Duck beak development image recognition method based on infrared thermal imaging technology

Abstract

The invention relates to the technical field of image recognition, and particularly discloses a duck beak development image recognition method based on an infrared thermal imaging technology, and the method comprises the steps: collecting a duck beak infrared image of a target duck object through an infrared imager, carrying out the image noise reduction processing of the duck beak infrared image, and generating a new image sample through a data enhancement technology; according to the method, duck beak infrared images are processed to obtain an enhanced data set of the duck beak infrared images, and then image feature extraction and fine-grained feature enhancement processing are performed on the generated duck beak infrared images through an image processing technology based on deep learning to extract duck beak feature representation of the duck beak infrared images; and carrying out essential feature mining on duck beak feature representation of each duck beak infrared image so as to intelligently identify the development stage of the duck beak. Therefore, the duck beak development state identification accuracy and efficiency can be effectively improved, and an efficient and accurate duck beak development evaluation tool is provided for the poultry breeding industry.

Description

Technical Field

[0001] The present application relates to the field of image recognition technology, and more specifically, to a duck beak development image recognition method based on infrared thermal imaging technology. Background Art

[0002] In modern poultry farming, understanding the growth and development of ducks is essential for optimizing feeding management and improving production efficiency. The development of duck beaks is an important organ for ducks to eat, preen feathers, and interact with the environment. Its development is directly related to the health and productivity of ducks. Traditional methods for evaluating duck beak development usually rely on visual observation or simple measuring tools. This method is not only time-consuming and labor-intensive, but also difficult to achieve accuracy requirements, making it difficult to achieve accurate, rapid, and large-scale testing.

[0003] With the rapid development of computer vision technology, especially the increasing application of infrared thermal imaging technology in the fields of biomedicine and agriculture, new ideas have been provided for the automated and intelligent identification of the development status of duck beaks. Infrared thermal imaging technology can capture the infrared energy radiated from the surface of an object and generate a thermal image that reflects the temperature distribution of the object. This feature enables the technology to non-contactly monitor the physiological state changes of organisms, including temperature differences in tissues, blood circulation, etc., which has potential value in evaluating the growth, inflammation or damage of duck beaks.

[0004] However, in actual application scenarios, infrared images are often affected by differences in shooting angles and the diversity of environmental conditions. When performing image recognition processing on them, the network model may mistakenly learn noise information or irrelevant features in the image, thereby affecting the accuracy of recognition.

[0005] Therefore, an optimized duck beak development image recognition method based on infrared thermal imaging technology is expected. Summary of the invention

[0006] In order to solve the above technical problems, the present application is proposed. An embodiment of the present application provides a method for recognizing duck beak development images based on infrared thermal imaging technology, which uses an infrared imager to collect infrared images of the duck beak of a target duck object, performs image noise reduction processing on the infrared images, and then generates new image samples through data enhancement technology to obtain an enhanced data set of the duck beak infrared image, and then uses deep learning-based image processing technology to perform image feature extraction and fine-grained feature enhancement processing on each generated duck beak infrared image to extract the duck beak feature representation of each duck beak infrared image, and then, by mining the essential features of the duck beak feature representation of each duck beak infrared image, the development stage of the duck beak can be intelligently identified. In this way, the accuracy and efficiency of identifying the developmental status of the duck beak can be effectively improved, and an efficient and accurate duck beak development assessment tool can be provided for the poultry farming industry.

[0007] According to one aspect of the present application, a method for recognizing duck beak development images based on infrared thermal imaging technology is provided, which comprises:

[0008] An infrared imager is used to collect an infrared image of the duck beak of the target duck;

[0009] Performing image denoising and data enhancement processing on the duck beak infrared image to obtain an enhanced data set of the duck beak infrared image;

[0010] Extracting duck beak features of each duck beak infrared image in the enhanced data set of duck beak infrared images respectively to obtain a set of semantically encoded feature maps of duck beak feature infrared images;

[0011] Performing a hierarchical enhancement process based on feature significance on each of the duck beak feature infrared image semantic coding feature maps in the set of duck beak feature infrared image semantic coding feature maps to obtain a set of duck beak feature infrared image enhanced semantic coding feature maps;

[0012] Based on the core significant features of the set of semantic coding feature maps of the duck beak characteristic infrared image enhancement, the recognition result of the duck beak development stage is determined.

[0013] Compared with the prior art, the duck beak development image recognition method based on infrared thermal imaging technology provided by the present application uses an infrared imager to collect the infrared image of the duck beak of the target duck object, performs image noise reduction processing on the image, and then generates a new image sample through data enhancement technology to obtain an enhanced data set of the duck beak infrared image, and then uses deep learning-based image processing technology to extract image features and fine-grained feature enhancement processing on each generated duck beak infrared image to extract the duck beak feature representation of each duck beak infrared image, and then, by mining the essential features of the duck beak feature representation of each duck beak infrared image, the development stage of the duck beak can be intelligently identified. In this way, the accuracy and efficiency of duck beak development status recognition can be effectively improved, and an efficient and accurate duck beak development assessment tool can be provided for the poultry farming industry. BRIEF DESCRIPTION OF THE DRAWINGS

[0014] By describing the embodiments of the present application in more detail in conjunction with the accompanying drawings, the above and other purposes, features and advantages of the present application will become more apparent. The accompanying drawings are used to provide a further understanding of the embodiments of the present application and constitute a part of the specification. Together with the embodiments of the present application, they are used to explain the present application and do not constitute a limitation of the present application. In the accompanying drawings, the same reference numerals generally represent the same components or steps.

[0015] Figure 1 The present invention is a flowchart of a duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application.

[0016] Figure 2 Schematic diagram of data flow of a duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application.

[0017] Figure 3 This is a flowchart of sub-step S4 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application.

[0018] Figure 4 This is a flowchart of sub-step S41 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application.

[0019] Figure 5 This is a flowchart of sub-step S42 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application.

[0020] Figure 6 This is a flowchart of sub-step S5 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application.

[0021] Figure 7 This is a flowchart of sub-step S51 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application. DETAILED DESCRIPTION

[0022] As shown in this application and claims, unless the context clearly indicates an exception, the words "a", "an", "an" and / or "the" do not refer to the singular and may also include the plural. Generally speaking, the terms "include" and "comprise" only indicate the inclusion of the steps and elements that have been clearly identified, and these steps and elements do not constitute an exclusive list. The method or device may also include other steps or elements.

