Preserved egg quality detection method, program product, medium, device and preserved egg screening system

By integrating machine learning models that can be separated in depth, feature pyramid networks and image attention neural networks, the problems of inefficient and insufficient accuracy of traditional egg quality detection methods are solved, and efficient and accurate egg quality detection is achieved.

CN119073244APending Publication Date: 2024-12-06HUBEI POLYTECHNIC UNIV
View PDF 8 Cites 0 Cited by

Patent Information

Application Number
CN202411481610.8
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-10-23
Publication Date
2024-12-06

AI Technical Summary

Technical Problem

Traditional egg quality detection methods are inefficient and insufficiently accurate, making it difficult to meet consumers' requirements for improving product quality.

Method used

Using machine learning models integrating deep separable convolutional networks, feature pyramid networks and image attention neural networks for image feature extraction and quality type recognition, combined with histogram equalization, contrast adjustment, denoising and quantization processing, through softmax The classifier determines the quality type of the egg.

Benefits of technology

It improves the accuracy and efficiency of quality detection of eggs, reduces the influence of human factors, realizes a fast and non-destructive detection process, and significantly enhances the ability to identify complex features of eggs images.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN119073244A_ABST
    Figure CN119073244A_ABST
Patent Text Reader

Abstract

The invention discloses a preserved egg quality detection method, a program product, a medium, a device and a preserved egg screening system, and the method comprises the steps: obtaining a pre-trained machine learning model which is integrated with a deep separable convolutional network, a feature pyramid network and an image attention neural network; acquiring a preserved egg image of a to-be-detected preserved egg, and performing feature extraction on the preserved egg image based on the machine learning model to obtain a preserved egg feature map of the to-be-detected preserved egg; and determining a target quality type of the to-be-detected preserved egg based on the preserved egg feature map. According to the technical scheme provided by the invention, the quality detection efficiency of the preserved eggs can be ensured, and meanwhile, the quality detection accuracy of the preserved eggs is improved.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The present application belongs to the technical field of preserved egg quality detection, and in particular, relates to a preserved egg quality detection method, program product, medium, device and preserved egg screening system. Background Art

[0002] At present, in the production scenario of preserved eggs, the quality inspection of preserved eggs is the key to ensure product safety and quality. However, the traditional manual inspection method has problems such as low efficiency and inconsistent judgment standards; and as consumers' requirements for product quality increase, it becomes urgent to introduce efficient and accurate methods of preserved egg quality inspection. Based on this, how to improve the accuracy of preserved egg quality inspection while ensuring the efficiency of preserved egg quality inspection is a technical problem that needs to be solved urgently. Summary of the invention

[0003] The embodiments of the present application provide a preserved egg quality detection method, program product, medium, device and preserved egg screening system, which can improve the accuracy of preserved egg quality detection while ensuring the efficiency of preserved egg quality detection to at least a certain extent.

[0004] Other features and advantages of the present application will become apparent from the following detailed description, or may be learned in part by the practice of the present application.

[0005] According to a first aspect of an embodiment of the present application, a method for detecting the quality of preserved eggs is provided, the method comprising: obtaining a pre-trained machine learning model, the machine learning model integrating a deep separable convolutional network, a feature pyramid network, and an image attention neural network; obtaining a preserved egg image of a preserved egg to be detected, and performing feature extraction on the preserved egg image based on the machine learning model to obtain a preserved egg feature map of the preserved egg to be detected; based on the preserved egg feature map, determining a target quality type of the preserved egg to be detected.

[0006] In some embodiments of the present application, based on the aforementioned scheme, the machine learning model is trained through the following steps: obtaining preserved egg sample images of different quality types, and the preserved egg sample images are annotated with corresponding quality type labels; based on the preserved egg sample images of different quality types, training a pre-constructed machine learning model framework to obtain the machine learning model.

[0007] In some embodiments of the present application, based on the aforementioned scheme, the method further includes: performing quantization and pruning processing on the machine learning model.

[0008] In some embodiments of the present application, based on the aforementioned scheme, before performing feature extraction on the preserved egg image based on the machine learning model, the method also includes: performing histogram equalization and contrast adjustment on the preserved egg image; denoising the preserved egg image through Gaussian filtering algorithm and median filtering algorithm; and standardizing the preserved egg image to a set image size.

