Part recognition method, device and storage medium

Abstract

The present disclosure provides a part identification method and device and a storage medium, relates to the technical field of communication, and can improve part identification accuracy. The method comprises the following steps: acquiring images of a plurality of feature parts of a part to be identified; determining a part model set corresponding to each feature part in the plurality of feature parts according to the images of the plurality of feature parts; and determining a part model of the part to be identified according to the part model set corresponding to each feature part. The present disclosure is used for improving the part identification accuracy.

Description

Technical Field

[0001] This disclosure relates to the field of communication technology, and in particular to a method, apparatus and storage medium for identifying parts. Background Technology

[0002] Currently, part recognition methods are mainly based on image recognition and 3D (3D) modeling recognition. However, image recognition is easily affected by ambient lighting and the size and features of the parts, making it difficult to extract effective features from images, leading to decreased accuracy. 3D modeling recognition, on the other hand, requires high computing power and precise 3D camera accuracy, resulting in significant recognition errors when multiple parts have high similarity or when the 3D camera's precision is limited. Therefore, improving the accuracy of part recognition has become an urgent technical problem to be solved. Summary of the Invention

[0003] This application provides a part identification method, apparatus, and storage medium that can improve the accuracy of part identification.

[0004] To achieve the above objectives, this application adopts the following technical solution:

[0005] In a first aspect, a part identification method is provided, comprising: acquiring images of multiple feature parts of a part to be identified; determining a set of part models corresponding to each feature part based on the images of the multiple feature parts; and determining the part model of the part to be identified based on the set of part models corresponding to each feature part.

[0006] In conjunction with the first aspect above, in one possible implementation, determining the set of part models corresponding to each of the multiple feature parts based on images of multiple feature parts includes: performing the following target operation on each feature part based on the images of the multiple feature parts to determine the set of part models corresponding to each feature part; the target operation includes: detecting the image of the target feature part according to a preset image detection algorithm to determine a first set of part models corresponding to the target feature part; the target feature part is one of the multiple feature parts; extracting the image region of the feature part from the image of the feature part, processing the image region according to a preset image classification algorithm to determine a second set of part models corresponding to the target feature part; and determining the set of part models corresponding to the target feature part based on the first set of part models and the second set of part models.

[0007] In conjunction with the first aspect above, in one possible implementation, the image of the target feature region includes images of the target feature region acquired from multiple viewpoints; the first part model set includes the first part model of the target feature region from multiple viewpoints; the second part model set includes the second part model of the target feature region from multiple viewpoints; determining the part model set corresponding to the target feature region based on the first part model set and the second part model set includes: determining whether the first part model set and the second part model set are the same; in response to the first part model set and the second part model set being the same, determining whether the part model set corresponding to the target feature region is the first part model set or the second part model set; in response to the first part model set and the second part model set being different, determining the first confidence level corresponding to the first part model of the target feature region from the i-th viewpoint and the second confidence level corresponding to the second part model from the i-th viewpoint; i is a positive integer; the first part model is the part model in the first part model set; the second part model is the part model in the second part model set; comparing the first confidence level and the second confidence level, determining the target confidence level; determining the part model from the i-th viewpoint in the part model set corresponding to the target feature region as the part model corresponding to the target confidence level.

[0008] In conjunction with the first aspect mentioned above, in one possible implementation, when the first set of part models and the second set of part models are different, the part model from the i-th viewpoint in the set of part models corresponding to the target feature part... Satisfy the following formula:

[0009]

[0010] in, W represents the model number of the first part from the i-th perspective. d The first confidence level; Let W be the model number of the second part from the i-th perspective; c α is the second confidence level; β is the scaling factor of the first part model set; β is the scaling factor of the second part model set; and thres is the threshold for distinguishing between the first part model set and the second part model set.

[0011] In conjunction with the first aspect above, in one possible implementation, the part model of the part to be identified is determined based on the set of part models corresponding to each feature part, including: determining the intersection of the set of part models corresponding to each feature part; determining whether the number of part models in the intersection is greater than a preset value; in response to the number of part models in the intersection being less than or equal to the preset value, determining the part models in the intersection as the part models of the part to be identified; and in response to the number of part models in the intersection being greater than the preset value, determining the part model with the highest confidence level in the intersection as the part model of the part to be identified.

