with the traditional weight measurement identification method, it can effectively improve the accuracy and efficiency of identifying chickens in the feeding area; compared with the traditional chicken feeding behavior detection method, it can effectively improve the accuracy and efficiency of chicken feeding behavior detection and automatically identify its feeding behavior. The method includes the following steps: S101, acquiring images of chickens in a feeding area in real time and performing preprocessing operations.

[0026] Connect the USB near-infrared / visible light camera to the industrial computer via USB and install it just above the feeding area. Use the camera to collect images of chickens in the feeding area, obtain real-time monitoring video through the image recording function, and then save the acquired monitoring video in frames. Import the obtained chicken images into the computer system for preprocessing.

[0027] Specifically, the preprocessing operation includes cropping the chicken image to a fixed size with a uniform resolution of 1280*720 in pixels; ensuring the consistency and efficiency of subsequent processing; and performing adaptive histogram equalization and high-pass filtering on the cropped chicken image.

[0028] S102, inputting the preprocessed chicken image into a chicken recognition model in the feeding area, and outputting target frame information of the chicken.

[0029] The chicken identification model in the feeding area is obtained by training the improved YOLOvX (X=5 / 6 / 7 / 8 / 9 / 10) target detection model. The improvements include adding the EMA attention mechanism, replacing the backbone network with HGNetV2, and adopting the WIoU loss function.

[0030] Before step S102, the improved YOLOvX target detection model is also trained, specifically including: (1) Obtain images of chickens in the feeding area and preprocess them, then label the preprocessed images of chickens.

[0031] The pre-processed images are annotated with chickens. The chickens are marked with boxes of the same color to facilitate visual identification and subsequent analysis.

[0032] The labeled chicken images are divided into training and test sets in a ratio of 9:1. The training set data is used to further train and optimize the YOLOvX (X=5 / 6 / 7 / 8 / 9 / 10) object detection model and ensure that the model can achieve high-precision chicken recognition on the unseen test set.

[0033] (2) Use the labeled chicken images to train the improved YOLOvX object detection model.

[0034] The improved YOLOvX object detection model is trained using the annotated chicken images in the training set.

[0035] Specifically, the number of training rounds is 300, the batch size of training is 16, the initial learning rate is 1×10-4, the decay is 5×10-4, and the SGD optimizer is used.

[0036] By adopting the improved YOLOvX target detection model, the accuracy and efficiency of identifying chickens in the feeding area can be improved. During the training process, the improved YOLOvX target detection model is used to learn the features of the annotated chicken images, which can accurately distinguish the chickens from the background.

[0037] After the training is completed, the trained improved YOLOvX target detection model is optimized and compressed for deployment using ONNX technology, and the algorithm using the model is redesigned to obtain a chicken recognition model in the feeding area. The specific algorithm includes efficient reasoning through ONNXRuntime to improve the reasoning speed, and combined with post-processing steps such as non-maximum suppression (NMS) to accurately identify chickens in the feeding area.

[0038] Using the chicken identification model in the feeding area can lower the equipment and computing power threshold for using this method.

[0039] Further, step S102 includes: (1) The preprocessed chicken images are input into the feeding area chicken recognition model, which extracts image features through a series of convolutional layers; (2) The feature map is processed by a specific detection head, which uses anchor points to predict the position of the chicken; (3) Use the non-maximum suppression (NMS) method for each predicted chicken to reduce overlapping detection results and ensure that each chicken is detected only once; (4) The target box information of the chicken is output, and the category is chicken (broiler).

[0040] The output target frame information is used for statistical analysis, including location and quantity, subsequent storage, and serial port information transmission.

[0041] This embodiment uses the chicken recognition model in the feeding area to detect the chicken images in each feeding area and outputs the target frame information of all the chickens.

[0042] S103, calculating the number of chickens according to the target frame information; if the number of chickens in the same chicken image exceeds one, the chickens in the corresponding feeding area are driven out of the feeding area to ensure that there is only one chicken in the feeding area.

