Component identification method and device, equipment and storage medium
By training a multimodal neural network model and using quadruples of data for component identification and region localization, the problems of low efficiency and poor adaptability in the identification of used mobile phone components are solved, and non-destructive and automated component detection is achieved.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 转转一零二四(北京)科技有限公司
- Filing Date
- 2025-12-29
- Publication Date
- 2026-05-01
AI Technical Summary
In existing technologies, the identification of used mobile phone components is inefficient and costly, and it is difficult to cover all possible component variations. Traditional algorithms are unable to handle the diversity of component shapes, and static feature libraries cannot dynamically adapt to changes in component shapes.
By collecting quadruple data of model-component-region-image, a multimodal neural network model is trained to learn the morphology, position and contextual relationship of different components in the whole machine X-ray image, thereby realizing component name recognition and region localization.
It enables non-destructive and automated component identification and area positioning, improves detection efficiency and accuracy, adapts to changes in component shape, and provides a technical foundation for large-scale multi-model testing.
Smart Images
Figure CN121962034A_ABST
Abstract
Description
Technical Field
[0001] This application relates to the field of image processing technology, and in particular to a method, apparatus, device, and storage medium for identifying electronic components. Background Technology
[0002] With the rapid development of the second-hand electronics market, users' demand for assessing device performance, authenticity, and condition is increasing. However, there is a wide variety of second-hand mobile phone models, and components of the same model may appear in various forms due to differences in suppliers, production batches, or modifications.
[0003] The disassembly process is currently mainly carried out manually, that is, by manually disassembling the equipment, taking X-ray images of the components, and manually labeling the component names, types and morphological characteristics.
[0004] However, this approach is inefficient, costly, and difficult to cover all possible component variations. Summary of the Invention
[0005] This application provides a method, apparatus, device, and storage medium for identifying electronic components, in order to improve the braking safety of electronic devices in emergency situations.
[0006] In a first aspect, embodiments of this application provide a method for identifying electronic components, including:
[0007] Obtain a full-machine X-ray image of the model to be tested;
[0008] The whole machine X-ray image of the model to be tested is input into the multimodal model to obtain the component name and corresponding area of each component in the model to be tested;
[0009] The multimodal model is obtained by training a preset neural network model based on the quadruple data corresponding to each component in multiple models and the whole machine X-ray images corresponding to each model. Each quadruple data includes: component name, the first model corresponding to the component, the X-ray image corresponding to the component, and the region corresponding to the component.
[0010] In one or more embodiments, before inputting the whole-machine X-ray image of the machine model to be tested into the multimodal model to obtain the component names and corresponding regions of each component in the machine model to be tested, the method further includes:
[0011] Obtain the quadruple data corresponding to each component in the multiple models;
[0012] The multimodal model is obtained by training a preset neural network model based on the quadruple data corresponding to each component in the multiple models and the whole machine X-ray images corresponding to the multiple models.
[0013] In one or more embodiments, the method further includes:
[0014] After calibrating the component names and corresponding regions of each component in the model under test, the multimodal model is updated based on the calibrated component names and regions of each component in the model under test.
[0015] In one or more embodiments, the method further includes:
[0016] Based on the component name and corresponding region of each component, the same component is clustered to obtain different regions of different components.
[0017] Based on different regions of different components, the different morphological characteristics of different components are determined.
[0018] In one or more embodiments, the method further includes:
[0019] Obtain the feature vector corresponding to the X-ray image of the component to be inspected;
[0020] Based on the feature vector corresponding to the X-ray image of the component to be detected, the target component to which the component to be detected belongs is determined by the different morphological features of different components.
[0021] Based on the feature vector corresponding to the X-ray image of the component to be detected, the different morphological features of the target component are updated.
[0022] In one or more embodiments, the method further includes:
[0023] If the target component to which the component to be detected belongs is not determined among the different morphological characteristics of different components, the multimodal model is updated based on the X-ray image of the component to be detected.