[0012] In conjunction with the first aspect mentioned above, in one possible implementation, before acquiring images of multiple feature parts of the part to be identified, the method further includes: acquiring the part level of the part to be identified; if the part level of the part to be identified is a first preset level, determining the feature parts to be acquired and the image acquisition viewpoint of the part to be identified; acquiring images of the feature parts to be acquired according to the image acquisition viewpoint; if the part level of the part to be identified is a second preset level, determining the feature parts to be acquired of the part to be identified; acquiring images of the feature parts to be acquired; if the part level of the part to be identified is a third preset level, randomly acquiring images of multiple feature parts of the part to be identified.

[0013] Secondly, a part identification device is provided, comprising: a processing unit. The processing unit is configured to acquire images of multiple feature parts of a part to be identified; the processing unit is further configured to determine a set of part models corresponding to each feature part based on the images of the multiple feature parts; the processing unit is further configured to determine the part model of the part to be identified based on the set of part models corresponding to each feature part.

[0014] In conjunction with the second aspect above, in one possible implementation, the processing unit is specifically configured to: perform the following target operation for each feature part based on images of multiple feature parts, to determine the set of part models corresponding to each feature part; the target operation includes: detecting the image of the target feature part according to a preset image detection algorithm, and determining the first set of part models corresponding to the target feature part; the target feature part is one of the multiple feature parts; extracting the image region of the feature part from the image of the feature part, processing the image region according to a preset image classification algorithm, and determining the second set of part models corresponding to the target feature part; and determining the set of part models corresponding to the target feature part based on the first set of part models and the second set of part models.

[0015] In conjunction with the second aspect above, in one possible implementation, the image of the target feature region includes images of the target feature region acquired from multiple viewpoints; the first part model set includes the first part model of the target feature region from multiple viewpoints; the second part model set includes the second part model of the target feature region from multiple viewpoints; the processing unit is specifically configured to: determine whether the first part model set and the second part model set are the same; in response to the first part model set and the second part model set being the same, determine whether the part model set corresponding to the target feature region is the first part model set or the second part model set; in response to the first part model set and the second part model set being different, determine the first confidence level corresponding to the first part model of the target feature region from the i-th viewpoint and the second confidence level corresponding to the second part model from the i-th viewpoint; i is a positive integer; the first part model is the part model in the first part model set; the second part model is the part model in the second part model set; compare the first confidence level and the second confidence level to determine the target confidence level; determine the part model from the i-th viewpoint in the part model set corresponding to the target feature region as the part model corresponding to the target confidence level.

[0016] In conjunction with the second aspect above, one possible implementation is that, when the first set of part models and the second set of part models are different, the part model from the i-th viewpoint in the set of part models corresponding to the target feature part... Satisfy the following formula:

[0017]

[0018] in, W represents the model number of the first part from the i-th perspective. d The first confidence level; Let W be the model number of the second part from the i-th perspective; c α is the second confidence level; β is the scaling factor of the first part model set; β is the scaling factor of the second part model set; and thres is the threshold for distinguishing between the first part model set and the second part model set.

[0019] In conjunction with the second aspect above, in one possible implementation, the processing unit is specifically used to: determine the intersection of the sets of part models corresponding to each feature part; determine whether the number of part models in the intersection is greater than a preset value; in response to the number of part models in the intersection being less than or equal to the preset value, determine the part models in the intersection as the part models of the part to be identified; in response to the number of part models in the intersection being greater than the preset value, determine the part model with the highest confidence level in the intersection as the part model of the part to be identified.

[0020] In conjunction with the second aspect above, in one possible implementation, the processing unit is specifically used for: obtaining the part level of the part to be identified; when the part level of the part to be identified is a first preset level, determining the feature parts to be collected and the image acquisition viewpoint of the part to be identified; acquiring an image of the feature parts to be collected according to the image acquisition viewpoint; when the part level of the part to be identified is a second preset level, determining the feature parts to be collected of the part to be identified; acquiring an image of the feature parts to be collected; and when the part level of the part to be identified is a third preset level, randomly acquiring images of multiple feature parts of the part to be identified.