[0023] Although the present application makes various references to certain modules in the system according to the embodiments of the present application, any number of different modules can be used and run on the user terminal and / or server. The modules are only illustrative, and different aspects of the system and method can use different modules.

[0024] Flowcharts are used in the present application to illustrate the operations performed by the system according to the embodiments of the present application. It should be understood that the preceding or following operations are not necessarily performed accurately in order. On the contrary, various steps may be processed in reverse order or simultaneously as required. Meanwhile, other operations may also be added to these processes, or a certain step or several steps of operations may be removed from these processes.

[0025] Below, the exemplary embodiments according to the present application will be described in detail with reference to the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present application, rather than all the embodiments of the present application, and it should be understood that the present application is not limited to the exemplary embodiments described here.

[0026] It is worth noting that in this application, all actions to obtain data are carried out in compliance with the relevant data protection laws and policies of the country where the data is located, and with the authorization given by the owner of the corresponding device.

[0027] In response to the technical problems described in the above background technology, the present application proposes an optimized duck beak development image recognition method based on infrared thermal imaging technology, which uses an infrared imager to collect the infrared image of the duck beak of the target duck object, performs image noise reduction processing on it, and then generates new image samples through data enhancement technology to obtain an enhanced data set of the duck beak infrared image, and then uses deep learning-based image processing technology to extract image features and fine-grained feature enhancement processing on each generated duck beak infrared image to extract the duck beak feature representation of each duck beak infrared image, and then, by mining the essential features of the duck beak feature representation of each duck beak infrared image, the development stage of the duck beak can be intelligently identified. In this way, the accuracy and efficiency of the recognition of the developmental status of the duck beak can be effectively improved, and an efficient and accurate duck beak development assessment tool can be provided for the poultry farming industry.

[0028] Figure 1 The present invention is a flowchart of a duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application. Figure 2 Schematic diagram of data flow of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application. Figure 1 and Figure 2 As shown, the duck beak development image recognition method based on infrared thermal imaging technology includes the following steps: S1, using an infrared imager to collect an infrared image of a duck beak of a target duck object; S2, performing image denoising and data enhancement processing on the duck beak infrared image to obtain an enhanced data set of the duck beak infrared image; S3, respectively extracting duck beak features of each duck beak infrared image in the enhanced data set of the duck beak infrared image to obtain a set of semantic coding feature maps of duck beak feature infrared images; S4, respectively performing hierarchical enhancement processing based on feature significance on each duck beak feature infrared image semantic coding feature map in the set of duck beak feature infrared image semantic coding feature maps to obtain a set of enhanced semantic coding feature maps of duck beak feature infrared images; S5, determining the recognition result of the duck beak development stage based on the core significant features of the set of enhanced semantic coding feature maps of the duck beak feature infrared image.

[0029] In the above-mentioned duck beak development image recognition method based on infrared thermal imaging technology, the step S1 uses an infrared imager to collect an infrared image of the duck beak of the target duck object. It should be understood that infrared thermal imaging technology is not limited by light conditions and can generate a temperature distribution map by detecting the infrared radiation emitted by the object. Therefore, it is very suitable for obtaining thermal imaging images of parts such as duck beaks. In the technical solution of the present application, by collecting the infrared image of the duck beak of the target duck object, non-contact monitoring of the development status of the duck beak can be achieved, providing effective raw data for the identification of duck beak development.

[0030] In order to ensure the professionalism and accuracy of infrared image acquisition, it is necessary to first select a suitable infrared imager. An ideal infrared imager should have high resolution to capture subtle temperature changes and structural features, and should be sensitive enough to detect small temperature differences, and cover a wide temperature measurement range to adapt to different temperature conditions. In addition, the imager should have a fast response time, portability, ease of operation, and a good level of protection to facilitate use in the complex environment of the farm. After selecting the equipment, calibration is required to ensure the consistency and accuracy of the measurement results, and familiarize yourself with the environmental conditions in which you will work, including parameters such as temperature and humidity, and adjust the equipment settings if necessary to compensate for environmental interference. The safety and protection level of the equipment must also be considered to cope with various challenges in the farm.

[0031] Before officially starting the acquisition, you need to prepare the working environment thoroughly. Block background heat sources to prevent the heat from other high-temperature objects in the surrounding environment, such as heaters and lights, from confusing the actual temperature distribution of the duck's beak. Try to shoot in an environment with a relatively stable temperature, such as adjusting the indoor temperature through an air-conditioning system, or choose a time of day when the temperature fluctuates less. Considering that wind speed may change the surface temperature of the duck's beak, it is best to work in windless or light wind conditions; you can also build a simple barrier to block the airflow. These measures help reduce the impact of external factors on image quality.

[0032] In order to obtain high-quality infrared images of duck beaks, the choice of shooting angle and distance is particularly critical. It is recommended to approach the duck beak from the front as much as possible to obtain the most direct view, which is helpful for subsequent analysis. Adjust the position of the imager according to the height of the duck when it stands, so that the lens is roughly near the horizontal line of the duck beak, and maintain a relatively fixed shooting distance to ensure that the proportions of all images are consistent, which is convenient for comparison between different individuals. At the same time, multiple sampling can increase the reliability and representativeness of the data. Take several more photos for each duck and select the clearest and most representative images for subsequent analysis. In addition, some small objects with known temperatures can be placed around the duck beak as references to help correct the temperature readings in the image. Record the date, time, location, environmental conditions and other information in detail for each shot. These auxiliary data are very important for post-processing.

[0033] Considering the duck’s behavior and health status, keeping the duck in a calm and relaxed state can provide a more natural and realistic temperature distribution. This can be achieved by soothing it with a gentle voice or giving it a short rest. Checking the duck beforehand for any signs of illness, especially those that may cause local temperature abnormalities, such as inflammation or infection, will help explain the special temperature patterns in the image. These details not only improve the quality of the image, but also provide a more accurate data basis for subsequent analysis.

