Hatching egg incubation and fertilization rate detection system based on YOLO algorithm and photoelectric acquisition

Abstract

The invention discloses a hatching egg incubation and fertilization rate detection system based on a YOLO algorithm and photoelectric acquisition, and the system comprises a control module which is used for obtaining a corresponding hatching egg detection image through the combination of the YOLO algorithm, carrying out the detection and analysis through a suitable model, and rapidly and accurately distinguishing a fertilized egg from an unfertilized egg; the shading module is used for forming a dark environment in the hatching egg detection process to avoid interference of other light rays; the image acquisition module is used for acquiring a hatching egg detection image irradiated by the LED light and transmitting the hatching egg detection image to the control module; the illumination module is used for providing hatching light in the vertical direction for the hatching eggs to be detected; and the conveying module is used for driving the hatching eggs to be detected to enter the shading module for shooting and collecting the detection images.

Description

Technical field:

[0001] The present invention relates to a hatching egg fertilization rate detection system based on a YOLO algorithm and photoelectric acquisition. Background technology:

[0002] Hatching refers to the process in which an animal embryo breaks through the egg membrane and begins to live freely in the outside world. It generally refers to oviparous animals. In corresponding agricultural projects, in order to increase the hatching yield of breeding eggs, corresponding testing of fertilized eggs is required.

[0003] Existing methods for detecting fertilized eggs during hatching primarily utilize spectroscopy and deep learning. Near-infrared equipment, spectral imaging systems, and spectroscopy principles are combined to collect data on eggs under different conditions to determine whether they are fertilized. However, this acquisition method is often only applicable to image recognition of single eggs, which can lead to excessively high overall equipment costs for detecting a large number of eggs. Furthermore, this technology is susceptible to interference from the detection equipment and light sources, resulting in prolonged detection times. Summary of the invention:

[0004] The embodiment of the present invention provides a hatching egg fertilization rate detection system based on the YOLO algorithm and photoelectric acquisition. The system has a reasonable structural design and integrates the YOLO algorithm into the photoelectric detection equipment to distinguish the developmental differences between fertilized and unfertilized eggs. Low-cost and short-term hatching egg fertilization detection is achieved through an image acquisition control platform. The system can perform image acquisition and detection actions for multiple eggs in a single batch, avoid interference from detection equipment and light sources, improve detection efficiency, obtain accurate hatching fertilization rates, ensure that detection accuracy can meet actual application requirements, and solve the problems existing in the prior art.

[0005] The technical solution adopted by the present invention to solve the above technical problems is:

[0006] A hatching egg fertilization rate detection system based on the YOLO algorithm and photoelectric acquisition, the detection system includes:

[0007] A control module is used to combine the YOLO algorithm to obtain corresponding egg detection images and use an applicable model for detection and analysis to quickly and accurately distinguish fertilized eggs from unfertilized eggs;

[0008] A shading module is used to create a dark environment during the egg detection process to avoid interference from other light;

[0009] An image acquisition module is used to acquire images of the eggs irradiated by the LED light and transmit the images to the control module;

[0010] An illumination module, which is used to provide vertical hatching light for the eggs to be inspected;

[0011] The transmission module is used to drive the eggs to be tested into the shading module for shooting and collecting the test images.

[0012] The shading module includes a darkroom, which is constructed of acrylic plates and is equipped with light-proof curtains at both ends of the darkroom to ensure that the eggs are in a dark environment when entering the darkroom.

[0013] The image acquisition module includes a camera arranged in a vertical direction to take pictures of the eggs to be inspected to obtain inspection images;

[0014] The detection images can be captured and collected at different incubation time points and different incubation light intensities, and processed in real time using OpenCV, including pre-processing and analysis of the detection images; the processing includes image scaling, grayscale conversion, and denoising to ensure that the input data is consistent with the input size required by the model, thereby improving the detection accuracy of the model;

[0015] The OpenCV can be used to visualize the detection results. During the image recognition process, YOLO is used to identify multiple targets in the image and output corresponding bounding boxes and classification labels. The detected targets are then marked with rectangular boxes through OpenCV, and corresponding category labels and confidence levels are set to intuitively display the monitoring results.

[0016] The lighting module includes an LED light board, which includes 30 groups of lights. Each group of lights consists of 9 high-intensity cold white LED lights. Each group of lights is evenly distributed in a circle, and parallel circuits are used within and between groups for independent power supply.