[0021] Thirdly, this disclosure provides a part identification device, which includes a processor and a memory; wherein the memory is used to store computer execution instructions, and when the part identification device is running, the processor executes the computer execution instructions stored in the memory to cause the part identification device to perform the part identification method as described in the first aspect and any possible implementation thereof.

[0022] Fourthly, this disclosure provides a computer-readable storage medium storing instructions that, when executed by a processor of a part identification device, enable the part identification device to perform the part identification method as described in the first aspect and any possible implementation thereof.

[0023] Fifthly, this disclosure provides a computer program product containing instructions that, when run on a part identification device, causes the part identification device to perform the part identification method as described in the first aspect and any possible implementation thereof.

[0024] In a sixth aspect, this disclosure provides a chip including a processor and a communication interface coupled to the processor, the processor being used to run computer programs or instructions to implement the part identification method as described in the first aspect and any possible implementation thereof.

[0025] Specifically, the chip provided in this application embodiment also includes a memory for storing computer programs or instructions.

[0026] In this disclosure, the name of the aforementioned part identification device does not limit the device or functional module itself. In actual implementation, these devices or functional modules may appear under other names. As long as the function of each device or functional module is similar to that of this disclosure, it falls within the scope of the claims of this disclosure and its equivalents.

[0027] These or other aspects of this disclosure will become more readily apparent in the following description.

[0028] The technical solution provided in this disclosure brings at least the following beneficial effects:

[0029] In this scheme, the part identification device acquires images of multiple feature parts of the part to be identified. Based on the images of these feature parts, the device determines the set of part models corresponding to each feature part. Finally, based on the set of part models corresponding to each feature part, the device determines the part model of the part to be identified. In this way, by extracting multiple feature parts of the part and the corresponding set of part models, the device determines the part model of the part to be identified, thus improving the accuracy of part identification. Attached Figure Description

[0030] Figure 1 This is a schematic diagram of the hardware structure of a part identification device provided in an embodiment of this application;

[0031] Figure 2 This is a schematic diagram of a parts identification device module provided in an embodiment of this application;

[0032] Figure 3 This is a schematic flowchart of a method for identifying parts provided in an embodiment of this application;

[0033] Figure 4 This is another schematic flowchart illustrating a method for identifying parts provided in an embodiment of this application;

[0034] Figure 5 This is another schematic flowchart illustrating a method for identifying parts provided in an embodiment of this application;

[0035] Figure 6 This is another schematic flowchart illustrating a method for identifying parts provided in an embodiment of this application;

[0036] Figure 7 This is another schematic flowchart illustrating a method for identifying parts provided in an embodiment of this application;

[0037] Figure 8 This is another schematic flowchart illustrating a method for identifying parts provided in an embodiment of this application;

[0038] Figure 9 This is a schematic diagram of the workflow of a parts identification system provided in an embodiment of this application;

[0039] Figure 10 This is a schematic diagram of the structure of a part identification device provided in an embodiment of this application. Detailed Implementation

[0040] The part identification method, apparatus and storage medium provided in this disclosure are described in detail below with reference to the accompanying drawings.

[0041] In this article, the term "and / or" is merely a description of the relationship between related objects, indicating that there can be three relationships. For example, A and / or B can represent three situations: A exists alone, A and B exist simultaneously, and B exists alone.

[0042] The terms "first" and "second," etc., used in the specification and drawings of this application are used to distinguish different objects or to distinguish different treatments of the same object, rather than to describe a specific order of objects.

[0043] Furthermore, the terms "comprising" and "having," and any variations thereof, used in the description of this application are intended to cover non-exclusive inclusion. For example, a process, method, system, product, or apparatus that includes a series of steps or units is not limited to the steps or units listed, but may optionally include other steps or units not listed, or may optionally include other steps or units inherent to such process, method, product, or apparatus.