[0034] During the actual acquisition process, attention should also be paid to controlling external variables to minimize the impact of external factors on the image, such as by fixing the shooting position and angle, using blackout cloth or other materials to shield unnecessary heat sources, etc. Ensure that the parameters and conditions used throughout the acquisition process are as consistent as possible, including ambient temperature, humidity, wind speed, and imager settings. Take appropriate safety measures to protect the ducks from being frightened or injured, while also ensuring the safety of the staff.

[0035] In summary, using an infrared imager to collect infrared images of duck beaks is a technically demanding task, involving equipment selection, environmental preparation, shooting parameter setting, image quality optimization, etc. Each link requires careful planning and strict execution to ensure accurate and reliable image data.

[0036] In the above-mentioned duck beak development image recognition method based on infrared thermal imaging technology, the step S2 performs image denoising and data enhancement processing on the duck beak infrared image to obtain an enhanced data set of the duck beak infrared image. It should be understood that the present application takes into account that in the actual acquisition process of infrared images, the acquired infrared images are often affected by factors such as the thermal noise, readout noise, and environmental electromagnetic interference of the sensor itself, and contain noise to varying degrees. Therefore, in order to improve the image quality and enhance the feature visibility, the present application further performs image denoising processing on the duck beak infrared image to remove or weaken unnecessary signal components in the image, so that the image is clearer and the recognizability of the duck beak features is enhanced, thereby obtaining a denoised duck beak infrared image. In an embodiment of the present application, methods such as median filtering, Gaussian filtering, or bilateral filtering can be used to remove random noise in the image. Median filtering is suitable for removing salt and pepper noise and can effectively retain image edge information; Gaussian filtering smoothes the image by weighted averaging and is suitable for reducing Gaussian distributed random noise; bilateral filtering combines the dual weighting of spatial neighborhood pixels and intensity similarity to remove noise and maintain edge details. In specific implementation, the most appropriate filtering method can be selected according to the specific type of noise, or multiple filtering techniques can be combined to achieve the best noise reduction effect. For example, after initially applying Gaussian filtering to remove most of the random noise, bilateral filtering is used to further refine the image edges to ensure the clarity of the duck beak structure features.

[0037] At the same time, considering that in practical applications, the posture, angle, environmental conditions, etc. of the duck will affect the performance of the infrared image, so that overfitting is prone to occur when only relying on the actual collected duck beak infrared image to identify the developmental state of the duck beak. Therefore, in order to improve the adaptability and robustness of the model to different situations, the present application further processes the de-noised duck beak infrared image through data enhancement technology (such as rotation, flipping, scaling, cropping, brightness adjustment, etc.), and generates image samples in various forms of variation from the limited original image to obtain an enhanced data set of the duck beak infrared image. In a specific implementation, random rotation, horizontal or vertical flipping and other methods can be used to simulate the image of the duck in different postures, while scaling and cropping can help simulate images taken at different distances and angles. Brightness adjustment can simulate images under different lighting conditions, thereby enhancing the adaptability of the model to various environmental changes. Specifically, by rotating the image at random angles, the different orientations of the duck's head are simulated, which increases the model's ability to understand multi-view images; by flipping the image horizontally or vertically, the diversity of the training set is expanded, especially for organisms with strong symmetry (such as duck beaks). This method is simple and effective; by adjusting the image size and randomly cropping the area of ​​interest from it, it helps to simulate the effects of shooting at different distances and angles, while ensuring that the focus (i.e., the duck beak) is always in the center of the image; by changing the overall brightness level of the image, the situation of shooting under different lighting conditions is simulated, and the model's adaptability to lighting changes is enhanced. These data enhancement techniques can be automated through programming to ensure that each original image can generate multiple variants, thereby significantly expanding the amount of training data. In addition, more complex transformations, such as affine transformations and color jittering, can be introduced to further enrich the content of the data set. In this way, it helps to improve the model's generalization ability for the characteristics of duck beak infrared images and reduce the risk of overfitting.

[0038] In the above-mentioned duck beak development image recognition method based on infrared thermal imaging technology, the step S3 extracts the duck beak features of each duck beak infrared image in the enhanced data set of the duck beak infrared image to obtain a set of semantic coding feature maps of the duck beak feature infrared image. In a specific example of the present application, the step S3 includes: inputting each duck beak infrared image in the enhanced data set of the duck beak infrared image into a duck beak feature extractor based on a convolutional neural network model to obtain a set of semantic coding feature maps of the duck beak feature infrared image. It should be understood that in order to effectively extract the feature representation of the duck beak from the enhanced data set of the duck beak infrared image, the present application adopts a convolutional neural network (CNN) model with excellent performance in image processing tasks as a duck beak feature extractor, and performs feature extraction on each duck beak infrared image in the enhanced data set of the duck beak infrared image. Specifically, the convolutional neural network (CNN) model gradually extracts the local features of the duck beak infrared image through multiple layers of convolutional layers, and maintains key information while reducing the feature dimension through pooling layers to enhance the feature expression ability, so that it can effectively learn the feature information such as the edge, texture, structure and shape of the duck beak infrared image, and generate a set of semantically encoded feature maps of the duck beak feature infrared image, so as to achieve a deep understanding of the duck beak morphology.

[0039] In the above-mentioned duck beak development image recognition method based on infrared thermal imaging technology, the step S4 performs hierarchical enhancement processing based on feature significance on each duck beak feature infrared image semantic coding feature map in the set of duck beak feature infrared image semantic coding feature maps to obtain a set of duck beak feature infrared image enhanced semantic coding feature maps. It should be understood that considering that the duck beak infrared image usually contains a lot of background information that is irrelevant to the developmental state of the duck beak, it may interfere with the model's recognition of duck beak features. Therefore, the present application proposes a feature hierarchical enhancement method based on feature significance, which performs pixel-level feature significance measurement on each duck beak feature infrared image semantic coding feature map in the set of duck beak feature infrared image semantic coding feature maps to distinguish between significant features related to the developmental state of the duck beak and background noise, thereby achieving focused enhancement of the duck beak features and obtaining a set of duck beak feature infrared image enhanced semantic coding feature maps. Among them, Figure 3 FIG. 4 is a flowchart of sub-step S4 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application. Figure 3As shown, the step S4 includes the steps of: S41, performing pixel-based semantic saliency measurement on the semantic coding feature map of the duck beak feature infrared image to obtain a fine-grained semantic saliency distribution matrix of the duck beak feature infrared image; S42, based on the fine-grained semantic saliency distribution matrix of the duck beak feature infrared image, performing feature-explicit hierarchical modulation on the semantic coding feature map of the duck beak feature infrared image to obtain an enhanced semantic coding feature map of the duck beak feature infrared image.