[0009] In some embodiments of the present application, based on the aforementioned scheme, determining the target quality type of the preserved egg to be tested based on the preserved egg characteristic map includes: based on a global average pooling algorithm, converting the preserved egg characteristic map into a feature vector of a set length; based on the feature vector, determining the probability distribution of the preserved egg to be tested belonging to each quality type through a softmax classifier; and determining the quality type with the highest probability as the target quality type of the preserved egg to be tested.

[0010] According to a second aspect of an embodiment of the present application, a computer program product is provided, which includes computer instructions, which are stored in a computer-readable storage medium and are suitable for being read and executed by a processor, so that a computer device having the processor executes to implement the operations performed by the method described in any one of the first aspects above.

[0011] According to a third aspect of an embodiment of the present application, a computer-readable storage medium is provided, in which at least one computer program instruction is stored. The at least one computer program instruction is loaded and executed by a processor to implement the operations performed by the method described in any one of the first aspects above.

[0012] According to the fourth aspect of the embodiments of the present application, a preserved egg quality detection device is provided, which includes one or more processors and one or more memories, and at least one program code is stored in the one or more memories. The at least one program code is loaded and executed by the one or more processors to implement the method described in any embodiment of the first aspect above.

[0013] According to a fifth aspect of an embodiment of the present application, a preserved egg screening system is provided, the preserved egg screening system comprising: a preserved egg quality detection device as described in the fifth aspect above, used to execute a method as described in any embodiment of the first aspect above; a conveying component, the conveying component being used to convey the preserved eggs to be inspected; a preserved egg image acquisition component, the preserved egg image acquisition component being arranged on the conveying component, used to capture the preserved egg image of the preserved egg to be inspected placed on the conveying component, and sending the preserved egg image to the preserved egg quality detection device; a flap assembly, the flap assembly being arranged on the conveying component, used to remove the preserved eggs to be inspected of a preset quality type from the conveying component according to the detection result of the preserved egg quality detection device.

[0014] In some embodiments of the present application, based on the aforementioned scheme, the preserved egg image acquisition component includes a darkroom, and a light source and an image acquisition device arranged in the darkroom, the conveying chain of the conveying component passes through the darkroom, the light source is arranged on one side of the conveying chain, and the image acquisition device is arranged on the other side of the conveying chain.

[0015] Based on the technical solution proposed in the present application, the machine learning model used to extract the quality characteristics of preserved eggs can achieve effective fusion and attention to features of different scales in the preserved egg image by integrating a deep separable convolutional network, a feature pyramid network and an image attention neural network, and significantly enhances the machine learning model's ability to recognize complex features of the preserved egg image, thereby improving the efficiency and accuracy of feature extraction of the preserved egg image. It can be seen that the quality inspection of the preserved eggs to be inspected through the technical solution proposed in the present application can improve the accuracy of the preserved egg quality inspection while ensuring the efficiency of the preserved egg quality inspection.

[0016] It should be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the present application. BRIEF DESCRIPTION OF THE DRAWINGS

[0017] The drawings herein are incorporated into the specification and constitute a part of the specification, showing embodiments consistent with the present application, and together with the specification, are used to explain the principles of the present application. Obviously, the drawings described below are only some embodiments of the present application, and for ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:

[0018] Figure 1 A schematic diagram of a preserved egg screening system in one embodiment of the present application is shown;

[0019] Figure 2 An example diagram showing different quality types of preserved eggs in one embodiment of the present application is shown;

[0020] Figure 3 A flow chart showing a method for detecting the quality of preserved eggs in one embodiment of the present application is shown;

[0021] Figure 4 A machine learning model framework diagram in one embodiment of the present application is shown;

[0022] Figure 5 A schematic structural diagram of a preserved egg quality detection device in one embodiment of the present application is shown. DETAILED DESCRIPTION

[0023] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0024] In addition, described feature, structure or characteristic can be combined in one or more embodiments in any suitable manner. In the following description, many specific details are provided to provide a full understanding of the embodiments of the present application. However, those skilled in the art will appreciate that the technical scheme of the present application can be put into practice without one or more of the specific details, or other methods, components, devices, steps, etc. can be adopted. In other cases, known methods, devices, realizations or operations are not shown or described in detail to avoid blurring the various aspects of the application.

[0025] The block diagrams shown in the accompanying drawings are merely functional entities and do not necessarily correspond to physically independent entities. That is, these functional entities may be implemented in software form, or in one or more hardware modules or integrated circuits, or in different networks and / or processor devices and / or microcontroller devices.