[0044] It should be noted that in the embodiments of this application, the words "exemplary" or "for example" are used to indicate examples, illustrations, or explanations. Any embodiment or design scheme described as "exemplary" or "for example" in the embodiments of this application should not be construed as being more preferred or advantageous than other embodiments or design schemes. Specifically, the use of the words "exemplary" or "for example" is intended to present the relevant concepts in a specific manner.

[0045] Figure 1 This is a schematic diagram of the structure of a part identification device provided in an embodiment of this disclosure. Figure 1 As shown, the part identification device 100 includes at least one processor 101, a communication line 102, and at least one communication interface 104, and may also include a memory 103. The processor 101, memory 103, and communication interface 104 are connected via the communication line 102.

[0046] The processor 101 may be a central processing unit (CPU), an application-specific integrated circuit (ASIC), or one or more integrated circuits configured to implement the embodiments of this disclosure, such as one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs).

[0047] Communication line 102 may include a path for transmitting information between the aforementioned components.

[0048] The communication interface 104 is used to communicate with other devices or communication networks. It can use any transceiver-like device, such as Ethernet, radio access network (RAN), wireless local area network (WLAN), etc.

[0049] The memory 103 may be a read-only memory (ROM) or other type of static storage device capable of storing static information and instructions, random access memory (RAM) or other type of dynamic storage device capable of storing information and instructions, or electrically erasable programmable read-only memory (EEPROM), compact disc read-only memory (CD-ROM) or other optical disc storage, optical disc storage (including compressed optical discs, laser discs, optical discs, digital universal optical discs, Blu-ray discs, etc.), magnetic disk storage media or other magnetic storage devices, or any other medium capable of including or storing desired program code having the form of instructions or data structures and accessible by a computer, but not limited thereto.

[0050] In one possible design, the memory 103 can exist independently of the processor 101, meaning the memory 103 can be an external memory of the processor 101. In this case, the memory 103 can be connected to the processor 101 via a communication line 102 to store execution instructions or application code, and its execution is controlled by the processor 101 to implement the space measurement and determination method provided in the following embodiments of this disclosure. In another possible design, the memory 103 can also be integrated with the processor 101, meaning the memory 103 can be an internal memory of the processor 101. For example, the memory 103 can be a cache, which can be used to temporarily store some data and instruction information.

[0051] As one possible implementation, processor 101 may include one or more CPUs, for example Figure 1 CPU0 and CPU1 in the example. Alternatively, the part identification device 100 may include multiple processors, such as CPU0 and CPU1. Figure 1 The processors 101 and 107 are included. Alternatively, the part identification device 100 may also include an output device 105 and an input device 106.

[0052] Through the above description of the implementation methods, those skilled in the art will clearly understand that, for the sake of convenience and brevity, only the division of the above functional modules is used as an example. In practical applications, the above functions can be assigned to different functional modules as needed, that is, the internal structure of the network node can be divided into different functional modules to complete all or part of the functions described above. The specific working process of the system, modules, and network nodes described above can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0053] In related technologies, due to the wide variety of automotive parts and the high similarity between different parts, manual visual identification suffers from large errors, low efficiency, and high cost. Existing technologies include many automated identification methods for automotive parts, such as identification and localization of typical automotive parts based on RGB-D (Red, Green, Blue) cameras, automotive part identification methods based on spatial shape context features, and automotive part visual inspection systems.

[0054] In this study, a machine vision-based recognition and localization system was designed, using Kinect as the RGB-D camera, to identify and locate typical automotive parts based on an RGB-D camera. The system utilizes the Scale-invariant Feature Transform (SIFT) algorithm to extract feature points from the target part, compares the similarity between these feature points and those of the target part, and then uses depth images to further identify the target area. Finally, by calculating the position parameters of the target part, its three-dimensional coordinates are obtained, completing the localization process.