[0040] Figure 4 FIG. 4 is a flowchart of sub-step S41 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application. Figure 4 As shown, the step S41 includes the steps of: S411, performing pixel-granularity feature decoupling on the semantic encoding feature map of the duck beak feature infrared image to obtain a set of pixel-granularity local decoupling feature vectors of the duck beak feature infrared image; S412, inputting each duck beak feature infrared image pixel-granularity local decoupling feature vector in the set of the duck beak feature infrared image pixel-granularity local decoupling feature vectors into a hyperbolic space mapper to obtain a set of hyperbolic space modulated duck beak feature infrared image pixel-granularity local decoupling feature vectors; S413, calculating the fine-grained semantic saliency measurement coefficient of each hyperbolic space modulated duck beak feature infrared image pixel-granularity local decoupling feature vector in the set of the hyperbolic space modulated duck beak feature infrared image pixel-granularity local decoupling feature vectors to obtain the fine-grained semantic saliency distribution matrix of the duck beak feature infrared image formed by arranging multiple duck beak feature infrared image fine-grained semantic saliency measurement coefficients.

[0041] More specifically, the step S411 is expressed as follows:

[0042] decouple(F i )={f1,f2,...,f i ,...,f m}

[0043] Among them, F i represents the semantic coding feature map of the infrared image of the duck beak feature, decouple(·) represents the pixel granularity feature decoupling, f1, f2, f i and f m They respectively represent the first, second, i-th and m-th duck beak feature infrared image pixel granularity local decoupling feature vectors, and m is the number of feature vectors in the set of duck beak feature infrared image pixel granularity local decoupling feature vectors.

[0044] That is, by performing a pixel-granularity feature decoupling operation on the semantic coding feature map of the duck beak feature infrared image, it is decomposed into a series of independent duck beak feature infrared image pixel-granularity local decoupled feature vectors to eliminate unnecessary coupling relationships between pixels and improve the ability to understand the features of each pixel position.

[0045] More specifically, the step S412 is expressed as follows:

[0046] h i =W1f i W2

[0047] Where W1 and W2 represent the first weight parameter matrix and the second weight parameter matrix respectively, h i represents the pixel-granular local decoupling feature vector of the i-th hyperbolic spatially modulated duck beak feature infrared image.

[0048] Specifically, considering that hyperbolic geometric space has better adaptability than Euclidean space when processing hierarchical data sets, the present application further sends the obtained set of local decoupled feature vectors of duck beak feature infrared image pixel granularity into a hyperbolic space mapper for domain transformation, that is, mapping from Euclidean geometric space to hyperbolic geometric space, so as to utilize the characteristics of hyperbolic geometric space to better reveal the hierarchical structure and intrinsic correlation of duck beak features.

[0049] More specifically, the step S413 is expressed by the formula:

[0050]

[0051] in,‖·‖ 2 represents the square of the norm of the eigenvector, represents the logarithmic function with base 2, s i Represents the fine-grained semantic saliency coefficient of the infrared image of the i-th duck beak feature.

[0052] That is, in the hyperbolic space, by defining a metric function to quantitatively evaluate the semantic significance of each pixel-scale feature, and by arranging the obtained semantic significance metric coefficients into a matrix form according to the corresponding pixel space positions, the fine-grained semantic significance distribution matrix of the duck beak feature infrared image is obtained, so as to reveal the overall semantic significance distribution of the semantic encoding feature map of the duck beak feature infrared image, and intuitively show the semantic significance differences of the duck beak feature infrared image at different pixel positions, thereby providing an important basis for subsequent feature selection.

[0053] Figure 5 FIG. 4 is a flowchart of sub-step S42 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application. Figure 5As shown, the step S42 includes the steps of: S421, inputting the fine-grained semantic significance distribution matrix of the duck beak feature infrared image into a feature explicit hierarchical precipitation module based on a multi-layer mask function to obtain a fine-grained semantic significance distribution weight matrix of the duck beak feature infrared image; S422, calculating the position point multiplication of the fine-grained semantic significance distribution weight matrix of the duck beak feature infrared image and each feature matrix along the channel dimension of the semantic coding feature map of the duck beak feature infrared image to obtain the enhanced semantic coding feature map of the duck beak feature infrared image.

[0054] More specifically, the step S421 is expressed by the formula:

[0055]

[0056] Where, mask(·) represents the multi-layer mask function, δ represents the mask threshold, S i ' represents the weight value of the ith position in the fine-grained semantic saliency distribution weight matrix of the duck beak feature infrared image.

[0057] That is, a multi-layer mask function is used to perform layered precipitation on the fine-grained semantic significance distribution matrix of the duck beak feature infrared image. By applying different masks to features at different levels, selective amplification and suppression of features at different pixel positions in the semantic encoding feature map of the duck beak feature infrared image is achieved, and a corresponding weight matrix is ​​generated.

[0058] More specifically, the step S422 is expressed by the formula:

[0059] F f =F i ⊙S'

[0060] Among them, ⊙ represents the point product by position, S' represents the fine-grained semantic saliency distribution weight matrix of the infrared image of duck beak features, and F f Represents the infrared image enhanced semantic coding feature map of the duck beak feature.

[0061] That is, by performing position-weighted operations on each feature matrix along the channel dimension of the original semantically encoded feature map of the duck beak feature infrared image, weighted adjustment of the original features is achieved, thereby effectively enhancing the significant features related to the developmental state of the duck beak, while suppressing irrelevant background noise, providing a more accurate feature representation for the final identification of the duck beak development stage.

[0062] In the above-mentioned duck beak development image recognition method based on infrared thermal imaging technology, the step S5 determines the recognition result of the duck beak development stage based on the core significant features of the set of semantic coding feature maps of the duck beak feature infrared image enhancement. Figure 6FIG. 5 is a flowchart of sub-step S5 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application. Figure 6 As shown, the step S5 includes the steps of: S51, extracting the core significant features of the set of the infrared image enhanced semantic coding feature maps of the duck beak features to obtain a duck beak feature significant clustering enhanced representation vector; S52, inputting the duck beak feature significant clustering enhanced representation vector into a classifier-based developmental stage recognition module to obtain the recognition result, and the recognition result is used to represent the developmental stage label of the duck beak.