[0026] The flowcharts shown in the accompanying drawings are only exemplary and do not necessarily include all the contents and operations / steps, nor must they be executed in the order described. For example, some operations / steps can be decomposed, and some operations / steps can be combined or partially combined, so the actual execution order may change according to actual conditions.

[0027] It should be noted that the "multiple" mentioned in this article refers to two or more. "And / or" describes the association relationship of associated objects, indicating that three relationships can exist. For example, A and / or B can mean: A exists alone, A and B exist at the same time, and B exists alone. The character " / " generally indicates that the associated objects before and after are in an "or" relationship.

[0028] It should also be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the objects used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those shown or described.

[0029] In order to make those skilled in the art better understand the present application, Figure 1 A brief description is given of the preserved egg screening system involved in this application.

[0030] See also Figure 1 , shows a schematic diagram of a preserved egg screening system in one embodiment of the present application. Figure 1 As shown, the preserved egg screening system described in the present application may include a conveying component, a preserved egg image acquisition component, a flip plate component and a preserved egg quality detection device.

[0031] The transmission assembly may include a driving wheel 101, a chain slideway 102, a chain 103, a slideway 107 and a driven wheel 108, and the transmission assembly may be used to transmit the preserved egg to be detected 105. Specifically, the driving wheel 101 may drive the chain 103 to realize the transmission of the preserved egg to be detected.

[0032] The preserved egg image acquisition component may include a preserved egg 105 to be detected, an image acquisition device 106 (such as a camera), a darkroom 110 and a light source 111. The preserved egg image acquisition component may be arranged in the conveying component, and the chain 103 of the conveying component passes through the darkroom 110. The light source 111 is arranged on one side (such as the lower side) of the chain 103, and the image acquisition device 106 is arranged on the other side (such as the upper side) of the chain 103. The image acquisition device 106 can acquire the preserved egg image of the preserved egg 105 to be detected placed on the conveying component, and send the preserved egg image to the preserved egg quality detection device. In the present application, a darkroom is set in the preserved egg screening system, and its advantage is that it can provide a stable background light environment for the image acquisition device 106 to shoot the preserved egg image, thereby reducing the interference of external light on image acquisition.

[0033] The flap assembly may include a flap 104 and a cylinder 109. The flap assembly may be arranged on the conveying assembly and may also remove the preserved eggs to be inspected of a preset quality type (such as preserved eggs that do not meet the quality requirements) from the conveying assembly according to the inspection results of the preserved egg quality inspection device.

[0034] The preserved egg quality detection device (not shown in the figure) may include an industrial computer for receiving the preserved egg image collected by the preserved egg image acquisition component 106, and processing and analyzing the preserved egg image based on the preserved egg quality detection method proposed in this application, and a PLC control device (not shown in the figure) that can respond to the instructions issued by the industrial computer based on the analysis results to control the operation of the transmission component, the preserved egg image acquisition component and the flip plate component. For example, when the industrial computer determines that a preserved egg belongs to a preset quality type of preserved egg, the PLC control device can control the cylinder 109 to push the flip plate 104, thereby removing the preserved egg from the preserved egg screening system.

[0035] In a specific embodiment, the image acquisition device 106 may be an industrial camera, and the light source 111 may be an LED par light.

[0036] In some other embodiments, the image acquisition device may also be a hyperspectral imager or an infrared imager or other device, which can capture richer spectral information or thermal radiation information, and provide more comprehensive data support for non-destructive testing of preserved egg quality.

[0037] In some other embodiments, the light source may also be other light sources with high brightness, long life and good transmittance, such as a laser light source or an LED array of a specific wavelength. These light sources have higher energy concentration and better spectral characteristics, which help to improve the quality of image acquisition. This application does not specifically limit the types of image acquisition devices and light sources.

[0038] Next, the present application will elaborate on the work flow of a specific preserved egg screening system, wherein the image detection device 106 may be an industrial camera.

[0039] First, the preserved egg image of the preserved egg to be tested is collected by the preserved egg image acquisition component. The preserved egg 105 to be tested is placed on the chain 103. As the chain 103 moves, the preserved egg 105 to be tested gradually approaches the industrial camera 106. When the preserved egg 105 to be tested passes through the photoelectric tube switch and enters the darkroom 110, the PLC control device receives the signal sent by the photoelectric tube switch and sends a photo command to the industrial camera 106. After receiving the photo command, the industrial camera 106 takes a photo of the preserved egg 105 to be tested and transmits the collected image information to the industrial computer, wherein the photoelectric tube switch can be set at the entrance of the darkroom 110, and can judge whether the preserved egg 105 to be tested has passed by sensing the irradiation of the light source 111, thereby sending a signal to the PLC control device that the preserved egg 105 to be tested has entered the darkroom 110.