[0055] The method for recognizing automotive parts based on spatial shape context features belongs to the category of automotive parts recognition. The specific steps include: constructing an offline automotive parts feature library; extracting features of the automotive parts to be recognized online, namely, extracting feature points of the automotive parts to be recognized in three-dimensional space, using the relative positional relationship of feature points formed by segmenting spatial regions as the features of the automotive parts, and performing feature matching and recognition with automotive parts in the offline automotive parts feature library.

[0056] The visual inspection system for automotive parts includes an optical imaging system, an image acquisition device, an image processing system, an intelligent decision-making mechanism, and a control execution mechanism. The image processing system performs image enhancement, segmentation, feature extraction, and pattern recognition.

[0057] However, existing image-based part recognition methods are susceptible to image quality issues due to variations in lighting intensity and viewing angle. Some automotive parts are too large or highly similar, making it difficult to capture their entirety in a single image. Furthermore, the three-dimensional features of parts are difficult to represent in a single picture. Feature extraction methods like SIFT are highly dependent on the environment; when lighting is insufficient or the object's features are not obvious, it is difficult to extract effective features. 3D modeling-based part recognition methods require sophisticated and expensive equipment, have a high barrier to entry for manual operation, and involve complex 3D modeling processes that consume significant computational resources. They also struggle to accurately distinguish between parts with similar shapes. Additionally, the limited accuracy of 3D cameras may introduce additional errors, making it difficult to meet recognition requirements in some cases. Therefore, improving the accuracy of part recognition has become a pressing technical problem.

[0058] In order to improve the accuracy of part identification, this application provides a method such as Figure 2 The part identification and classification system shown includes: a data acquisition device 201, a service module 202, a web management module 203, an algorithm module 204, a comprehensive analysis module 2041, a label scanning and recognition module 2042, and a part image recognition module 2043.

[0059] Among them, the data acquisition device 201 can capture and scan information about automotive parts, and can collect data on the labels of the parts and the appearance images of the parts respectively.

[0060] Service module 202 can be implemented through a server and serves as the overall management and control module for the system. Images acquired by the acquisition devices are uploaded to this server via the network and managed and viewed through the web management module 203.

[0061] The web management module 203 allows for remote viewing, statistics, and management of data and corresponding recognition result records, while also providing historical record query and analysis.

[0062] The comprehensive analysis module 2041 can perform comprehensive analysis on the part model identified by barcode scanning and the part model identified by part image recognition, obtain the part model result, compare the result image with the image in the database, confirm the final part recognition result, and upload the result to the database on the server.

[0063] The label scanning and recognition module 2042 is mainly used to identify the labels on the parts containers (including parts boxes, parts frames, etc.), obtain the default part model information specified by the worker or customer, and send it to the comprehensive analysis module 2041 for further analysis and processing.

[0064] The part image recognition module 2043 mainly identifies the acquired part image information, determines the specific model information of the part from the information of the actual part image, and sends it to the comprehensive analysis module 2041 for further analysis and processing. Among them, the label scanning recognition module 2042, the part image recognition module 2043, and the comprehensive analysis module 2041 can all be considered as a whole as the algorithm module 204.

[0065] To improve the accuracy of part recognition, a part recognition method is also provided. The part recognition device acquires images of multiple feature parts of the part to be recognized. Based on the images of the multiple feature parts, the part recognition device determines the set of part models corresponding to each feature part. The part recognition device then finds the intersection of the set of part models corresponding to each feature part to determine the part model of the part to be recognized. In this way, the part recognition device improves the accuracy of part recognition by extracting multiple feature parts of the part and the corresponding set of part models.

[0066] The part identification method provided in this disclosure can be applied to, for example, Figure 1 In the part identification device shown, such as Figure 3 As shown, the part identification method provided in this embodiment can be implemented through the following steps 301 to 303.

[0067] Step 301: The part recognition device acquires images of multiple feature parts of the part to be recognized.

[0068] In one possible implementation, the part recognition device takes multi-view, multi-part images of the part according to its level to obtain images of multiple feature parts of the part to be recognized.

[0069] Specifically, the parts identification device obtains the part's model number from its database using a label on the part's container. Based on this model number, it classifies the part's identification level. According to the identification level, the device takes multi-view, multi-part images of the part, acquiring images of multiple feature areas of the part to be identified.