[0063] Specifically, the step S51 extracts the core significant features of the set of enhanced semantic coding feature maps of the duck beak feature infrared image to obtain a duck beak feature significant clustering enhanced representation vector. It should be understood that, considering that the enhanced dataset of the duck beak infrared image is a new sample generated by making various changes to the original duck beak infrared image, although it can help the model learn more diverse representations, additional noise or non-essential changes are usually introduced in the data enhancement process. Therefore, in order to ensure that the model can focus on learning the core attributes of the duck beak development state, the present application further performs potential energy analysis and significance aggregation processing on the set of enhanced semantic coding feature maps of the duck beak feature infrared image, so as to deeply mine the essential characteristics of the duck beak development state from the feature set, thereby obtaining a duck beak feature significant clustering enhanced representation vector, so as to improve the accuracy of subsequent developmental stage identification. Among them, Figure 7 FIG. 5 is a flowchart of sub-step S51 of the duck beak development image recognition method based on infrared thermal imaging technology according to an embodiment of the present application. Figure 7 As shown, the step S51 includes the steps of: S511, flattening each duck beak feature infrared image enhancement semantic coding feature map in the set of duck beak feature infrared image enhancement semantic coding feature maps to obtain a set of duck beak feature infrared image enhancement semantic coding feature vectors; S512, calculating the duck beak feature infrared image pseudo-anchor semantic aggregation center representation vector based on the characteristic static potential energy distribution of the set of duck beak feature infrared image enhancement semantic coding feature vectors; S513, calculating the aggregation movement direction of each duck beak feature infrared image enhancement semantic coding feature vector in the set of duck beak feature infrared image enhancement semantic coding feature vectors relative to the duck beak feature infrared image pseudo-anchor semantic aggregation center representation vector to obtain a set of duck beak feature infrared image aggregation movement directions; S514, based on the set of duck beak feature infrared image aggregation movement directions, dynamically aggregating the set of duck beak feature infrared image enhancement semantic coding feature vectors toward the duck beak feature infrared image pseudo-anchor semantic aggregation center representation vector to obtain the duck beak feature significant clustering enhancement representation vector.

[0064] More specifically, the step S511 is expressed as follows:

[0065] V={v1,v2,...,v i ,...,v n}

[0066] Wherein, V represents the set of semantic encoding feature vectors of the infrared image enhancement of duck beak features, v1, v2, v i and v n They respectively represent the first, second, i-th and n-th duck beak feature infrared image enhanced semantic coding feature vectors in the set of duck beak feature infrared image enhanced semantic coding feature vectors, and n is the number of feature vectors in the set of duck beak feature infrared image enhanced semantic coding feature vectors.

[0067] That is, by flattening each duck beak feature infrared image enhanced semantic coding feature map, it is converted into a duck beak feature infrared image enhanced semantic coding feature vector to facilitate subsequent feature analysis.

[0068] More specifically, in a specific example of the present application, the step S512 includes: first, each duck beak feature infrared image enhanced semantic coding feature vector in the set of duck beak feature infrared image enhanced semantic coding feature vectors is input into the feature static potential energy measurement network to obtain a set of duck beak feature infrared image static potential energy measurement coefficients, which is expressed as follows:

[0069]

[0070] in, represents the eigenvalue of the jth position in the infrared image enhanced semantic encoding feature vector of the i-th duck beak feature, μ i and σ i denote the feature mean and feature variance of the i-th duck beak feature infrared image enhanced semantic coding feature vector, L denotes the feature scale value of the duck beak feature infrared image enhanced semantic coding feature vector, e i is the static potential energy measurement coefficient of the i-th duck beak feature infrared image enhanced semantic encoding feature vector.

[0071] That is, by measuring the potential energy level of the semantic encoding feature vector of each duck beak feature infrared image under static conditions, the importance and influence of each duck beak feature infrared image feature in the global image feature distribution, as well as the intrinsic connection between each duck beak feature infrared image feature, are revealed, thereby obtaining a set of static potential energy measurement coefficients of the duck beak feature infrared image.

[0072] Then, the set of duck beak characteristic infrared image static potential energy measurement coefficients is subjected to characteristic energy level gating screening to obtain a set of duck beak characteristic infrared image static potential energy weight factors, which is expressed as follows:

[0073]

[0074] Among them, w i represents the static potential energy measurement coefficient of the ith normalized duck beak feature infrared image, mask(·) represents the gated mask function, represents the gate threshold, w si Represents the static potential energy weight factor of the i-th duck beak feature infrared image.

[0075] That is, based on the gating mechanism, the set of static potential energy measurement coefficients of the obtained duck beak feature infrared image is gated and masked to screen out image features that contribute more to the characterization of the duck beak development stage and assign them corresponding weight factors, while suppressing image features with smaller contributions to reduce the influence of irrelevant information on the identification of the final development stage.

[0076] Next, the set of static potential energy weight factors of the duck beak feature infrared image is used as the weight distribution, and the weighted sum of the set of enhanced semantic coding feature vectors of the duck beak feature infrared image is calculated to obtain the pseudo-anchor semantic aggregation center representation vector of the duck beak feature infrared image, which is expressed as follows:

[0077]

[0078] Among them, v a Representation vector of pseudo-anchor semantic aggregation center of duck beak feature infrared image.

[0079] That is, based on the obtained weight factor, the set of enhanced semantic encoding feature vectors of the duck beak feature infrared image is weightedly aggregated to reveal the feature distribution trend center of the enhanced data set of the duck beak infrared image, and generate a pseudo-anchored semantic aggregation center representation vector of the duck beak feature infrared image, thereby providing effective guidance for the subsequent aggregation process of the duck beak infrared image features.