[0040] Secondly, the preserved egg image is processed and analyzed by the preserved egg quality detection device. After the industrial computer receives the preserved egg image information transmitted by the industrial camera 106, it processes and analyzes the image through a machine learning model, which can automatically identify the quality type of the preserved egg and give the detection result. Figure 2 , shows an example diagram of different quality types of preserved eggs in one embodiment of the present application. Figure 2 The quality types of preserved eggs shown can include at least 6 types (such as smooth preserved eggs, stinky preserved eggs, cat's eye eggs, knotted eggs, green preserved eggs and good preserved eggs, etc.). Figure 2 It can be seen that each quality type of preserved egg image has different image features. In this application, the preserved egg feature maps of these preserved eggs can be extracted through a machine learning model, and the quality type of the preserved eggs can be determined based on these preserved egg feature maps.

[0041] Then, the results of the preserved egg quality inspection are transmitted and decided. The industrial computer transmits the inspection results obtained by the machine learning model to the PLC control device in real time, so that the PLC control device can make a decision on whether to remove the preserved egg based on the inspection results.

[0042] Finally, the preserved eggs of different quality types are automatically removed or transferred. Under the control of the PLC control device, if the preserved eggs to be tested are determined to be preserved eggs of the preset quality type (such as preserved eggs that do not meet the quality requirements), the cylinder 109 and the flap 104 are controlled to remove the preserved eggs of the preset quality type from the conveying component; if the preserved eggs to be tested are determined to be preserved eggs that do not belong to the preset quality type (such as preserved eggs that meet the quality requirements), the preserved eggs that meet the quality requirements are allowed to continue to be conveyed to the packaging box by the conveying component.

[0043] It can be seen that, combined with the above-mentioned preserved egg screening process, the preserved egg screening system proposed in the present application can automatically, quickly and accurately realize the screening of preserved eggs, thereby greatly improving the efficiency of preserved egg screening.

[0044] Next, this application will elaborate on the proposed method for detecting the quality of preserved eggs.

[0045] See also Figure 3 , shows a flow chart of a method for detecting the quality of preserved eggs in one embodiment of the present application, the method for detecting the quality of preserved eggs can be executed by a device having a computing and processing function, refer to Figure 3 As shown, the preserved egg quality detection method at least includes steps 310 to 330, which are described in detail as follows:

[0046] Step 310, obtaining a pre-trained machine learning model, wherein the machine learning model integrates a deep separable convolutional network, a feature pyramid network, and an image attention neural network.

[0047] Step 320, obtaining a preserved egg image of the preserved egg to be tested, and performing feature extraction on the preserved egg image based on the machine learning model to obtain a preserved egg feature map of the preserved egg to be tested.

[0048] Step 330, determining the target quality type of the preserved egg to be tested based on the preserved egg characteristic graph.

[0049] In the present application, the machine learning model can be a deep learning model, which integrates technologies such as deep separable convolutional network, feature pyramid network (FPN) and image attention neural network (SE-Block), which not only effectively reduces model parameters and calculation amount, but also improves the feature extraction of preserved egg images and the accuracy of preserved egg classification, thereby improving the accuracy of preserved egg quality detection while ensuring the efficiency of preserved egg quality detection.

[0050] In some other embodiments, the machine learning model may be other machine learning models besides the deep learning model, such as support vector machine, random forest, etc., or a method combining traditional image processing technology (such as edge detection, threshold segmentation, etc.) with deep learning. In addition, new feature extraction and dimensionality reduction technologies (such as principal component analysis, autoencoder, etc.) may be introduced to optimize model performance, and this application does not make too many restrictions on this.

[0051] In this application, the architecture of the machine learning model can refer to Figure 4 The machine learning model framework diagram shown in the figure is used to build Figure 4 , showing a machine learning model framework diagram in one embodiment of the present application.

[0052] like Figure 4 As shown, the machine learning model may include a data acquisition module, an image processing module, a feature extraction module, a classification decision module, and an optimization deployment module.