[0024] Secondly, embodiments of this application provide a component identification device, comprising:
[0025] The acquisition module is used to acquire the full-machine X-ray image of the model to be inspected;
[0026] The processing module is used to input the whole X-ray image of the machine model to be tested into the multimodal model to obtain the component name and corresponding area of each component in the machine model to be tested;
[0027] The multimodal model is obtained by training a preset neural network model based on the quadruple data corresponding to each component in multiple models and the whole machine X-ray images corresponding to each model. Each quadruple data includes: component name, the first model corresponding to the component, the X-ray image corresponding to the component, and the region corresponding to the component.
[0028] In one or more embodiments, before inputting the whole-machine X-ray image of the machine to be inspected into the multimodal model to obtain the component names and corresponding regions of each component in the machine to be inspected, the processing module is further configured to:
[0029] Obtain the quadruple data corresponding to each component in the multiple models;
[0030] The multimodal model is obtained by training a preset neural network model based on the quadruple data corresponding to each component in the multiple models and the whole machine X-ray images corresponding to the multiple models.
[0031] In one or more embodiments, the processing module is further configured to:
[0032] After calibrating the component names and corresponding regions of each component in the model under test, the multimodal model is updated based on the calibrated component names and regions of each component in the model under test.
[0033] In one or more embodiments, the processing module is further configured to:
[0034] Based on the component name and corresponding region of each component, the same component is clustered to obtain different regions of different components.
[0035] Based on different regions of different components, the different morphological characteristics of different components are determined.
[0036] In one or more embodiments, the processing module is further configured to:
[0037] Obtain the feature vector corresponding to the X-ray image of the component to be inspected;
[0038] Based on the feature vector corresponding to the X-ray image of the component to be detected, the target component to which the component to be detected belongs is determined by the different morphological features of different components.
[0039] Based on the feature vector corresponding to the X-ray image of the component to be detected, the different morphological features of the target component are updated.
[0040] In one or more embodiments, the processing module is further configured to:
[0041] If the target component to which the component to be detected belongs is not determined among the different morphological characteristics of different components, the multimodal model is updated based on the X-ray image of the component to be detected.
[0042] Thirdly, embodiments of this application provide an electronic device, including: a memory and a processor;
[0043] The memory stores computer-executed instructions;
[0044] The processor executes computer execution instructions stored in the memory, causing the processor to perform the first aspect and / or various possible implementations of the first aspect as described above.
[0045] Fourthly, embodiments of this application provide a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, are used to implement the first aspect and / or various possible implementations of the first aspect.
[0046] Fifthly, embodiments of this application provide a computer program product, including a computer program that, when executed by a processor, implements the first aspect and / or various possible implementations of the first aspect.
[0047] The component identification method, apparatus, device, and storage medium provided in this application acquire an X-ray image of the entire machine model to be tested; input the X-ray image of the entire machine model to be tested into a multimodal model to obtain the component name and corresponding region of each component in the machine model to be tested; wherein, the multimodal model is obtained by training a preset neural network model based on the four-tuple data corresponding to each component in multiple machine models and the X-ray images of the entire machine models corresponding to multiple machine models, and each four-tuple data includes: component name, the first machine model corresponding to the component, the X-ray image corresponding to the component, and the region corresponding to the component. This technical solution achieves non-invasive and automated identification and region localization of internal components of the entire machine by acquiring the X-ray image of the entire machine model to be tested and inputting it into a multimodal model trained based on the four-tuple data. The multimodal model is trained by fusing quadruple data of machine model, components, regions, and X-ray images. This establishes a multi-dimensional mapping relationship from the macroscopic image of the entire machine to the microscopic features of components, enabling accurate output of the names of each component and their precise locations in the overall machine image without disassembly. This effectively solves the problems of traditional component inspection relying on manual disassembly, which is inefficient and prone to equipment damage. It achieves rapid and non-destructive identification of complex internal structures, significantly improving the automation and accuracy of component inspection. Attached Figure Description
[0048] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.
[0049] Figure 1 Flowchart of the component identification method provided in the embodiments of this application Figure 1 ;
[0050] Figure 2 Schematic diagram of X-ray images of some components provided in the embodiments of this application;
[0051] Figure 3 Flowchart of the component identification method provided in the embodiments of this application Figure 2 ;
[0052] Figure 4 Flowchart of the component identification method provided in the embodiments of this application Figure 3 ;
[0053] Figure 5 A schematic diagram of the structure of the component identification device provided in the embodiments of this application;
[0054] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application.