[0070] In one example, for automotive parts that are difficult to identify due to similar textures and shapes, the part identification device obtains the part's model and identification level through a label on the part's container. Based on the part's identification level, the device captures images of multiple feature areas of the part from different perspectives, obtaining images of these feature areas.

[0071] Step 302: The part identification device determines the set of part models corresponding to each of the multiple feature parts based on the images of the multiple feature parts.

[0072] In one possible implementation, the part recognition device extracts multiple feature parts from an image of multiple feature parts. Based on the extracted feature parts, the part recognition device retrieves a set of part models corresponding to each feature part from a database.

[0073] For example, automotive parts with similar shapes may have different model numbers but share some common features. For instance, rearview mirrors from different car models may have multiple identical features. The part identification device determines the set of part models corresponding to each of these identical features based on images of these features.

[0074] Step 303: The part identification device determines the part model of the part to be identified based on the set of part models corresponding to each feature part.

[0075] In one possible implementation, the part identification device performs an intersection operation on the set of part models corresponding to each feature part obtained in step 302 to determine the part model of each feature part of the part to be identified.

[0076] Specifically, the part identification device performs an intersection operation on the set of part models corresponding to each feature part. If the operation result has multiple part models, the part model with the higher confidence level is determined as the part model of the part to be identified. The image in the database corresponding to the part model is then compared with the multiple feature images captured by the part identification device to further determine the final part model to be identified.

[0077] The above solution offers at least the following advantages: The part identification device determines the set of part models corresponding to each of the multiple feature parts based on images of those feature parts. Then, based on the set of part models corresponding to each feature part, the part identification device determines the part model of the part to be identified. In this way, by extracting multiple feature parts of the part and the corresponding set of part models, the part identification device determines the part model of the part to be identified, thus improving the accuracy of part identification.

[0078] Combination Figure 3 ,like Figure 4 As shown, in step 302, the part identification device can perform the following target operation for each of the multiple feature parts based on the images of the feature parts, to determine the set of part models corresponding to each feature part.

[0079] This target operation can be achieved through the following steps 401-403.

[0080] Step 401: The part recognition device detects the image of the target feature part according to the preset image detection algorithm and determines the first part model set corresponding to the target feature part.

[0081] The target feature region is one of multiple feature regions.

[0082] In one possible implementation, the part recognition device uploads multiple feature images to a server and detects the part features using a preset image detection algorithm. The part recognition device detects the image of the target feature area according to the preset image detection algorithm, obtaining the location of the part feature in the image and determining the first set of part models corresponding to the target feature area.

[0083] Step 402: The part identification device extracts the image region of the feature part from the image of the feature part, processes the image region according to the preset image classification algorithm, and determines the second part model set corresponding to the target feature part.

[0084] In one possible implementation, the part recognition device identifies the location of the part's feature area in the image using a preset detection algorithm. The part recognition device then crops out the image of the feature area and feeds it separately into a preset image classification algorithm. Based on the extracted image area of ​​the feature area, the part recognition device processes the image area using the preset image classification algorithm to determine the second set of part models corresponding to the target feature area.

[0085] Step 403: The part identification device determines the set of part models corresponding to the target feature based on the first set of part models and the second set of part models.

[0086] The images of the target feature include images of the target feature acquired from multiple viewpoints; the first part model set includes the first part model of the target feature from multiple viewpoints; and the second part model set includes the second part model of the target feature from multiple viewpoints.

[0087] In one possible implementation, the part identification device compares the first part model of the target feature part from multiple perspectives in the first part model set with the second part model of the target feature part from multiple perspectives in the second part model set to determine the part model set corresponding to the target feature part.

[0088] The above solution offers at least the following advantages: The part recognition device detects images of target feature areas using a preset image detection algorithm and determines a first set of part models corresponding to the target feature areas. Based on the first set of part models and a second set of part models, the part recognition device determines a set of part models corresponding to the target feature areas. In this way, the part recognition device can determine the set of part models corresponding to each feature area based on images of multiple feature areas.