[0053] In the present application, the image of the preserved egg to be tested is obtained as follows: Figure 4 The data acquisition module shown can capture the preserved egg images of the preserved eggs to be tested in a continuous and uninterrupted manner through a high-resolution industrial camera to ensure that the preserved egg images are clear, detailed, and free of blur, laying a solid foundation for subsequent analysis. For the problem of variable lighting conditions, the intelligent exposure adjustment algorithm and white balance optimization algorithm can be used to enable the machine learning model to automatically adapt to different lighting conditions, thereby ensuring the stable and reliable quality of the preserved egg images.

[0054] In addition, the inventors of the present application found that the traditional method of quality inspection of preserved eggs is complicated to operate, has low inspection efficiency, and may cause damage to the preserved eggs to be inspected. Based on this, the inventors of the present application proposed to perform non-contact image acquisition of the preserved eggs to be inspected through an image acquisition device, and process and analyze the preserved egg images in combination with a machine learning model, thereby realizing a fast and non-destructive inspection process.

[0055] In the above step 310, the training of the machine learning model can be specifically performed according to the following steps 311 to 312:

[0056] Step 311, obtaining preserved egg sample images of different quality types, wherein the preserved egg sample images are annotated with corresponding quality type labels.

[0057] Step 312: Based on the preserved egg sample images of different quality types, a pre-built machine learning model framework is trained to obtain the machine learning model.

[0058] In this application, in order to solve the problem of strong subjectivity and unstable results of traditional manual detection methods, this application constructs a machine learning model framework and trains the machine learning model framework through a large number of preserved egg sample images labeled with corresponding quality types, so as to obtain a machine learning model, so that the machine learning model can extract key features in the preserved egg image, thereby determining the quality type of the preserved egg to be detected. The method of detecting the quality of preserved eggs through a machine learning model can reduce the impact of human factors on the quality detection of preserved eggs, and improve the stability and repeatability of the quality detection of preserved eggs.

[0059] Based on the above steps, the training of the machine learning model can also be performed according to the following step 213:

[0060] Step 313: quantize and prune the machine learning model.

[0061] In this application, if Figure 4 In the optimization deployment module shown, the machine learning model uses quantization processing technology and pruning processing technology, which can significantly reduce the size of the machine learning model and greatly improve the calculation speed of the machine learning model. Among them, the quantization processing technology can convert the weights and activation values ​​of the machine learning model from floating point numbers to fixed point numbers (such as INT8), which can not only effectively reduce the size of the machine learning model, but also improve the calculation speed of the machine learning model.

[0062] The pruning technology further reduces the size of the machine learning model by removing unimportant weights (such as weights close to zero) in the machine learning model, and keeps the machine learning model at a high computing speed. In terms of the method of pruning, the machine learning model proposed in this application adopts an importance-based pruning method, that is, when pruning the machine learning model, multiple factors such as the absolute value of the weight, gradient information, and activation value statistics are comprehensively considered to ensure the accuracy and effectiveness of the pruning process.

[0063] In this application, the machine learning model after quantization and pruning can also be converted into a format suitable for industrial equipment (such as industrial computers) (such as TensorFlow Lite, ONNX, etc.), and an easy-to-integrate API interface is provided to achieve seamless docking of functions such as preserved egg image input, preserved egg quality detection result output and error handling. In addition, deployment tests can also be carried out on industrial equipment to verify the efficiency and accuracy of the machine learning model to ensure that the machine learning model can operate stably and reliably in practical applications. In this way, not only the operating efficiency of the machine learning model is improved, but also the deployment cost can be reduced, which provides strong support for the widespread application of preserved egg quality detection devices.

[0064] In the above step 320, before extracting features from the preserved egg image based on the machine learning model, the following steps 321 to 323 may be specifically performed:

[0065] Step 321, performing histogram equalization processing and contrast adjustment processing on the preserved egg image.

[0066] Step 322: De-noising the preserved egg image by using a Gaussian filter algorithm and a median filter algorithm.

[0067] Step 323, standardizing the preserved egg image to a set image size.

[0068] In this application, if Figure 4 The image processing module shown in the present application can perform histogram equalization and contrast adjustment on the preserved egg image, thereby enhancing the features in the preserved egg image, such as key features such as cracks and stains on the surface of the preserved egg, making the features of the preserved egg image more distinct and easy to identify.

[0069] In addition, the preserved egg image can be denoised using Gaussian filtering algorithm and median filtering algorithm, which can effectively remove random noise in the preserved egg image and further improve the quality of the preserved egg image.

[0070] Finally, the machine learning model can also standardize all input preserved egg images to a set image size, which not only meets the input requirements of the machine learning model, but also ensures the consistency and efficiency of the processing process. The above method can improve the accuracy of data acquisition and the effect of image processing, providing a high-quality data foundation for subsequent feature extraction modules and classification decision modules.