[0017] The conveying module includes a conveyor belt and an egg tray. The conveyor belt is used to drive the eggs to be tested into the shading module for testing. The egg tray adopts a structure of 5 rows and 6 columns, with a total of 30 grids. Each grid is a completely black and opaque square. Two thorns are respectively provided on the top four sides of each grid for fixing a rubber sponge pad. The rubber sponge pad is used to converge the light beam of the light source and reduce the interference of the light sources between different grids. A right-angled trapezoidal support structure is added at the midpoint of the four sides of the bottom of each grid to stabilize and fix the position of the eggs to be tested, thereby ensuring the uniformity of light and the stability of the eggs during the detection process, and providing reliable protection for the hatching detection of breeding eggs.

[0018] The control module uses PyCharm to provide support for the development environment, with built-in debugging tools and an interactive Python console, making model debugging and data preprocessing more convenient, and capable of viewing and modifying data output in real time; the control module uses PySide6 as a framework for graphical user interface development, which is used to build a visual interface for the image detection system and perform image loading, model selection, and result display on the visual interface; the control module selects different images to be tested for detection through image loading, realizes arbitrary switching of multiple models in the research content through model selection, and displays the image detection results intuitively and clearly through result display.

[0019] The graphical user interface includes a login interface and a detection interface; the detection interface includes a model switching interface, an image detection interface and a configuration information interface. The model switching interface is used to switch the homepage of the detection model, the image detection interface is used to upload pictures and perform detection, and the configuration information interface is used to modify the image display size and the detection model IOU threshold.

[0020] The present invention adopts the above structure, obtains corresponding egg detection images through the control module in combination with the YOLO algorithm, adopts an applicable model to perform detection and analysis, and quickly and accurately distinguishes fertilized eggs from unfertilized eggs; forms a dark environment during the egg detection process through the shading module to avoid interference from other light; collects the egg detection images irradiated by LED light through the image acquisition module and transmits them to the control module; provides vertical hatching light for the eggs to be detected through the illumination module; and drives the eggs to be detected into the shading module through the transmission module to capture and collect the detection images, so as to distinguish fertilized eggs from unfertilized eggs. The invention has the advantages of accuracy, high efficiency, economy and practicality. Description of the drawings:

[0021] Figure 1 It is a structural schematic diagram of the present invention.

[0022] Figure 2 It is a schematic diagram of the process of the present invention.

[0023] Figure 3 This is the system login interface of the present invention.

[0024] Figure 4 It is the model switching interface of the present invention.

[0025] Figure 5 This is the image detection interface of the present invention.

[0026] Figure 6 It is the configuration information interface of the present invention.

[0027] Figure 7 Schematic diagram of the device of the present invention. Specific implementation method:

[0028] In order to clearly illustrate the technical features of this solution, the present invention is described in detail below through specific implementation methods and in conjunction with the accompanying drawings.

[0029] like Figure 1-7 As shown in the figure, a hatching egg fertility rate detection system based on the YOLO algorithm and photoelectric acquisition includes:

[0030] A control module is used to combine the YOLO algorithm to obtain corresponding egg detection images and use an applicable model for detection and analysis to quickly and accurately distinguish fertilized eggs from unfertilized eggs;

[0031] A shading module is used to create a dark environment during the egg detection process to avoid interference from other light;

[0032] An image acquisition module is used to acquire images of the eggs irradiated by the LED light and transmit the images to the control module;

[0033] An illumination module, which is used to provide vertical hatching light for the eggs to be inspected;

[0034] The transmission module is used to drive the eggs to be tested into the shading module for shooting and collecting the test images.

[0035] The shading module includes a darkroom, which is constructed of acrylic plates and is equipped with light-proof curtains at both ends of the darkroom to ensure that the eggs are in a dark environment when entering the darkroom.

[0036] The image acquisition module includes a camera arranged in a vertical direction to take pictures of the eggs to be inspected to obtain inspection images;

[0037] The detection images can be captured and collected at different incubation time points and different incubation light intensities, and processed in real time using OpenCV, including pre-processing and analysis of the detection images; the processing includes image scaling, grayscale conversion, and denoising to ensure that the input data is consistent with the input size required by the model, thereby improving the detection accuracy of the model;

[0038] The OpenCV can be used to visualize the detection results. During the image recognition process, YOLO is used to identify multiple targets in the image and output corresponding bounding boxes and classification labels. The detected targets are then marked with rectangular boxes through OpenCV, and corresponding category labels and confidence levels are set to intuitively display the monitoring results.

[0039] The lighting module includes an LED light board, which includes 30 groups of lights. Each group of lights consists of 9 high-intensity cold white LED lights. Each group of lights is evenly distributed in a circle, and parallel circuits are used within and between groups for independent power supply.