[0080] More specifically, the step S513 calculates the aggregated motion direction of each duck beak feature infrared image enhanced semantic coding feature vector in the set of duck beak feature infrared image enhanced semantic coding feature vectors relative to the duck beak feature infrared image pseudo anchor semantic aggregation center representation vector to obtain a set of duck beak feature infrared image aggregated motion directions. The duck beak feature infrared image aggregated motion direction is the arc cosine function value between the duck beak feature infrared image enhanced semantic coding feature vector and the duck beak feature infrared image pseudo anchor semantic aggregation center representation vector, which is expressed by the formula:

[0081]

[0082] Among them, ‖·‖ represents the modulus of the vector, arccos(·) is the inverse cosine function, θ i Represents the aggregation movement direction of the i-th duck beak feature infrared image enhanced semantic encoding feature vector relative to the duck beak feature infrared image pseudo-anchor semantic aggregation center representation vector.

[0083] That is, by calculating the aggregation movement direction of each duck beak feature infrared image enhanced semantic coding feature vector relative to the duck beak feature infrared image pseudo-anchor semantic aggregation center representation vector, the degree of change of each image feature relative to the center point is quantified, thereby revealing the relative relationship between each duck beak image feature and the group behavior pattern.

[0084] More specifically, the step S514 is expressed as follows:

[0085]

[0086] Where sin(·) represents the sine function, W i and b i Denote the weight matrix and the bias term respectively, and X denotes the significant clustering enhancement representation vector of the duck beak feature.

[0087] That is, based on the aggregation movement direction and feature static potential energy measurement coefficient of each duck beak feature infrared image enhanced semantic coding feature vector, each duck beak feature infrared image enhanced semantic coding feature vector is dynamically aggregated toward the duck beak feature infrared image pseudo-anchored semantic aggregation center representation vector to obtain the duck beak feature significant clustering enhanced representation vector, thereby effectively mining the core significant features of the duck beak development state and achieving an in-depth understanding and accurate representation of the duck beak development state.

[0088] Specifically, in step S52, the duck beak feature significant clustering enhanced representation vector is input into a developmental stage recognition module based on a classifier to obtain the recognition result, and the recognition result is used to represent the developmental stage label of the duck beak. Specifically, the classifier adopts a multi-layer perceptron structure, and through layer-by-layer processing and feature abstraction of the fully connected layer, it can extract high-level features related to duck beak development from the duck beak feature significant clustering enhanced representation vector, and through the softmax function of the output layer, the extracted high-level features are mapped into a probability distribution for representing the developmental state of the duck beak belonging to each developmental stage, thereby determining the final output result, that is, the developmental stage label, such as "duckling period", "fattening period", "late fattening period", etc.

[0089] In a preferred example of the present application, inputting the duck beak feature significant clustering reinforcement characterization vector into a classifier-based developmental stage recognition module to obtain a recognition result includes:

[0090] First, the distance between each pair of eigenvalues ​​of the duck beak feature significant clustering reinforcement characterization vector is calculated, such as the L2 distance, and the square root of the distance is taken to obtain the duck beak feature significant clustering reinforcement distance representation matrix, which is expressed as follows:

[0091]

[0092] Among them, D i,j Indicates the significant clustering reinforcement distance of duck beak features. The element value at position (i, j) in the distance matrix is ​​represented by x i 、x j They represent the eigenvalues ​​of the i-th and j-th positions in the duck beak feature significant clustering and reinforcement representation vector, d(·,·) represents the distance metric function, and X represents the duck beak feature significant clustering and reinforcement representation vector;

[0093] Secondly, the duck beak feature significant clustering reinforcement autocorrelation matrix of the duck beak feature significant clustering reinforcement representation vector is obtained, which is expressed as follows:

[0094]

[0095] Where D′ represents the duck beak feature significant clustering reinforcement autocorrelation matrix, and X is the row vector, (·) T represents the transpose of a vector, Represents matrix multiplication;

[0096] Next, after matrix multiplication of the duck beak feature significant clustering reinforcement characterization vector and the duck beak feature significant clustering reinforcement distance representation matrix, it is further matrix multiplied with the matrix product of the duck beak feature significant clustering reinforcement distance representation matrix and the duck beak feature significant clustering reinforcement self-correlation matrix to obtain the duck beak feature significant clustering reinforcement multi-level mapping vector, which is expressed as follows:

[0097]

[0098] Where D″ represents the duck beak feature significant clustering reinforcement multi-level mapping vector, and D represents the duck beak feature significant clustering reinforcement distance representation matrix;

[0099] Then, the duck beak feature significant clustering reinforcement multi-level mapping vector is interpolated with the duck beak feature significant clustering reinforcement associated eigenvector (if the eigenvalue is insufficient, interpolation or zero filling) composed of the eigenvalue of the duck beak feature significant clustering reinforcement auto-association matrix to obtain an optimized duck beak feature significant clustering reinforcement representation vector;

[0100] Finally, the optimized duck beak feature significant clustering reinforcement representation vector is input into a classifier-based developmental stage recognition module to obtain a recognition result.

[0101] Specifically, considering that each duck beak feature infrared image enhancement semantic coding feature map in the set of duck beak feature infrared image enhancement semantic coding feature maps respectively represents the explicit enhancement coding features of the image semantic space dimension of each duck beak infrared image in the enhanced data set of duck beak infrared images, when performing feature aggregation based on feature dynamics analysis, the insufficient feature dynamic correspondence of each local image semantic feature will also lead to the lack of decision-making of the aggregated feature instances of the duck beak feature significant clustering enhancement representation vector, thereby affecting the accuracy of the recognition result obtained by its input into the developmental stage recognition module based on the classifier.

[0102] Therefore, the present application uses a linear target mapping representation based on the similarity distance representation matrix of the duck's beak feature significantly clustered and enhanced representation vector to perform a secondary target mapping representation based on a multi-level distribution hierarchy for the complete similarity instantiation of the self-correlation of the duck's beak feature significantly clustered and enhanced representation vector, and compensates for the negative impact factor of association mismatch through an associated fusion kernel bias to improve the degree of instance determination of the eigenvalues ​​of the duck's beak feature significantly clustered and enhanced representation vector under similarity constraints, that is, the degree of significance of the eigenvalues ​​as instances for generative regression determination, thereby improving the accuracy of the recognition results obtained by inputting the duck's beak feature significantly clustered and enhanced representation vector into a developmental stage recognition module based on a classifier.