[0055] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation
[0056] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.
[0057] With the rapid development of the second-hand electronics market, users' demand for assessing device performance, authenticity, and condition is increasing. However, there is a wide variety of second-hand mobile phone models, and components of the same model may appear in various forms due to differences in suppliers, production batches, or modifications.
[0058] In existing technologies, the identification of used mobile phone components and the construction of datasets mainly rely on the following methods:
[0059] 1) Manual disassembly and labeling: Using manual disassembly equipment, X-ray images of components are taken, and the component names, types, and morphological characteristics are manually labeled.
[0060] 2) Current image recognition algorithms classify components based on predefined rules or shallow machine learning models (such as SVM and random forest).
[0061] 3) Static feature library management: By manually maintaining the component feature library, the morphological features of known components are stored as fixed templates.
[0062] However, the existing implementations described above have the following technical problems:
[0063] 1) Low data acquisition efficiency: The manual disassembly and labeling process is cumbersome and difficult to quickly cover a large number of models and component variations.
[0064] 2) Weak model generalization ability: Traditional algorithms have difficulty handling the diversity of component shapes, especially the low accuracy of recognizing new shapes.
[0065] 3) Difficulty in maintaining the feature library: Static feature libraries cannot dynamically adapt to changes in the form of components and require frequent manual updates, resulting in poor system flexibility.
[0066] To address the technical problems existing in the prior art, the inventors of this application propose the following concept: By collecting and associating quadruple data of model-component-region-image, a multimodal neural network model is trained. This model learns the implicit patterns of different components in the shape, position, and contextual relationships of the whole machine X-ray image, thereby automatically completing component name recognition and region localization based solely on the whole machine X-ray image without prior disassembly information. This transforms the traditional disassembly and inspection process, which relies on manual experience, into an automated identification process based on data-driven and algorithmic modeling. This not only solves the technical challenges of non-destructive testing but also provides a scalable technical foundation for large-scale, multi-model intelligent component analysis.
[0067] The technical solution of this application and how it solves the above-mentioned technical problems will be described in detail below with specific embodiments. The subject of this technical solution is an electronic device, such as a control unit in an electronic device.
[0068] The following specific embodiments can be combined with each other, and the same or similar concepts or processes may not be described again in some embodiments. The embodiments of this application will be described below with reference to the accompanying drawings.
[0069] Figure 1 Flowchart of the component identification method provided in the embodiments of this application Figure 1 ,like Figure 1 As shown, the method includes:
[0070] Step 11: Obtain the full-machine X-ray image of the model to be tested;
[0071] In this step, non-destructive X-ray imaging is performed on the device to be inspected to obtain a complete image reflecting its internal structure, that is, to obtain a whole-machine X-ray image.
[0072] In one possible implementation, the model to be inspected is placed in the imaging area of an X-ray device. X-rays penetrate the device and form a grayscale image, in which components of different densities will exhibit different contrasts of brightness and darkness.
[0073] For example, for a mobile phone, X-ray images can clearly show the outline and location of its internal components such as the motherboard, battery, and camera module, without disassembling the casing, thus maintaining the physical integrity of the device and providing standardized input data for subsequent intelligent identification.
[0074] Step 12: Input the whole machine X-ray image of the model to be tested into the multimodal model to obtain the component name and corresponding area of each component in the model to be tested.
[0075] The multimodal model is obtained by training a preset neural network model based on the quadruple data corresponding to each component in multiple models and the whole machine X-ray images corresponding to multiple models. Each quadruple data includes: component name, the first model corresponding to the component, the X-ray image corresponding to the component, and the region corresponding to the component.
[0076] In this step, a trained multimodal model is used to identify and locate components in the input whole-machine X-ray image, and output the component name and corresponding area of each component in the model to be inspected.
[0077] In one possible implementation, the training data for the multimodal model comes from a large amount of previously accumulated quadruple data (e.g., camera module - model 13 - independent component X-ray image - upper left region of the image) and the whole machine X-ray image of the corresponding model, enabling the multimodal model to learn to segment and identify each component from the whole machine image.