[0040] The conveying module includes a conveyor belt and an egg tray. The conveyor belt is used to drive the eggs to be tested into the shading module for testing. The egg tray adopts a structure of 5 rows and 6 columns, with a total of 30 grids. Each grid is a completely black and opaque square. Two thorns are respectively provided on the top four sides of each grid for fixing a rubber sponge pad. The rubber sponge pad is used to converge the light beam of the light source and reduce the interference of the light sources between different grids. A right-angled trapezoidal support structure is added at the midpoint of the four sides of the bottom of each grid to stabilize and fix the position of the eggs to be tested, thereby ensuring the uniformity of light and the stability of the eggs during the detection process, and providing reliable protection for the hatching detection of breeding eggs.

[0041] The control module uses PyCharm to provide support for the development environment, with built-in debugging tools and an interactive Python console, making model debugging and data preprocessing more convenient, and capable of viewing and modifying data output in real time; the control module uses PySide6 as a framework for graphical user interface development, which is used to build a visual interface for the image detection system and perform image loading, model selection, and result display on the visual interface; the control module selects different images to be tested for detection through image loading, realizes arbitrary switching of multiple models in the research content through model selection, and displays the image detection results intuitively and clearly through result display.

[0042] The graphical user interface includes a login interface and a detection interface; the detection interface includes a model switching interface, an image detection interface and a configuration information interface. The model switching interface is used to switch the homepage of the detection model, the image detection interface is used to upload pictures and perform detection, and the configuration information interface is used to modify the image display size and the detection model IOU threshold.

[0043] The working principle of the hatching and fertilization rate detection system of hatching eggs based on the YOLO algorithm and photoelectric acquisition in the embodiment of the present invention is as follows: the YOLO algorithm is integrated into the photoelectric detection equipment to distinguish the developmental differences between fertilized and unfertilized eggs, and low-cost and short-term hatching and fertilization detection of hatching eggs is realized through the image acquisition control platform. It can perform image acquisition and detection actions for multiple hatching eggs in a single batch, avoid interference from detection equipment and light sources, improve detection efficiency, obtain accurate hatching fertilization rates, and ensure that the detection accuracy can meet actual application requirements.

[0044] Since most existing conventional technologies can only capture and identify images of a single egg at a time, the equipment cost is too high. At the same time, detection methods that rely solely on deep learning algorithms are easily interfered with by light sources, which prolongs the detection time and reduces the overall detection efficiency.

[0045] The overall solution mainly includes a control module, which is used to combine the YOLO algorithm to obtain the corresponding egg detection images and use the applicable model for detection and analysis, so as to quickly and accurately distinguish fertilized eggs from unfertilized eggs; a shading module, which is used to create a dark environment during the egg detection process to avoid interference from other light; an image acquisition module, which is used to collect the egg detection images illuminated by LED light and transmit them to the control module; an illumination module, which is used to provide vertical incubation light for the eggs to be detected; and a transmission module, which is used to drive the eggs to be detected into the shading module for shooting and collecting the detection images.

[0046] In actual use, Hy-Line white eggs were selected for testing. The YOLOv8 model was used to place 300 unfertilized eggs and 300 fertilized eggs in an incubator, and detection images of Hy-Line white eggs at different incubation time points were collected. This can realize the detection image collection of 30 eggs in a single batch, and can realize the detection of fertilized eggs 48 hours after incubation at a low cost.

[0047] Specifically, the eggs to be tested are taken out of the incubator, the LED light board, conveyor belt and camera are turned on, the test images are collected and saved, and the test images are transmitted to the control module to select the appropriate model for testing, and then the test analysis results are output.

[0048] Preferably, the shading module includes a darkroom, which is constructed with acrylic panels and is equipped with opaque curtains at both ends of the openings to ensure that the breeding eggs are in a dark environment when entering the darkroom. Curtains made of opaque polyester material are provided at both ends of the openings. The width of the darkroom matches the width of the conveyor belt, and the height meets the requirement of the camera to collect images of 30 eggs in 5 rows and 6 columns within the field of view.

[0049] Preferably, the image acquisition module includes a camera arranged in a vertical direction to take pictures of the eggs to be tested to obtain test images; the camera is a Nikon Z30, which can be used to shoot and collect images in combination with different incubation time nodes and different incubation light intensities, and use OpenCV to perform real-time processing of the test images, including preprocessing and analysis of the test images; the processing includes image scaling, grayscale conversion and denoising to ensure that the input data is consistent with the input size required by the model, thereby improving the detection accuracy of the model.

