CN-122023237-A - Chicken number detection method, device, equipment, medium and program product

CN122023237ACN 122023237 ACN122023237 ACN 122023237ACN-122023237-A

Abstract

The application discloses a method, a device, equipment, a medium and a program product for detecting the quantity of chickens, belonging to the technical field of poultry management; and inputting the chicken shooting images into a chicken quantity detection model to obtain chicken quantity detection results output by the chicken quantity detection model, wherein the chicken quantity detection model is constructed based on a YOLO model and combines a reversible column backbone network and an attention mechanism. According to the application, the obtained chicken shooting images of the target area are input into the pre-trained chicken quantity detection model, the characteristics in the chicken images are extracted by utilizing the chicken quantity detection model, and the chicken quantity detection result is obtained by automatic identification and detection, so that the chicken quantity can be automatically, efficiently and accurately monitored, the labor cost and the inventory pressure are obviously reduced, and the cultivation efficiency and the management level are improved.

Inventors

FANG CHENG
WU ZHENLONG
MA CHUANG
YANG JIKANG
Chao Xiaohuan
LUO QINGBIN
ZHANG XIQUAN
ZHANG TIEMIN

Assignees

华南农业大学

Dates

Publication Date: 20260512
Application Date: 20251212

Claims (10)

1. A method for detecting the number of chickens, which is characterized by comprising the following steps: Acquiring a chicken shooting image of a target area; Inputting the chicken shooting image into a chicken quantity detection model to obtain a chicken quantity detection result output by the chicken quantity detection model; The chicken quantity detection model is constructed based on a YOLO model and combines a reversible column backbone network with an attention mechanism.
2. The method of claim 1, wherein inputting the chicken shot image into a chicken quantity detection model to obtain a chicken quantity detection result output by the chicken quantity detection model comprises: inputting the chicken shooting image into a chicken quantity detection model; And detecting and determining the range of the chicken coops through the chicken number detection model, detecting and determining the number of chicken in the chicken coops, and outputting the chicken number as a chicken number detection result.
3. The method according to claim 2, wherein the detecting and determining the range of the chicken coops by the chicken number detecting model, and further detecting and determining the number of chicken in the chicken coops, and outputting the result as the chicken number detecting result, comprises: Identifying the boundary frame of the chicken coop fence in the chicken shooting image through the chicken quantity detection model, extracting the minimum abscissa and the maximum abscissa of the boundary frame of the chicken coop fence, and defining a chicken coop range; Detecting and identifying the number and the state of the chickens in the range of the chicken coop, and outputting a chicken number detection result.
4. The method according to claim 1, wherein the method further comprises: according to the number information of the chicken coops, determining a chicken quantity detection result of multi-frame chicken shooting images of the chicken coops; Under the condition that the abnormal frames exist in the detection result of the number of the multi-frame chicken shooting images of the chicken coops, determining the chicken shooting images of the abnormal frames and corresponding chicken coops numbering information, and generating abnormal warning information; Under the condition that abnormal frames do not exist in the chicken quantity detection results of the multi-frame chicken shooting images of the chicken coops, determining the chicken quantity detection result with the highest numerical value in the chicken quantity detection results of all the multi-frame chicken shooting images, and uploading the chicken quantity detection results as the chicken quantity detection results of the chicken coops.
5. The method of claim 1, further comprising, after the capturing of the image of the chicken of the target area: performing image enhancement on the chicken shooting image, and cutting the chicken shooting image to a preset specification; wherein the image enhancement includes adaptive histogram equalization and high pass filtering.
6. The method of claim 1, wherein the chicken quantity detection model is trained based on the steps of: Acquiring a plurality of historical chicken shooting images, and acquiring chicken quantity detection results corresponding to the historical chicken shooting images; taking the images shot by each historical chicken as a sample, taking the detection results of the number of chicken corresponding to the images shot by each historical chicken as sample tags corresponding to the samples, and constructing a training data set; and pre-training the chicken quantity detection model by using the training data set.
7. A chicken quantity detection device, the device comprising: The image acquisition module is used for acquiring a chicken shooting image of the target area; the chicken detection module is used for inputting the chicken shooting image into a chicken quantity detection model so as to obtain a chicken quantity detection result output by the chicken quantity detection model; The chicken quantity detection model is constructed based on a YOLO model and combines a reversible column backbone network with an attention mechanism.
8. An electronic device comprising a memory storing a computer program and a processor implementing the method of any of claims 1 to 6 when the computer program is executed by the processor.
9. A computer readable storage medium storing a computer program, characterized in that the computer program, when executed by a processor, implements the method of any one of claims 1 to 6.
10. A computer program product comprising a computer program, characterized in that the computer program, when executed by a processor, implements the method of any one of claims 1 to 6.