[0078] For example, after inputting a full-body X-ray image of a laptop into the model, the model can analyze the image pixel by pixel and output structured information such as: solid-state drive: located in the right-center area of the image; memory modules: located side by side to the left of the solid-state drive.
[0079] For example, Figure 2 This is a schematic diagram of X-ray images of some components provided in the embodiments of this application, such as... Figure 2 As shown, X-ray images of multiple components are presented.
[0080] The component identification method provided in this application involves acquiring an X-ray image of the entire machine of the model to be tested; inputting the X-ray image of the entire machine of the model to be tested into a multimodal model to obtain the component name and corresponding region of each component in the model to be tested; wherein, the multimodal model is obtained by training a preset neural network model based on the four-tuple data corresponding to each component in multiple models and the X-ray images of the entire machine corresponding to multiple models, and each four-tuple data includes: component name, the first model corresponding to the component, the X-ray image corresponding to the component, and the region corresponding to the component. This technical solution achieves non-invasive and automated identification and region localization of internal components of the entire machine by acquiring the X-ray image of the entire machine of the model to be tested and inputting it into a multimodal model trained based on four-tuple data. The multimodal model is trained by fusing the four-tuple data of model-component-region-X-ray image, which can establish a multi-dimensional mapping relationship from the macroscopic image of the entire machine to the microscopic component features, thereby accurately outputting the name of each component and its precise position in the image of the entire machine without disassembling the machine. It effectively solves the problems of traditional component testing relying on manual disassembly, which is inefficient and prone to damaging equipment. It enables rapid and non-destructive identification of the internal structure of complex models, significantly improving the automation level and identification accuracy of component testing.
[0081] Based on the above embodiments, Figure 3 Flowchart of the component identification method provided in the embodiments of this application Figure 2 ,like Figure 3 As shown, the method further includes the following steps before step 12:
[0082] Step 31: Obtain the quadruple data corresponding to each component in multiple models;
[0083] In this step, multiple models are first physically disassembled, each component is separated, and each independent component is X-rayed. At the same time, the model to which the component belongs, the component name, and the location information of the corresponding area in the whole machine X-ray image are recorded, thus forming a four-tuple data entry of component name-model-component X-ray image-area.
[0084] For example, after disassembling a mobile phone, separate X-ray images of its motherboard, battery, and camera can be collected and recorded, such as: motherboard - model A - motherboard X-ray image - central area of the whole machine image, etc.
[0085] Step 32: Based on the quadruple data corresponding to each component in multiple models and the whole machine X-ray images corresponding to multiple models, train the preset neural network model to obtain a multimodal model.
[0086] In this step, based on the quadruple data collected in step 31 and the full-machine X-ray images corresponding to each model, a preset neural network model (such as a multimodal Transformer or a region recognition network) is trained so that the neural network model can establish a mapping relationship from the full-machine X-ray image to the names and regions of internal components.
[0087] During training, the neural network model learns to align the whole machine image with the local features and positional information of the components in the quadruple, thereby obtaining a trained neural network model, denoted as a multimodal model.
[0088] For example, after training with quadruple data from multiple laptops or mobile phones and whole-machine X-ray images, the neural network model can learn to identify from new whole-machine images that the keyboard module is located in the lower region and the heat dissipation module is located in the rear region.
[0089] Optionally, the following can also be performed: after calibrating the component names and corresponding areas of each component in the model to be tested, update the multimodal model based on the calibrated component names and areas of each component in the model to be tested.
[0090] In this implementation, in order to continuously optimize the recognition performance and generalization ability of the multimodal model, the multimodal model can be updated based on manual calibration and feedback mechanisms.
[0091] In one possible implementation, after the multimodal model outputs the names and regions of the components to be tested, the results are verified and corrected by professionals.
[0092] For example, correcting misidentifications, adjusting the position of the region box, or adding unidentified components can form an accurate correspondence between the calibrated model, component, and region.
[0093] Subsequently, these calibrated data are used as new training samples to update the original multimodal model through incremental learning or in conjunction with memory mechanisms.
[0094] For example, if a multimodal model fails to accurately identify a new camera module in a new mobile phone, the calibrator can label its correct name and location, add the sample to the training set, and update the multimodal model parameters so that the model can correctly identify similar components in the next identification.