[0071] In a specific embodiment, the set image size may be a fixed size of 224x224 pixels, or may be other set sizes. The present application does not make any specific limitation on the set image size.

[0072] In some other embodiments, during the image processing stage, deep learning-based image super-resolution reconstruction technology may be used to improve the resolution and detail expression of the image. This application does not impose too many limitations on this.

[0073] Continue to combine Figure 4In the feature extraction module, the machine learning model proposed in this application can use a deep separable convolutional network as a basic building block, which can not only greatly reduce the number of parameters and the amount of calculation of the machine learning model, but also maintain a strong feature extraction capability, achieving dual optimization of efficiency and performance. The convolution layer in the deep separable convolutional network can be configured with a suitable convolution kernel size, step size and filling strategy to ensure the stability of the size of the preserved egg feature map. The ReLU activation function can also be introduced to increase the nonlinearity of the machine learning model and further enhance the ability to express the preserved egg features. In addition, the machine learning model can also integrate a feature pyramid network (FPN), and through bottom-up and top-down feature fusion paths, the different scale features of the preserved egg image can be fully enhanced, which can improve the machine learning model's ability to recognize complex defects on the surface of the preserved egg to be detected. In the feature extraction module, an image attention neural network (SE-Block) can also be introduced, and the importance of the preserved egg feature map channel can be dynamically evaluated to achieve the recalibration of the preserved egg features, further improving the model's attention to key features. In terms of parameter setting, this application is adjusted according to specific experimental data to ensure that the model performance is optimal. The machine learning model integrates a deep separable convolutional network, a feature pyramid network and an image attention neural network, which not only improves the accuracy and efficiency of feature extraction, but also provides strong support for subsequent classification decision modules.

[0074] In some other embodiments, in addition to the traditional feature extraction methods based on the color features, texture features and shape features of the preserved egg, the high-level features of the preserved egg image can also be automatically learned and extracted through the feature extraction layer in the deep learning model. In addition, a more comprehensive feature set can be constructed by combining multi-source data such as spectral information and thermal radiation information, and this application does not make too many restrictions on this.

[0075] In the above step 330, the target quality type of the preserved egg to be tested is determined based on the preserved egg characteristic graph, which can be specifically performed according to the following steps 331 to 333:

[0076] Step 331, based on the global average pooling algorithm, convert the preserved egg feature map into a feature vector of a set length.

[0077] Step 332: Based on the feature vector, determine the probability distribution of the preserved egg to be tested belonging to each quality type through a softmax classifier.

[0078] Step 333, determining the quality type with the highest probability as the target quality type of the preserved egg to be tested.

[0079] Continue to combine Figure 4 ,like Figure 4The classification decision module shown is a key module for the preserved egg quality detection method proposed in this application to realize intelligent judgment of the quality type of preserved eggs. In the stage of integrating preserved egg features, the classification decision module can use a global average pooling algorithm to efficiently convert the preserved egg feature map into a feature vector of a set length, which can not only retain the key features in the preserved egg feature map, but also reduce the complexity of calculation. In the classification stage, the preserved egg quality detection method proposed in this application can use a softmax classifier as a classification decision tool. The softmax classifier can determine the probability distribution of each quality type of the preserved egg to be detected based on the input feature vector, and provide rich and accurate information for the decision of the quality type of preserved eggs. The weights and biases of the softmax classifier can be automatically learned and optimized during the training process through the back propagation algorithm, so as to ensure the accuracy and generalization ability of the classification decision. The design of the classification decision module can not only simplify the classification process, but also improve the accuracy and robustness of the classification, and provide a strong guarantee for the intelligentization of preserved egg quality detection.

[0080] In some other embodiments, in the process of constructing the classification decision module, in addition to using the deep learning model, an integrated learning method (such as Boosting, Bagging, etc.) can also be used to combine multiple weak classifiers into a strong classifier. In addition, the problem of insufficient labeled data can also be addressed by using new learning methods such as semi-supervised learning and unsupervised learning, and this application does not make too many restrictions on this.