Description

Chicken number detection method, device, equipment, medium and program product Technical Field The application relates to the technical field of poultry management, in particular to a chicken number detection method, a device, equipment, a medium and a program product. Background Currently, in a large-scale poultry farm, accurate counting of the number of live chickens is a key for realizing intelligent accurate management and health monitoring. In the related art, the traditional manual checking method needs to check the chickens in the henhouse for many times every day by a inspector, and the quantity and the state of each cage chicken are manually recorded. However, in practical application, the traditional checking method has the problems of larger labor capacity, heavier labor capacity, poorer manual checking accuracy and the like. In summary, the technical problems in the related art are to be improved. Disclosure of Invention The embodiment of the application provides a chicken number detection method, a device, equipment, a medium and a program product, which can automatically, efficiently and accurately monitor chicken number, obviously reduce labor cost and inventory pressure and improve cultivation efficiency and management level. In one aspect, the embodiment of the application provides a chicken number detection method, which comprises the following steps: Acquiring a chicken shooting image of a target area; Inputting the chicken shooting image into a chicken quantity detection model to obtain a chicken quantity detection result output by the chicken quantity detection model; The chicken quantity detection model is constructed based on a YOLO model and combines a reversible column backbone network with an attention mechanism. Optionally, the inputting the chicken shot image to a chicken quantity detection model to obtain a chicken quantity detection result output by the chicken quantity detection model includes: inputting the chicken shooting image into a chicken quantity detection model; And detecting and determining the range of the chicken coops through the chicken number detection model, detecting and determining the number of chicken in the chicken coops, and outputting the chicken number as a chicken number detection result. Optionally, the detecting and determining the range of the chicken coops through the chicken number detecting model, further detecting and determining the number of chicken in the chicken coops, and outputting the chicken number as a chicken number detecting result, including: Identifying the boundary frame of the chicken coop fence in the chicken shooting image through the chicken quantity detection model, extracting the minimum abscissa and the maximum abscissa of the boundary frame of the chicken coop fence, and defining a chicken coop range; Detecting and identifying the number and the state of the chickens in the range of the chicken coop, and outputting a chicken number detection result. Optionally, the method further comprises: according to the number information of the chicken coops, determining a chicken quantity detection result of multi-frame chicken shooting images of the chicken coops; Under the condition that the abnormal frames exist in the detection result of the number of the multi-frame chicken shooting images of the chicken coops, determining the chicken shooting images of the abnormal frames and corresponding chicken coops numbering information, and generating abnormal warning information; Under the condition that abnormal frames do not exist in the chicken quantity detection results of the multi-frame chicken shooting images of the chicken coops, determining the chicken quantity detection result with the highest numerical value in the chicken quantity detection results of all the multi-frame chicken shooting images, and uploading the chicken quantity detection results as the chicken quantity detection results of the chicken coops. Optionally, after the capturing of the image of the chicken in the target area, the method further includes: performing image enhancement on the chicken shooting image, and cutting the chicken shooting image to a preset specification; wherein the image enhancement includes adaptive histogram equalization and high pass filtering. Optionally, the chicken number detection model is trained based on the following steps: Acquiring a plurality of historical chicken shooting images, and acquiring chicken quantity detection results corresponding to the historical chicken shooting images; taking the images shot by each historical chicken as a sample, taking the detection results of the number of chicken corresponding to the images shot by each historical chicken as sample tags corresponding to the samples, and constructing a training data set; and pre-training the chicken quantity detection model by using the training data set. In another aspect, an embodiment of the present application provides a chicken number detecting device,