[0095] The component identification method provided in this application acquires four-tuple data corresponding to each component in multiple machine models; based on the four-tuple data corresponding to each component in multiple machine models and the X-ray images of the whole machine corresponding to multiple machine models, a preset neural network model is trained to obtain a multimodal model. Firstly, this training mechanism effectively establishes a cross-level mapping relationship from the macroscopic structure of the whole machine to the microscopic features of the components, enabling the model to learn the spatial distribution patterns and morphological characteristics of components within different machine models; secondly, multimodal data fusion training enhances the model's ability to understand the complex contextual relationships of components in the whole machine environment, significantly improving the accuracy of distinguishing similar components or overlapping areas; finally, the trained model possesses strong generalization ability, enabling accurate identification and positioning of components within multiple unknown machine models without retraining for specific machine models, providing an efficient and reliable technical foundation for automated non-destructive testing of large-scale, multi-model equipment.
[0096] Based on the above embodiments, Figure 4 Flowchart of the component identification method provided in the embodiments of this application Figure 3 ,like Figure 4 As shown, the method also includes:
[0097] Step 41: Based on the component name and corresponding region of each component, cluster the same component to obtain different regions of different components;
[0098] In this step, in order to form a component region knowledge base, the component results identified by the multimodal model can be categorized and integrated.
[0099] In one possible implementation, based on the names of each component and their corresponding regions output by the multimodal model, cluster analysis is performed on the data of different regions corresponding to components with the same name, thereby summarizing the typical distribution patterns and regional characteristics of the same type of components in the whole machine X-ray image.
[0100] For example, for the camera component category, the location and outline of multiple camera areas from different models can be clustered to form distribution patterns such as the front camera being mostly located in the upper left area of the device and the rear main camera being mostly located in the upper right area of the device.
[0101] Step 42: Determine the different morphological characteristics of different components based on their different regions.
[0102] In this step, based on the typical regions of each component obtained by clustering in step 41, the morphological features of different components in the same region in the X-ray image are further extracted, thereby achieving fine-grained differentiation of component morphology.
[0103] In one possible implementation, within a defined component region, feature extraction and pattern analysis are performed on the image to identify morphological differences (such as size, shape, internal structure, etc.) and classify them into different morphological subclasses.
[0104] For example, within the battery component area, different forms such as square lithium-ion batteries and pouch polymer batteries can be further distinguished, each with different X-ray density distribution and contour characteristics.
[0105] Optionally, this method can also perform the following implementation:
[0106] Step 1: Obtain the feature vector corresponding to the X-ray image of the component to be inspected;
[0107] In this implementation, the deep feature representation, i.e., the feature vector, of the X-ray image of the component to be inspected can be extracted.
[0108] In one possible implementation, the X-ray image of the component to be inspected is input into a trained feature extraction network (e.g., a convolutional neural network) to transform the X-ray image into a high-dimensional feature vector that can effectively characterize the visual and structural properties of the component.
[0109] For example, an X-ray image of a camera module can be processed by feature extraction to obtain a fixed-dimensional vector, which encodes key information such as the shape, size, and internal component layout of the module.
[0110] Step 2: Based on the feature vector corresponding to the X-ray image of the component to be detected, determine the target component to which the component to be detected belongs based on the different morphological features of different components.
[0111] In this implementation, the feature vector obtained in step 1 is used to perform similarity matching or classification decision with the existing component morphology feature library to determine the specific component category to which the component belongs, i.e., the target component.
[0112] In one possible implementation, the distance or similarity between the feature vector and each known component category and its different morphological feature vectors is calculated, and the category with the closest distance or the highest similarity is determined as the target component.
[0113] For example, if the extracted feature vector is closest to the shape feature of a square lithium-ion battery under the battery category in the feature library, then the component to be detected is determined to be a square lithium-ion battery.
[0114] Step 3: Update the different morphological features of the target component based on the feature vector corresponding to the X-ray image of the component to be inspected.
[0115] In this implementation, the morphological features of the target component are iteratively updated and optimized based on the newly detected component feature data.
[0116] In one possible implementation, the feature vector of the component to be detected is fused with the existing morphological features of the target component or incorporated into a memory mechanism. That is, through feature averaging, clustering and recombination, or feature update methods based on attention weights, the feature representation of the morphology is made more comprehensive and typical.