[0103] After the duck beak development stage is determined, the development stage tag information of each duck will be entered into the farm's database. This includes not only the duck's identification (such as ear tag number), sampling date and growth environment parameters (temperature, humidity, etc.), but also weight change trends and other health indicators. These data are then displayed intuitively in the form of charts or dashboards to help managers quickly understand the development of groups and individual ducks. For example, a time series graph can show the average development progress of a group over time, or a pie chart can be used to compare the proportional distribution of different development stages.

[0104] Based on this information, farms can adopt personalized feeding strategies for ducks at different developmental stages. For ducklings, providing high-protein, easily digestible food to promote healthy growth is the key; after entering the fattening period, gradually increase the proportion of feed with higher energy density, while paying attention to the supplementation of minerals and vitamins to support muscle growth; in the later stage of fattening, it is particularly important to control the amount of feed to avoid excessive obesity affecting meat quality, and start preparing for the work before marketing. In addition, it is also necessary to adjust the temperature and humidity in the house, the duration of light and other factors according to the needs of different developmental stages, and create a climate conducive to the growth of ducks. For example, provide additional warming measures for ducklings in cold seasons; strengthen ventilation and cooling in summer.

[0105] At the same time, regular inspections of the health status of ducks, especially those in critical developmental periods, can effectively reduce morbidity and mortality by detecting and addressing possible problems early. For example, close attention should be paid to the development of the immune system of ducks that have just entered the fattening period, and vaccinations should be given in a timely manner; in the late fattening period, disease prevention and control efforts should be strengthened to avoid large-scale epidemics caused by stress reactions.

[0106] Farms can also use the duck beak development stage tags to predict the production trend in the future and arrange production activities reasonably. In the short term, determine the number of ducks slaughtered daily or weekly to ensure stable market supply; from a long-term perspective, plan the breeding scale and technical investment for the next cycle in advance based on market demand forecasts. Collect a large number of actual cases for scientific research and analysis to explore more scientific and reasonable breeding models, such as studying the growth differences of different breeds of ducks under the same feeding conditions to provide a basis for genetic improvement; test the effects of new feed formulas or additives to find new ways to improve production performance.

[0107] Finally, it is very important to establish a continuous improvement mechanism. This mechanism not only involves regular review of existing operating procedures and finding potential improvement points, but also encourages front-line employees to put forward improvement suggestions to create a good atmosphere of full participation. At the same time, continue to track the development trend of new technologies and promptly introduce application results suitable for the actual situation of the farm. Through such a circular feedback process, not only can the breeding benefits be significantly improved, but also the health and welfare of ducks can be guaranteed, and the entire industry can be promoted to a more efficient and environmentally friendly direction.

[0108] In summary, the method for recognizing duck beak development images based on infrared thermal imaging technology according to the embodiment of the present application is explained, which uses an infrared imager to collect infrared images of the duck beak of the target duck object, performs image noise reduction processing on the infrared images, and then generates new image samples through data enhancement technology to obtain an enhanced data set of the infrared images of the duck beak, and then uses image processing technology based on deep learning to perform image feature extraction and fine-grained feature enhancement processing on each generated infrared image of the duck beak to extract the feature representation of the duck beak of each infrared image of the duck beak, and then, by mining the essential features of the feature representation of the duck beak of each infrared image of the duck beak, the development stage of the duck beak can be intelligently identified. In this way, the accuracy and efficiency of the recognition of the developmental status of the duck beak can be effectively improved, and an efficient and accurate duck beak development assessment tool can be provided for the poultry farming industry.

[0109] The basic principle of the present invention is described above in conjunction with specific embodiments. However, it should be pointed out that the advantages, strengths, effects, etc. mentioned in the present invention are only examples and not limitations, and it cannot be considered that these advantages, strengths, effects, etc. must be possessed by each embodiment of the present invention. In addition, the specific details of the above embodiments are only for the purpose of illustration and facilitation of understanding, rather than limitation, and the above details do not limit the present invention to being implemented by adopting the above specific details.

[0110] In the above embodiments, the description of each embodiment has its own emphasis. For the parts that are not described or recorded in detail in a certain embodiment, please refer to the relevant description of other embodiments. In the several embodiments provided by the present invention, it should be understood that the disclosed system and method can be implemented in other ways. For example, the system embodiment described above is only schematic. For example, the unit division is only a logical function division, and there may be other division methods in actual implementation. The units described as separate components may or may not be physically separated, and the components displayed as units may or may not be physical units, that is, they may be located in one place, or they may be distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the scheme of this embodiment.

[0111] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and scope of the equivalent elements of the claims be included in the invention. Any reference to a figure in a claim should not be considered as limiting the claim to which it relates.

[0112] In addition, it is obvious that the word "comprising" does not exclude other units or steps, and the singular does not exclude the plural. The multiple units stated in the system claims can also be implemented by one unit through software or hardware.

[0113] Finally, it should be noted that the above description has been given for the purpose of illustration and description. In addition, the above embodiments are only used to illustrate the technical solution of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, it should be understood by those skilled in the art that the technical solution of the present invention can be modified or replaced by equivalents without departing from the spirit and scope of the technical solution of the present invention.

Claims

1. A duck beak development image recognition method based on infrared thermal imaging technology, characterized in that: include: An infrared imager is used to collect an infrared image of the duck beak of the target duck; Performing image denoising and data enhancement processing on the duck beak infrared image to obtain an enhanced data set of the duck beak infrared image; Extracting duck beak features of each duck beak infrared image in the enhanced data set of duck beak infrared images respectively to obtain a set of semantically encoded feature maps of duck beak feature infrared images; Performing a hierarchical enhancement process based on feature significance on each of the duck beak feature infrared image semantic coding feature maps in the set of duck beak feature infrared image semantic coding feature maps to obtain a set of duck beak feature infrared image enhanced semantic coding feature maps; Based on the core significant features of the set of semantic coding feature maps of the duck beak characteristic infrared image enhancement, the recognition result of the duck beak development stage is determined.

2. The duck beak development image recognition method based on infrared thermal imaging technology according to claim 1 is characterized in that: Extracting duck beak features of each duck beak infrared image in the enhanced data set of duck beak infrared images respectively to obtain a set of semantic coding feature maps of duck beak feature infrared images, including: Each duck beak infrared image in the enhanced data set of the duck beak infrared image is respectively input into a duck beak feature extractor based on a convolutional neural network model to obtain a set of semantically encoded feature maps of the duck beak feature infrared images.