[0089] Combination Figure 4 ,like Figure 5 As shown, step 403 can be implemented through the following steps 501-505.

[0090] Step 501: The part identification device determines whether the first set of part models and the second set of part models are the same.

[0091] In one possible implementation, the part identification device determines whether the first part model of the target feature part in the first part model set under multiple views is the same as the second part model of the target feature part in the second part model set under multiple views.

[0092] Step 502: The part identification device, in response to the fact that the first part model set and the second part model set are the same, determines that the part model set corresponding to the target feature part is the first part model set or the second part model set.

[0093] Step 503: In response to the fact that the first part model set and the second part model set are different, the part identification device determines the first confidence level corresponding to the first part model and the second confidence level corresponding to the second part model in the i-th view of the target feature part.

[0094] Where i is a positive integer; the first part model is the part model in the first part model set; the second part model is the part model in the second part model set.

[0095] In one possible implementation, if the target feature part has different part models in the first part model set and the second part model set under the i-th view, the part identification device determines the first confidence level corresponding to the first part model of the target feature part under the i-th view and the second confidence level corresponding to the second part model under the i-th view.

[0096] Step 504: The part identification device compares the first confidence level and the second confidence level to determine the target confidence level.

[0097] In one possible implementation, the part recognition device determines the target confidence level of the target feature part in the i-th view based on the first confidence level corresponding to the first part model and the second confidence level corresponding to the second part model, as well as parameters such as the scaling factor and the discrimination threshold under the preset image detection algorithm and the preset image classification algorithm.

[0098] Step 505: The part identification device determines the part model in the i-th view of the part model set corresponding to the target feature part as the part model corresponding to the target confidence level.

[0099] In one possible implementation, the part identification device determines the part model from the i-th viewpoint in the set of part models corresponding to the target feature part as the part model corresponding to the target confidence level, based on the target confidence level of the target feature part.

[0100] One example is the part model from the i-th viewpoint in the set of part models corresponding to the target feature. Satisfy the following formula 1:

[0101]

[0102] in, W represents the model number of the first part from the i-th perspective. d The first confidence level; Let W be the model number of the second part from the i-th perspective; c α is the second confidence level; β is the scaling factor of the first part model set; β is the scaling factor of the second part model set; and thres is the threshold for distinguishing between the first part model set and the second part model set.

[0103] It should be noted that the scaling factor α of the first set of part models, the scaling factor β of the second set of part models, and the threshold thres for distinguishing between the first set of part models and the second set of part models are hyperparameters and need to be set according to the actual experimental conditions.

[0104] Furthermore, when |W d -W c When |≥thres|, it indicates that the confidence level of the part model corresponding to one algorithm is much higher than that of the part model corresponding to the other algorithm. In this case, the part identification device selects the model with the higher confidence level as the part model corresponding to the target confidence level. Conversely, if the confidence levels of the part models corresponding to the two algorithms are relatively close, the part identification device determines the scaling factor α of the preset image detection algorithm and the scaling factor β of the preset image classification algorithm based on the accuracy of the preset image detection algorithm and the accuracy of the preset image classification algorithm under different perspectives. The part identification device determines the part model corresponding to the target confidence level by increasing the weight of the scaling factor.

[0105] The above scheme brings at least the following beneficial effects. The part identification device determines whether a first set of part models and a second set of part models are the same. If they are the same, the part identification device, in response to the same set of part models, determines that the set of part models corresponding to the target feature is either the first set of part models or the second set of part models. If they are different, the part identification device, in response to the different set of part models, determines the first confidence level corresponding to the first part model of the target feature in the i-th view, and the second confidence level corresponding to the second part model in the i-th view. The part identification device compares the first confidence level and the second confidence level to determine the target confidence level. The part identification device determines the part model in the i-th view of the set of part models corresponding to the target feature as the target confidence level of the target feature in the i-th view. Thus, by using the first and second confidence levels of the part models to be identified, the part identification device performs secondary authentication and fusion on the part models obtained from the preset image detection algorithm and the preset image classification algorithm, improving the accuracy of the identification of the part models to be identified.