[0117] For example, if a new sample is detected that is slightly different from the existing camera morphology features but belongs to the same subclass, the features of the sample can be fused into the morphology category, thereby enhancing the robustness and coverage of the multimodal model in recognizing morphological changes of this type of component.
[0118] Correspondingly, if the target component to which the component to be tested belongs is not determined among the different morphological characteristics of different components, the multimodal model is updated based on the X-ray image of the component to be tested.
[0119] In this implementation, when the feature vector of the component to be detected cannot match any existing component morphological features, it indicates that the component may belong to a new morphological variant or a new category that has not yet been recorded in the system.
[0120] At this point, the X-ray image of the component and its contextual information within the whole machine will be used to perform targeted updates to the multimodal model.
[0121] In one possible implementation, the unknown component and the image and region information of the whole machine in which it is located are first labeled as new samples to form new quadruple data. Then, through incremental learning or online training, the samples of the quadruple data are added to the training set of the multimodal model to adjust the parameters of the multimodal model so that it can identify such components in the future.
[0122] For example, if a new model is equipped with a heat dissipation module that has never appeared before, the system will not be able to match the existing category during the first detection. It will then mark it as an unknown heat dissipation module and update the multimodal model by combining the overall image and location information of the device, so that the multimodal model can correctly identify the new form in the next detection.
[0123] The component identification method provided in this application clusters the same component based on its name and corresponding region to obtain different regions for different components; and determines the different morphological characteristics of different components based on their different regions. This scheme, by clustering the identified components according to their name and region information, can automatically extract and summarize the possible morphological variations and spatial distribution patterns of the same type of component in various models or production batches from massive amounts of data, thereby establishing a complete morphological feature map of various components. This effectively solves the technical problem that current methods are difficult to adapt to component process updates, model iterations, and individual differences. It can not only identify standard-shaped components but also accurately identify atypical shapes caused by production deviations, design changes, or physical wear.
[0124] Based on the above method embodiments, the following technical effects can be achieved:
[0125] Technical effect 1: The component disassembly is photographed with X-ray images and the model-component-X-ray image pair is recorded. The component area-name recognition-type determination network has strong generalization ability.
[0126] Technical effect 2: For a large amount of collected data, it is easy to distinguish the labels and categories of whole machine images, and can quickly build various datasets required by the algorithm;
[0127] Technical effect 3: By utilizing the memory mechanism, representative features are stored for a long time, and new features can be recorded incrementally, enabling the model to automatically recognize new forms.
[0128] Based on the above method embodiments, Figure 5 This is a schematic diagram of the structure of the component identification device provided in the embodiments of this application, as shown below. Figure 5 As shown, this device includes:
[0129] The acquisition module 51 is used to acquire the whole X-ray image of the model to be tested;
[0130] The processing module 52 is used to input the whole X-ray image of the model to be tested into the multimodal model to obtain the component name and corresponding area of each component in the model to be tested;
[0131] The multimodal model is obtained by training a preset neural network model based on the quadruple data corresponding to each component in multiple models and the whole machine X-ray images corresponding to multiple models. Each quadruple data includes: component name, the first model corresponding to the component, the X-ray image corresponding to the component, and the region corresponding to the component.
[0132] In one or more embodiments, before inputting the whole-machine X-ray image of the model to be tested into the multimodal model to obtain the component names and corresponding regions of each component in the model to be tested, the processing module 52 is further configured to:
[0133] Obtain the quadruple data corresponding to each component in multiple models;
[0134] Based on the quadruple data corresponding to each component in multiple models and the whole machine X-ray images corresponding to multiple models, a preset neural network model is trained to obtain a multimodal model.
[0135] In one or more embodiments, the processing module 52 is further configured to:
[0136] After calibrating the component names and corresponding areas of each component in the model to be tested, the multimodal model is updated based on the calibrated component names and areas of each component in the model to be tested.
[0137] In one or more embodiments, the processing module 52 is further configured to:
[0138] Based on the component name and corresponding region of each component, the same component is clustered to obtain different regions of different components.
[0139] Based on different regions of different components, the different morphological characteristics of different components are determined.