[0050] OpenCV is an open-source computer vision library widely used for real-time image processing and computer vision tasks. It provides a rich set of image processing algorithms and functions, helping developers perform operations such as image preprocessing, feature extraction, image enhancement, geometric transformations, and image segmentation, making it suitable for a variety of computer vision tasks. In this application, OpenCV primarily handles image processing, object detection result visualization, and image storage.

[0051] During the image recognition process, YOLO is used to identify multiple targets in the image and output the corresponding bounding boxes and classification labels. Then, OpenCV is used to mark the detected targets with rectangular boxes, set the corresponding category labels and confidence levels, and intuitively display the monitoring results.

[0052] For the illumination module and transmission module, the illumination module includes an LED light board, which includes 30 groups of lights. Each group of lights consists of 9 high-intensity cold white LED lights. Each group of lights is evenly distributed in a circle, and parallel circuits are used within and between groups for independent power supply.

[0053] The conveying module includes a conveyor belt and an egg tray. The conveyor belt is used to drive the eggs to be tested into the shading module for testing. The egg tray adopts a structure of 5 rows and 6 columns, with a total of 30 grids. Each grid is a completely black and opaque square. Two convex thorns are set on the top four sides of each grid for fixing rubber sponge pads. The rubber sponge pads are used to converge the light beam of the light source and reduce the interference of light sources between different grids. A right-angled trapezoidal support structure is added at the midpoint of the four sides of the bottom of each grid to stabilize the position of the eggs to be tested, thereby ensuring the uniformity of light and the stability of the eggs during the detection process, providing reliable protection for egg hatching detection.

[0054] The core component of this application is the control module, which uses PyCharm to provide support for the development environment, has built-in debugging tools and an interactive Python console, making model debugging and data preprocessing more convenient, and capable of viewing and modifying data output in real time; the control module uses PySide6 as a framework for graphical user interface development, which is used to build a visual interface for the image detection system and perform image loading, model selection and result display in the visual interface; the control module selects different images to be tested for detection through image loading, realizes arbitrary switching of multiple models in the research content through model selection, and displays the image detection results intuitively and clearly through result display.

[0055] For the graphical user interface, it includes a login interface and a detection interface; the detection interface includes a model switching interface, an image detection interface and a configuration information interface. The model switching interface is used to switch the homepage of the detection model, the image detection interface is used to upload pictures and perform detection, and the configuration information interface is used to modify the image display size and the detection model IOU threshold.

[0056] For this example, the test set consists of 45 hatching eggs and 45 unfertilized eggs, for a total of 90 eggs. To demonstrate the effectiveness of YOLOv8 model training, the best trained YOLOv8 model is used for testing and the test results are recorded.

[0057] Under different light intensity conditions, the performance of hatching egg detection will change with the incubation time, such as precision, recall rate, mAP50 and mAP50-95, and thus a comparison table of hatching egg detection performance under different lighting conditions and different incubation times and training set data are obtained.

[0058] Table 1 Training set data under light intensity of 180 lumens

[0059]

[0060] Table 2 Training set data under light intensity of 360 lumens

[0061]

[0062] Table 3 Training set data under light intensity of 540 lumens

[0063]

[0064] The training set data shows that under a light intensity of 540 lumens, the precision, mAP50, and mAP50-95 values are all higher, indicating that detection accuracy and overall performance are better under this light intensity condition. However, in terms of recall, the results under light intensities of 180 lumens and 360 lumens are more prominent, indicating that these light intensity conditions are more advantageous in capturing more true positive samples. This difference reflects the trade-off between precision and recall: at specific light intensities and incubation times, an increase in precision is often accompanied by a decrease in recall.

[0065] Optimal light intensity conditions require a balance between four indicators, and higher numbers were achieved at an incubation time of 48 hours. This shows that a 48-hour incubation time provides a more balanced solution under different light intensity conditions and is an important reference point for optimizing system performance.

[0066] We selected the YOLOv11 data augmentation model for analysis, using a light intensity of 540 lumens, which had the most significant effect. Various losses and evaluation indicators during the training process, including box regression in object detection, losses in classification and positioning tasks, as well as the precision and recall of the model on the training and validation sets, were used to adjust the model parameters according to corresponding change trends to achieve better performance.

[0067] It should be noted that the present application can perform model analysis through the F1-confidence curve, the precision-confidence curve, the precision-recall curve and the recall-confidence curve, so as to more accurately select a suitable data model.

[0068] In summary, the hatching and fertilization rate detection system for hatching eggs based on the YOLO algorithm and photoelectric acquisition in the embodiment of the present invention integrates the YOLO algorithm into the photoelectric detection equipment, distinguishes the developmental differences between fertilized and unfertilized eggs, and realizes low-cost and short-term hatching and fertilization detection of hatching eggs through the image acquisition control platform. It can perform image acquisition and detection actions for multiple eggs in a single batch, avoid interference from detection equipment and light sources, improve detection efficiency, obtain accurate hatching fertilization rates, and ensure that the detection accuracy can meet actual application requirements.