[0081] In addition to the above method, in some other embodiments of the present application, multi-sensor fusion technology can also be used, such as by combining multiple sensor technologies such as spectral imaging, infrared imaging, and acoustic detection to build a non-destructive testing system with multi-modal data fusion. By comprehensively utilizing the advantages of different sensors, a comprehensive and accurate evaluation of the quality of preserved eggs can be achieved; the non-destructive testing system can also be combined with the Internet of Things technology to achieve remote monitoring and real-time data transmission. By deploying sensor nodes and data acquisition terminals on the production line, the quality test results of preserved eggs can be uploaded to the cloud server in real time for analysis and processing. At the same time, remote access and query services are provided to users through mobile phone APP or web pages; the non-destructive testing system can also be integrated into an intelligent production line to achieve a deep integration of automated production and preserved egg quality control. Through linkage control with other production line equipment (such as automatic sorting machines, packaging machines, etc.), online detection of preserved egg quality, automatic sorting and packaging of preserved eggs, and other operations can be achieved.

[0082] Based on the technical solution proposed in this application, a machine learning model is designed and optimized for the task of preserved egg quality detection. The model integrates technologies such as deep separable convolutional network, feature pyramid network (FPN) and image attention neural network (SE-Block), effectively reducing model parameters and calculation amount, while greatly improving the accuracy of feature extraction and classification. This model optimization is a key embodiment of technological innovation and provides a more efficient and accurate solution for preserved egg quality detection. With the help of the feature pyramid network structure, the technical solution proposed in this application realizes the fusion of multi-scale features of preserved egg images. This technical improvement enables the machine learning model to capture the local detail features and global features of the preserved egg image at the same time, thereby comprehensively improving the machine learning model's ability to detect the quality of preserved eggs. The fusion of multi-scale features of the preserved egg image not only enhances the robustness of the model, but also further ensures the accuracy of the detection results. The technical solution proposed in this application also realizes the comprehensive automation and intelligence of preserved egg quality detection. Automated detection significantly reduces manual intervention and improves detection efficiency; while intelligent detection automatically analyzes the features of the preserved egg image through a machine learning model, significantly improving the accuracy of detection. This technological innovation makes preserved egg quality testing more efficient and convenient, and provides solid technical support for real-time preserved egg quality monitoring on the production line.

[0083] Based on the same inventive concept, an embodiment of the present application provides a computer-readable storage medium, in which at least one computer program instruction is stored. The at least one computer program instruction is loaded and executed by a processor to implement the operations performed by the method described above.

[0084] Based on the same inventive concept, an embodiment of the present application provides a computer program product, which includes computer instructions, which are stored in a computer-readable storage medium and are suitable for being read and executed by a processor, so that a computer device with the processor executes to implement the operations performed by the method described above.

[0085] Based on the same inventive concept, the present application also provides a preserved egg quality detection device, referring to Figure 5 , showing a schematic diagram of the structure of the preserved egg quality detection device in an embodiment of the present application, the preserved egg quality detection device includes one or more memories 504, one or more processors 502 and at least one computer program (computer program instruction) stored in the memory 504 and executable on the processor 502, and the processor 502 implements the method described above when executing the computer program.

[0086] Among them, Figure 5In the embodiment of the present invention, a bus architecture (represented by bus 500) is shown, which may include any number of interconnected buses and bridges, and bus 500 links various circuits including one or more processors represented by processor 502 and memory represented by memory 504. Bus 500 may also link various other circuits such as peripherals, voltage regulators, and power management circuits, which are well known in the art and are therefore not further described herein. Bus interface 505 provides an interface between bus 500 and receiver 501 and transmitter 503. Receiver 501 and transmitter 503 may be the same element, namely a transceiver, providing a unit for communicating with various other devices over a transmission medium. Processor 502 is responsible for managing bus 500 and general processing, while memory 504 may be used to store data used by processor 502 when performing operations.

[0087] The functions described herein may be implemented in hardware, software executed by a processor, firmware, or any combination thereof. If implemented in software executed by a processor, the functions may be stored as one or more instructions or codes on a computer-readable medium or transmitted via a computer-readable medium. Other examples and implementations are within the scope and spirit of the present application and the appended claims. For example, due to the nature of software, the functions described above may be implemented using software executed by a processor, hardware, firmware, hard wiring, or a combination of any of these. In addition, each functional unit may be integrated into a processing unit, each unit may exist physically separately, or two or more units may be integrated into one unit.

[0088] In the several embodiments provided in this application, it should be understood that the disclosed technical content can be implemented in other ways. Among them, the device embodiments described above are only schematic. For example, the division of the units can be a logical function division. There may be other division methods in actual implementation. For example, multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of units or modules, which can be electrical or other forms.