[0140] In one or more embodiments, the processing module 52 is further configured to:
[0141] Obtain the feature vector corresponding to the X-ray image of the component to be inspected;
[0142] Based on the feature vector corresponding to the X-ray image of the component to be tested, the target component to which the component to be tested belongs is determined by the different morphological features of different components.
[0143] Based on the feature vector corresponding to the X-ray image of the component to be inspected, the different morphological features of the target component are updated.
[0144] In one or more embodiments, the processing module 52 is further configured to:
[0145] The target component to which the component to be tested belongs was not determined among the different morphological characteristics of different components. The multimodal model was updated based on the X-ray image of the component to be tested.
[0146] It should be noted that the division of the various modules in the above device is merely a logical functional division. In actual implementation, they can be fully or partially integrated into a single physical element, or they can be physically separated. Furthermore, these modules can be implemented entirely in software through processing element calls, or entirely in hardware. Alternatively, some modules can be implemented through processing element calls in software, while others can be implemented in hardware. Moreover, these modules can be integrated together or implemented independently. The processing element here can be an integrated circuit with signal processing capabilities. During implementation, each step of the above method or each of the above modules can be completed through the integrated logic circuits in the hardware of the processor element or through software instructions.
[0147] As can be seen from the above, the component identification device provided in this embodiment can execute the method provided in the above method embodiment, and its implementation principle and technical effect are similar. This embodiment will not be described in detail here.
[0148] Figure 6 This is a schematic diagram of the structure of an electronic device provided in an embodiment of this application. Figure 6 As shown, the electronic device provided in this embodiment includes at least one processor 61 and a memory 62.
[0149] Optionally, the electronic device also includes a communication component 63.
[0150] The processor 61, memory 62, and communication component 63 are connected via bus 64.
[0151] In a specific implementation, at least one processor 61 executes computer execution instructions stored in memory 62, causing at least one processor 61 to perform the above-described method.
[0152] The specific implementation process of processor 61 can be found in the above method embodiments, and its implementation principle and technical effect are similar. It will not be repeated here.
[0153] In the above embodiments, it should be understood that the processor can be a Central Processing Unit (CPU), or other general-purpose processors, digital signal processors (DSPs), application-specific integrated circuits (ASICs), etc. The general-purpose processor can be a microprocessor or any conventional processor. The steps of the method disclosed in this invention can be directly implemented by a hardware processor, or implemented by a combination of hardware and software modules within the processor.
[0154] The memory may include random access memory (RAM) and may also include non-volatile memory (NVM), such as at least one disk storage device.
[0155] The bus can be an Industry Standard Architecture (ISA) bus, a Peripheral Component Interconnect (PCI) bus, or an Extended Industry Standard Architecture (EISA) bus, etc. Buses can be categorized as address buses, data buses, control buses, etc. For ease of illustration, the buses shown in the accompanying drawings are not limited to a single bus or a single type of bus.
[0156] This application also provides a computer program product, including a computer program that, when executed by a processor, implements the above-described method.
[0157] This application also provides a computer-readable storage medium storing computer-executable instructions, which, when executed by a processor, implement the above-described method.
[0158] The aforementioned readable storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (SRAM), electrically erasable programmable read-only memory (EEPROM), erasable programmable read-only memory (EPROM), programmable read-only memory (PROM), read-only memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk. The readable storage medium can be any available medium accessible to a general-purpose or special-purpose computer.
[0159] An exemplary readable storage medium is coupled to a processor, enabling the processor to read information from and write information to the readable storage medium. Of course, the readable storage medium can also be a component of the processor. The processor and the readable storage medium can reside in an Application Specific Integrated Circuit (ASIC). Alternatively, the processor and the readable storage medium can exist as discrete components in the device.
[0160] The division of units is merely a logical functional division; in actual implementation, there may be other division methods. For example, multiple units or components may be combined or integrated into another system, or some features may be ignored or not executed. Furthermore, the coupling or direct coupling or communication connection shown or discussed may be indirect coupling or communication connection through some interfaces, devices, or units, and may be electrical, mechanical, or other forms.