[0069] The above specific implementation manner cannot be used as a limitation on the protection scope of the present invention. For those skilled in the art, any replacement, improvement or transformation made to the implementation manner of the present invention falls within the protection scope of the present invention.

[0070] Any matters not described in detail in the present invention are well-known technologies to those skilled in the art.

Claims

1. The hatching egg fertilization rate detection system based on the YOLO algorithm and photoelectric acquisition is characterized by: The detection system comprises: A control module is used to combine the YOLO algorithm to obtain corresponding egg detection images and use an applicable model for detection and analysis to quickly and accurately distinguish fertilized eggs from unfertilized eggs; A shading module is used to create a dark environment during the egg detection process to avoid interference from other light; An image acquisition module is used to acquire images of the eggs irradiated by the LED light and transmit the images to the control module; An illumination module, which is used to provide vertical hatching light for the eggs to be inspected; The transmission module is used to drive the eggs to be tested into the shading module for shooting and collecting the test images.

2. The egg hatching fertilization rate detection system based on the YOLO algorithm and photoelectric acquisition according to claim 1, characterized in that: The shading module includes a darkroom, which is constructed of acrylic plates and is equipped with light-proof curtains at both ends of the darkroom to ensure that the eggs are in a dark environment when entering the darkroom.

3. The egg hatching fertilization rate detection system based on the YOLO algorithm and photoelectric acquisition according to claim 1, characterized in that: The image acquisition module includes a camera arranged in a vertical direction to take pictures of the eggs to be inspected to obtain inspection images; The detection images can be captured and collected at different incubation time points and different incubation light intensities, and processed in real time using OpenCV, including pre-processing and analysis of the detection images; the processing includes image scaling, grayscale conversion, and denoising to ensure that the input data is consistent with the input size required by the model, thereby improving the detection accuracy of the model; The OpenCV can be used to visualize the detection results. During the image recognition process, YOLO is used to identify multiple targets in the image and output corresponding bounding boxes and classification labels. The detected targets are then marked with rectangular boxes through OpenCV, and corresponding category labels and confidence levels are set to intuitively display the monitoring results.

4. The egg hatching fertilization rate detection system based on the YOLO algorithm and photoelectric acquisition according to claim 1, characterized in that: The lighting module includes an LED light board, which includes 30 groups of lights. Each group of lights consists of 9 high-intensity cold white LED lights. Each group of lights is evenly distributed in a circle, and parallel circuits are used within and between groups for independent power supply.

5. The egg hatching fertilization rate detection system based on the YOLO algorithm and photoelectric acquisition according to claim 1 is characterized in that: The conveying module includes a conveyor belt and an egg tray. The conveyor belt is used to drive the eggs to be tested into the shading module for testing. The egg tray adopts a structure of 5 rows and 6 columns, with a total of 30 grids. Each grid is a completely black and opaque square. Two thorns are respectively provided on the top four sides of each grid for fixing a rubber sponge pad. The rubber sponge pad is used to converge the light beam of the light source and reduce the interference of the light sources between different grids. A right-angled trapezoidal support structure is added at the midpoint of the four sides of the bottom of each grid to stabilize and fix the position of the eggs to be tested, thereby ensuring the uniformity of light and the stability of the eggs during the detection process, and providing reliable protection for the hatching detection of breeding eggs.

6. The egg hatching fertilization rate detection system based on the YOLO algorithm and photoelectric acquisition according to claim 1, characterized in that: The control module uses PyCharm as a development environment support, with built-in debugging tools and an interactive Python console, making model debugging and data preprocessing more convenient, and enabling real-time viewing and modification of data output; The control module uses PySide6 as a framework for graphical user interface development to build a visual interface for the image detection system and perform image loading, model selection, and result display on the visual interface; The control module selects different images to be tested for detection through image loading, realizes the arbitrary switching of multiple models in the research content through model selection, and displays the image detection results intuitively and clearly through result display.

7. The egg hatching fertilization rate detection system based on the YOLO algorithm and photoelectric acquisition according to claim 6, characterized in that: The graphical user interface includes a login interface and a detection interface; the detection interface includes a model switching interface, an image detection interface and a configuration information interface. The model switching interface is used to switch the homepage of the detection model, the image detection interface is used to upload pictures and perform detection, and the configuration information interface is used to modify the image display size and the detection model IOU threshold.

Images