3. The duck beak development image recognition method based on infrared thermal imaging technology according to claim 2 is characterized in that: The individual duck beak feature infrared image semantic coding feature maps in the set of duck beak feature infrared image semantic coding feature maps are respectively subjected to hierarchical enhancement processing based on feature significance to obtain a set of duck beak feature infrared image enhanced semantic coding feature maps, including: Performing pixel-granularity-based semantic saliency measurement on the semantic coding feature map of the duck beak feature infrared image to obtain a fine-grained semantic saliency distribution matrix of the duck beak feature infrared image; Based on the fine-grained semantic saliency distribution matrix of the duck beak feature infrared image, the semantic coding feature map of the duck beak feature infrared image is subjected to feature explicit hierarchical modulation to obtain the enhanced semantic coding feature map of the duck beak feature infrared image.

4. The duck beak development image recognition method based on infrared thermal imaging technology according to claim 3 is characterized in that: The semantic coding feature map of the duck beak feature infrared image is subjected to pixel-based semantic saliency measurement to obtain a fine-grained semantic saliency distribution matrix of the duck beak feature infrared image, including: Performing pixel granularity feature decoupling on the semantic coding feature map of the duck beak feature infrared image to obtain a set of pixel granularity local decoupling feature vectors of the duck beak feature infrared image; Inputting each duck beak feature infrared image pixel granularity local decoupling feature vector in the set of duck beak feature infrared image pixel granularity local decoupling feature vectors into a hyperbolic space mapper to obtain a set of hyperbolic space modulated duck beak feature infrared image pixel granularity local decoupling feature vectors; The fine-grained semantic saliency measurement coefficient of each hyperbolic spatial modulated duck beak feature infrared image pixel granularity local decoupling feature vector in the set of the hyperbolic spatial modulated duck beak feature infrared image pixel granularity local decoupling feature vector is calculated to obtain the fine-grained semantic saliency distribution matrix of the duck beak feature infrared image formed by arranging multiple duck beak feature infrared image fine-grained semantic saliency measurement coefficients.

5. The duck beak development image recognition method based on infrared thermal imaging technology according to claim 4 is characterized in that: Based on the fine-grained semantic saliency distribution matrix of the duck beak feature infrared image, the semantic coding feature map of the duck beak feature infrared image is subjected to feature explicit hierarchical modulation to obtain the enhanced semantic coding feature map of the duck beak feature infrared image, including: Inputting the fine-grained semantic saliency distribution matrix of the duck beak feature infrared image into a feature explicit hierarchical precipitation module based on a multi-layer mask function to obtain a fine-grained semantic saliency distribution weight matrix of the duck beak feature infrared image; The fine-grained semantic significance distribution weight matrix of the duck beak feature infrared image is calculated and multiplied by each feature matrix along the channel dimension of the semantic coding feature map of the duck beak feature infrared image at each position point to obtain the enhanced semantic coding feature map of the duck beak feature infrared image.

6. The duck beak development image recognition method based on infrared thermal imaging technology according to claim 5 is characterized in that: Based on the core significant features of the set of the infrared image-enhanced semantic coding feature map of the duck beak feature, the recognition result of the duck beak development stage is determined, including: Extracting the core significant features of the set of the duck beak feature infrared image enhanced semantic coding feature map to obtain a duck beak feature significant clustering enhanced representation vector; The duck beak feature significant clustering reinforcement representation vector is input into a classifier-based developmental stage recognition module to obtain the recognition result, and the recognition result is used to represent the developmental stage label of the duck beak.

7. The duck beak development image recognition method based on infrared thermal imaging technology according to claim 6 is characterized in that: Extracting the core significant features of the set of the infrared image enhanced semantic coding feature map of the duck beak feature to obtain a duck beak feature significant clustering enhanced representation vector, including: Flattening each duck beak feature infrared image enhancement semantic coding feature map in the set of duck beak feature infrared image enhancement semantic coding feature maps to obtain a set of duck beak feature infrared image enhancement semantic coding feature vectors; Based on the characteristic static potential energy distribution of the set of the duck beak characteristic infrared image enhanced semantic coding feature vectors, a pseudo-anchor semantic aggregation center representation vector of the duck beak characteristic infrared image is calculated; Calculating the aggregated motion direction of each duck beak feature infrared image enhanced semantic coding feature vector in the set of duck beak feature infrared image enhanced semantic coding feature vectors relative to the duck beak feature infrared image pseudo-anchor semantic aggregation center representation vector to obtain a set of duck beak feature infrared image aggregated motion directions; Based on the set of aggregation movement directions of the duck beak feature infrared image, the set of enhanced semantic coding feature vectors of the duck beak feature infrared image is dynamically aggregated toward the pseudo-anchored semantic aggregation center representation vector of the duck beak feature infrared image to obtain the duck beak feature significant clustering enhancement representation vector.

8. The duck beak development image recognition method based on infrared thermal imaging technology according to claim 7 is characterized in that: Based on the characteristic static potential energy distribution of the set of the duck beak feature infrared image enhanced semantic coding feature vectors, the duck beak feature infrared image pseudo-anchor semantic aggregation center representation vector is calculated, including: Inputting each duck beak feature infrared image enhanced semantic coding feature vector in the set of duck beak feature infrared image enhanced semantic coding feature vectors into a feature static potential energy measurement network to obtain a set of duck beak feature infrared image static potential energy measurement coefficients; Performing characteristic energy level gating screening on the set of duck beak characteristic infrared image static potential energy measurement coefficients to obtain a set of duck beak characteristic infrared image static potential energy weight factors; Taking the set of static potential energy weight factors of the duck beak feature infrared image as the weight distribution, the weighted sum of the set of enhanced semantic coding feature vectors of the duck beak feature infrared image is calculated to obtain the pseudo-anchor semantic aggregation center representation vector of the duck beak feature infrared image.

9. The duck beak development image recognition method based on infrared thermal imaging technology according to claim 8, characterized in that: The aggregation movement direction of the duck beak feature infrared image is the arccosine function value between the enhanced semantic coding feature vector of the duck beak feature infrared image and the pseudo-anchor semantic aggregation center representation vector of the duck beak feature infrared image.

Images