[0161] The units described as separate components may or may not be physically separate. The components shown as units may or may not be physical units; that is, they may be located in one place or distributed across multiple network units. Some or all of the units can be selected to achieve the purpose of this embodiment according to actual needs.
[0162] In addition, the functional units in the various embodiments of the present invention can be integrated into one processing unit, or each unit can exist physically separately, or two or more units can be integrated into one unit.
[0163] If a function is implemented as 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 this invention, or the part that contributes to the prior art, or a part of the technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions to cause a computer device (which may be a personal computer, server, or network device, etc.) to execute all or part of the steps of the methods of the various embodiments of this invention. The aforementioned storage medium includes various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.
[0164] Those skilled in the art will understand that all or part of the steps of the above-described method embodiments can be implemented by hardware related to program instructions. The aforementioned program can be stored in a computer-readable storage medium. When executed, the program performs the steps of the above-described method embodiments; and the aforementioned storage medium includes various media capable of storing program code, such as ROM, RAM, magnetic disks, or optical disks.
[0165] Finally, it should be noted that other embodiments of the invention will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This invention is intended to cover any variations, uses, or adaptations of the invention that follow the general principles of the invention and include common knowledge or customary techniques in the art not disclosed herein, and is not limited to the precise structures described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of the invention is limited only by the appended claims.
Claims
1. A method for identifying electronic components, characterized in that, include: Obtain a full-machine X-ray image of the model to be tested; The whole machine X-ray image of the model to be tested is input into the multimodal model to obtain the component name and corresponding area of each component in the model to be tested; The multimodal model is obtained by training a preset neural network model based on the quadruple data corresponding to each component in multiple models and the whole machine X-ray images corresponding to each model. Each quadruple data includes: component name, the first model corresponding to the component, the X-ray image corresponding to the component, and the region corresponding to the component.
2. The method according to claim 1, characterized in that, Before inputting the whole-machine X-ray image of the machine to be tested into the multimodal model to obtain the component names and corresponding regions of each component in the machine to be tested, the method further includes: Obtain the quadruple data corresponding to each component in the multiple models; The multimodal model is obtained by training a preset neural network model based on the quadruple data corresponding to each component in the multiple models and the whole machine X-ray images corresponding to the multiple models.
3. The method according to claim 2, characterized in that, The method further includes: After calibrating the component names and corresponding regions of each component in the model under test, the multimodal model is updated based on the calibrated component names and regions of each component in the model under test.
4. The method according to any one of claims 1-3, characterized in that, The method further includes: Based on the component name and corresponding region of each component, the same component is clustered to obtain different regions of different components. Based on different regions of different components, the different morphological characteristics of different components are determined.
5. The method according to claim 4, characterized in that, The method further includes: Obtain the feature vector corresponding to the X-ray image of the component to be inspected; Based on the feature vector corresponding to the X-ray image of the component to be detected, the target component to which the component to be detected belongs is determined by the different morphological features of different components. Based on the feature vector corresponding to the X-ray image of the component to be detected, the different morphological features of the target component are updated.
6. The method according to claim 5, characterized in that, The method further includes: If the target component to which the component to be detected belongs is not determined among the different morphological characteristics of different components, the multimodal model is updated based on the X-ray image of the component to be detected.
7. A component identification device, characterized in that, include: The acquisition module is used to acquire the full-machine X-ray image of the model to be inspected; The processing module is used to input the whole machine X-ray image of the machine to be tested into the multimodal model to obtain the component name and corresponding area of each component in the machine to be tested; The multimodal model is obtained by training a preset neural network model based on the quadruple data corresponding to each component in multiple models and the whole machine X-ray images corresponding to each model. Each quadruple data includes: component name, the first model corresponding to the component, the X-ray image corresponding to the component, and the region corresponding to the component.
8. An electronic device, characterized in that, include: Memory, processor; The memory stores computer-executed instructions; The processor executes computer execution instructions stored in the memory, causing the processor to perform the method as described in any one of claims 1-6.
9. A computer-readable storage medium, characterized in that, The computer-readable storage medium stores computer-executable instructions, which, when executed by a processor, are used to implement the method as described in any one of claims 1-6.
10. A computer program product, characterized in that, Includes a computer program, which, when executed by a processor, is used to implement the method as described in any one of claims 1-6.