[0043] If there is more than one target frame in a chicken image, it means that there are more than one chicken in the same feeding area, and then the serial port information is sent to the central processor to start the revolving door mechanism to drive the chickens away.

[0044] Specifically, the revolving door mechanism forms a three-door structure with an angle of 120 degrees, which divides the measuring instrument into two areas: feeding and waiting. The revolving door is driven by a worm gear reduction motor to achieve partitioned self-locking of the revolving door.

[0045] In order to support the real-time display of images recognized by the camera and the display of real-time results, this embodiment also provides a display interface developed based on pyqt5. The interface design focuses on user experience to ensure that all functions are easy to access and use. Specifically, it includes the following main components and corresponding functions: The data input and function operation module is used to receive the video results after recognition from the USB near-infrared / visible light camera. The administrator can select the required serial port as the data transmission channel through the interface.

[0046] The recognition result display module is used to display the results of chicken recognition in the feeding area. Each identified chicken is displayed in a rectangular frame of the same color, and the corresponding chicken text category is displayed above, so that users can quickly identify and count the chickens in the feeding area. The recognition results can be referred to Figure 2 .

[0047] The display module is used to display the number of detected and identified chickens in real time.

[0048] S104: When there is only one chicken in the feeding area, a video of the chicken in the feeding trough area is obtained in real time, and a single-frame chicken image is processed.

[0049] In one embodiment, the K210 / openMV development board is connected to an industrial computer via USB and installed on the left side of the feeding trough in the feeding area. The camera faces the feeding trough area and is connected to the central processor via a DuPont line. Visible light images of the feeding behavior of chickens in the feeding area are acquired in real time, and the acquired real-time monitoring images are imported into the computer system for further preprocessing.

[0050] Specifically, the processing operation includes cropping the image, and the resolution is unified to 214*214 (in pixels) to ensure the consistency and efficiency of subsequent processing. The subsequent processing includes Gaussian filtering, grayscale binarization, and Otsu threshold segmentation.

[0051] S105, performing frame difference image processing on the processed continuous multiple frame images, if there is no change, it indicates that the chickens are not pecking.

[0052] Detect the movement changes of chickens eating in the image. By comparing the differences between three consecutive frames of images, determine whether the chickens are eating to improve the accuracy and reliability of detection. If there is no difference between the three consecutive frames of images, it means that the chickens are not pecking, otherwise it means that the chickens are pecking.

[0053] S106, if the non-pecking time exceeds a preset time, the chickens are driven away.

[0054] If the pecking time exceeds the preset time, the serial port information is sent to the central controller to control the revolving door mechanism to automatically drive the chickens away.

[0055] In order to display the image processed by the frame difference method on the development board display screen, the inference results of eating and not eating are displayed in text in the upper left corner of the display screen, supporting real-time display of processed images and results.

[0056] The chicken feeding behavior detection provided by the present invention discriminates the chicken feeding behavior through the results of the frame difference method; the chicken identification in the feeding area can automatically identify and count the chickens in the feeding area by adopting advanced deep learning technology. Compared with the traditional manual method, it reduces the dependence on manpower, improves the degree of automation of data acquisition and data processing, and ensures the reliability and repeatability of the results.

[0057] Those skilled in the art will appreciate that all or part of the steps in the method for implementing the above embodiments may be completed by instructing related hardware through a program, and the corresponding program may be stored in a computer-readable storage medium.

[0058] It should be noted that although the method operations of the above embodiments are described in a specific order in the accompanying drawings, this does not require or imply that these operations must be performed in this specific order, or that all the operations shown must be performed to achieve the desired results. On the contrary, the steps depicted can be performed in a different order. Additionally or alternatively, some steps can be omitted, multiple steps can be combined into one step, and / or one step can be decomposed into multiple steps.