[0089] The units described as separate components may or may not be physically separated, and the components of the control device may or may not be physical units, that is, they may be located in one place or distributed in multiple units. Some or all of the units may be selected according to actual needs to achieve the purpose of the present embodiment.

[0090] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly or all or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium, including several instructions for a computer device (which can be a personal computer, a server or a network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present application. The aforementioned storage medium includes: U disk, read-only memory (ROM, Read-Only Memory), random access memory (RAM, Random Access Memory), mobile hard disk, disk or optical disk, etc., which can store computer program instructions.

[0091] The above description is only an embodiment of the present application and is not intended to limit the present application. For those skilled in the art, the present application may have various modifications and variations. Any modification, equivalent substitution, improvement, etc. made within the spirit and principle of the present application shall be included in the scope of the claims of the present application.

Claims

1. A method for detecting the quality of preserved eggs, characterized in that: The method comprises: Obtain a pre-trained machine learning model, wherein the machine learning model integrates a deep separable convolutional network, a feature pyramid network, and an image attention neural network; Acquire a preserved egg image of the preserved egg to be detected, and perform feature extraction on the preserved egg image based on the machine learning model to obtain a preserved egg feature map of the preserved egg to be detected; Based on the preserved egg characteristic graph, the target quality type of the preserved egg to be tested is determined.

2. The method according to claim 1, characterized in that The machine learning model is trained through the following steps: Acquire preserved egg sample images of different quality types, wherein the preserved egg sample images are annotated with corresponding quality type labels; Based on the preserved egg sample images of different quality types, a pre-built machine learning model framework is trained to obtain the machine learning model.

3. The method according to claim 2, characterized in that The method further comprises: The machine learning model is quantized and pruned.

4. The method according to claim 1, characterized in that: Before extracting features from the preserved egg image based on the machine learning model, the method further includes: Performing histogram equalization processing and contrast adjustment processing on the preserved egg image; De-noising the preserved egg image by using a Gaussian filter algorithm and a median filter algorithm; The preserved egg image is standardized to a set image size.

5. The method according to claim 1, characterized in that Determining the target quality type of the preserved egg to be tested based on the preserved egg characteristic graph includes: Based on the global average pooling algorithm, the preserved egg feature map is converted into a feature vector of a set length; Based on the feature vector, the probability distribution of the preserved egg to be tested belonging to each quality type is determined by a softmax classifier; The quality type with the highest probability is determined as the target quality type of the preserved egg to be tested.

6. A computer program product, characterized in that The computer program product includes computer instructions, which are stored in a computer-readable storage medium and are suitable for being read and executed by a processor, so that a computer device having the processor executes the method according to any one of claims 1 to 5.

7. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores at least one program code, and the at least one program code is loaded and executed by a processor to implement the operations performed by the method according to any one of claims 1 to 5.

8. A preserved egg quality detection device, characterized in that: The preserved egg quality detection device includes one or more processors and one or more memories, wherein at least one program code is stored in the one or more memories, and the at least one program code is loaded and executed by the one or more processors to implement the method according to any one of claims 1 to 5.

9. A preserved egg screening system, characterized in that: The preserved egg screening system comprises: The preserved egg quality detection device according to claim 8, used to perform the method according to any one of claims 1 to 5; A conveying component, the conveying component is used to convey the preserved eggs to be inspected; A preserved egg image acquisition component, which is disposed on the conveying component and is used to acquire a preserved egg image of a preserved egg to be inspected placed on the conveying component, and send the preserved egg image to the preserved egg quality inspection device; A flap assembly is arranged on the conveying assembly and is used to remove the preserved eggs to be inspected of a preset quality type from the conveying assembly according to the inspection results of the preserved egg quality inspection device.

10. The system according to claim 9, characterized in that The preserved egg image acquisition component includes a darkroom, and a light source and an image acquisition device arranged in the darkroom. The conveying chain of the conveying component passes through the darkroom. The light source is arranged on one side of the conveying chain, and the image acquisition device is arranged on the other side of the conveying chain.

Citation Information

Patent Citations

  • Nondestructive visual inspection grading method for internal quality of preserved eggs

    CN110006899A

  • Image classification method and device

    CN110163301A

  • Image target detection method and system based on convolutional neural network, and readable storage medium

    CN113052006A

  • Preserved egg crack detecting and sorting system and method based on deep learning

    CN114494922A

  • Microcosmic coded image copying detection method and device, mobile terminal and storage medium

    CN117830314A