[0059] Embodiment 2: like Figure 3As shown, this embodiment provides an intelligent monitoring system for determining poultry feed conversion rate for flat chicken farming, the system comprising a first acquisition module 301, an output module 302, a calculation module 303, a second acquisition module 304 and an identification module 305, wherein: The first acquisition module 301 is used to acquire images of chickens in the feeding area in real time and perform preprocessing operations; Output module 302, used for inputting the pre-processed chicken image into the chicken recognition model in the feeding area, and outputting the target frame information of the chicken; The calculation module 303 is used to calculate the number of chickens according to the target frame information; if the number of chickens in the same chicken image exceeds one, the chickens in the corresponding feeding area are driven out of the feeding area to ensure that there is only one chicken in the feeding area; The second acquisition module 304 is used to acquire the video of the chicken in the feeding trough area in real time when there is only one chicken in the feeding area, and to process the single-frame chicken image; The identification module 305 is used to perform frame difference image processing on the processed continuous multiple frames of images. If there is no change, it means that the chickens are not pecking; if the non-pecking time exceeds a preset time, the chickens are expelled.

[0060] The specific implementation of each module in this embodiment can refer to the above-mentioned embodiment 1, which will not be described one by one here; it should be noted that the system provided in this embodiment is only illustrated by the division of the above-mentioned functional modules. In actual applications, the above-mentioned functions can be assigned to different functional modules as needed, that is, the internal structure can be divided into different functional modules to complete all or part of the functions described above.

[0061] Embodiment 3: This embodiment provides a terminal device, which can be a computer. Figure 4As shown, a processor 402, a memory, an input device 403, a display 404 and a network interface 405 connected through a system bus 401, the processor is used to provide computing and control capabilities, the memory includes a non-volatile storage medium 406 and an internal memory 407, the non-volatile storage medium 406 stores an operating system, a computer program and a database, the internal memory 407 provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium, when the processor 402 executes the computer program stored in the memory, the chicken feeding behavior recognition method for flat chickens in the above-mentioned embodiment 1 is implemented, that is: real-time acquisition of the feeding area The method comprises the following steps: taking a chicken image and performing a preprocessing operation on the chicken image; inputting the preprocessed chicken image into a chicken recognition model in the feeding area, and outputting the target frame information of the chicken; calculating the number of chickens according to the target frame information; if the number of chickens in the same chicken image exceeds one, the chicken in the corresponding feeding area is driven out of the feeding area to ensure that there is only one chicken in the feeding area; when there is only one chicken in the feeding area, obtaining a chicken video in the feeding trough area in real time, and performing a processing operation on a single-frame chicken image; performing frame difference method image processing on the processed continuous multi-frame images, and if there is no change, it indicates that the chicken has not pecked at the food; if the non-pecking time exceeds a preset time, the chicken is driven out.

[0062] Embodiment 4: The present embodiment provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the chicken feeding behavior recognition method for flat-raised chickens of the above-mentioned embodiment 1, namely: obtaining images of chickens in a feeding area in real time and performing preprocessing operations; inputting the preprocessed chicken images into a chicken recognition model in the feeding area, and outputting target frame information of the chickens; calculating the number of chickens according to the target frame information; if the number of chickens in the same chicken image exceeds one, the chickens in the corresponding feeding area are expelled from the feeding area to ensure that there is only one chicken in the feeding area; when there is only one chicken in the feeding area, the chicken video in the feeding trough area is obtained in real time, and a single-frame chicken image is processed; frame difference method image processing is performed on the processed continuous multiple-frame images, and if there is no change, it indicates that the chicken has not pecked; if the non-pecking time exceeds a preset time, the chicken is expelled.

[0063] It should be noted that the computer-readable storage medium of the present embodiment may be a computer-readable signal medium or a computer-readable storage medium or any combination of the above two. The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, device or device, or any combination of the above. More specific examples of computer-readable storage media may include, but are not limited to: an electrical connection with one or more wires, a portable computer disk, a hard disk, a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disk read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the above.

[0064] The above is only a preferred embodiment of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical solution and inventive concept of the present invention within the scope disclosed by the present invention, which shall fall within the protection scope of the present invention.

Claims

1. A method for identifying chicken feeding behavior in flat-raised chickens, characterized in that: The method comprises: Acquire images of chickens in the feeding area in real time and perform pre-processing operations; The preprocessed chicken image is input into the chicken recognition model in the feeding area, and the target frame information of the chicken is output; Calculate the number of chickens based on the target frame information; if there is more than one chicken in the same chicken image, expel the chicken in the corresponding feeding area to ensure that there is only one chicken in the feeding area; When there is only one chicken in the feeding area, the video of the chicken in the feeding trough area is obtained in real time, and a single frame of the chicken image is processed; The processed continuous multi-frame images are processed by frame difference method. If there is no change, it means that the chickens are not pecking. If the chicken does not peck for longer than the preset time, it will be expelled.

2. The chicken feeding behavior identification method according to claim 1, characterized in that: The pre-processed chicken image is input into the chicken recognition model in the feeding area, and the target frame information of the chicken is output, including: The preprocessed chicken images are input into a chicken recognition model in the feeding area to extract feature maps; The feature map is processed by the detection head, and the detection head uses anchor points to predict the position of the chicken; A non-maximum suppression method was used for each predicted bird to ensure that each bird was detected only once; Output the target frame information of the chicken.

3. The chicken feeding behavior identification method according to any one of claims 1 and 2, characterized in that: The chicken identification model in the feeding area is obtained by optimizing and compressing the trained improved YOLOvX target detection model using ONNX technology, and redesigning the algorithm for ONNX Runtime combined with post-processing steps; The improved YOLOvX target detection model includes adding an EMA attention mechanism, replacing the backbone network with HGNetV2, and adopting a WIoU loss function; the value of X is any integer greater than 4 and less than 11.

4. The chicken feeding behavior identification method according to claim 3, characterized in that: Before the pre-processed chicken images are input into the feeding area chicken recognition model, the improved YOLOvX target detection model is trained, including: Acquire and pre-process images of chickens in the feeding area, and annotate the pre-processed images of chickens; The improved YOLOvX target detection model is trained using the labeled chicken images to obtain a trained improved YOLOvX target detection model.

5. The chicken feeding behavior identification method according to claim 4, characterized in that: During the training process, the improved YOLOvX target detection model is used to learn the labeled chicken image features, which can accurately distinguish the chickens from the background.

6. The chicken feeding behavior identification method according to claim 1, characterized in that: After obtaining the image of the chickens in the feeding area in real time, the information of the chickens in the feeding area and their target frames are displayed in real time; After obtaining the video of the chickens in the feeding trough area in real time, the text corresponding to the identified feeding behavior of the chickens in the feeding trough area is displayed in real time.

7. The chicken feeding behavior identification method according to any one of claims 1, 2, 4 to 6, characterized in that: The preprocessing operation includes cutting the chicken image into a fixed size, and performing adaptive histogram equalization and high-pass filtering on the cut chicken image.

8. The chicken feeding behavior identification method according to any one of claims 1, 2, 4 to 6, characterized in that: The processing operation includes cutting a single frame chicken image into a fixed size, and performing Gaussian filtering, grayscale binarization and Otsu threshold segmentation on the cut chicken image.

9. An intelligent monitoring system for determining poultry feed conversion rate for flat chicken farming, characterized in that: The system comprises: The first acquisition module is used to acquire images of chickens in the feeding area in real time and perform preprocessing operations; An output module is used to input the preprocessed chicken image into the chicken recognition model in the feeding area and output the target frame information of the chicken; A calculation module is used to calculate the number of chickens according to the target frame information; if there are more than one chicken in the same chicken image, the chicken in the corresponding feeding area is driven out of the feeding area to ensure that there is only one chicken in the feeding area; The second acquisition module is used to acquire the video of the chickens in the feeding trough area in real time when there is only one chicken in the feeding area, and to process the single-frame chicken image; The recognition module is used to perform frame difference image processing on the processed continuous multiple frames of images. If there is no change, it means that the chickens are not pecking; if the non-pecking time exceeds a preset time, the chickens are expelled.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the chicken feeding behavior identification method according to any one of claims 1 to 8 is implemented.

Images