[0106] Combination Figure 3 ,like Figure 6 As shown, step 303 can be implemented through the following steps 601-604.

[0107] Step 601: The part identification device determines the intersection of the set of part models corresponding to each feature part.

[0108] In one possible implementation, the part identification device determines the intersection of the set of part models corresponding to each characteristic part based on the set of part models corresponding to the determined target feature parts.

[0109] Specifically, the part recognition device obtains a set of features of the part from multiple perspectives based on a preset neural network. By using a voting method to filter out erroneous or duplicate features, the target feature set of the part is determined from multiple perspectives. The part identification device determines the set of part models corresponding to each feature based on the target feature set from multiple perspectives. The part identification device also determines the intersection of the set of part models corresponding to each feature.

[0110] In one example, the set of part models corresponding to each feature part satisfies the following formula 2:

[0111]

[0112] Among them, L iUsed to represent the i-th feature part; wrap is a mapping function used to map each feature part to a corresponding set of part types.

[0113] Furthermore, the intersection of the sets of part models corresponding to each feature part satisfies the following formula 3:

[0114]

[0115] in, This is used to represent the intersection of the set of part models corresponding to each feature part.

[0116] Step 602: The part identification device determines whether the number of part models in the intersection is greater than a preset value.

[0117] In one example, the preset value mentioned above is 1.

[0118] Step 603: The part identification device responds to the fact that the number of part models in the intersection is less than or equal to a preset value, and determines the part models in the intersection as the part models of the part to be identified.

[0119] In one possible implementation, if the number of part models in the intersection is not greater than a preset value, the part identification device determines the part models in the intersection as the part models of the part to be identified.

[0120] In one example, if the number of part models in the intersection is 1, that is, there is only one part model in the intersection, the part identification device determines that the part model in the intersection is the part model of the part to be identified.

[0121] If the number of part models in the intersection is less than 1, that is, if there are no part models in the intersection, it means that there is a shortage of part model information in the database. The part identification device can remind relevant staff to supplement the relevant part information in the database.

[0122] Step 604: In response to the fact that the number of part models in the intersection is greater than a preset value, the part identification device determines the part model with the highest confidence in the intersection as the part model of the part to be identified.

[0123] In one possible implementation, if the number of part models in the intersection is greater than a preset value, the part identification device determines the part model with the highest confidence level in the intersection as the part model of the part to be identified.

[0124] In one example, if the number of part models in the intersection is greater than 1, that is, there are multiple part models in the intersection, the part identification device determines the part model with the highest confidence level in the intersection as the part model of the part to be identified.

[0125] The above scheme brings at least the following beneficial effects: The part identification device determines the intersection of the set of part models corresponding to each feature part. The part identification device determines whether the number of part models in the intersection is greater than a preset value. If the number of part models in the intersection is less than or equal to the preset value, the part identification device determines the part models in the intersection as the part models of the part to be identified. If the number of part models in the intersection is greater than the preset value, the part identification device determines the part model with the highest confidence level in the intersection as the part model of the part to be identified. In this way, by further confirming the model of the part to be identified through the intersection of the set of part models corresponding to each feature part, the part identification device improves the accuracy of part identification.

[0126] Combination Figure 3 ,like Figure 7 As shown, steps 701-706 are included before step 301.

[0127] Step 701: The part identification device obtains the part grade of the part to be identified.

[0128] In one possible implementation, the part identification device obtains the part information of the part to be identified by scanning a barcode. The part identification device obtains the part level of the part to be identified through a classification algorithm.

[0129] Specifically, the parts identification device scans the label attached to the container of the part to be identified and retrieves the default information of the part to be identified from the database. Based on the default information of the part to be identified, the parts identification device classifies the parts to be identified using a hierarchical algorithm and performs multi-view, multi-part processing based on the characteristics of the parts.

[0130] It should be noted that when the part identification device cannot obtain the part information of the part to be identified by scanning the code, the part identification device notifies the relevant staff to add the part information to be identified.

[0131] Step 702: When the part level of the part to be identified is the first preset level, the part identification device determines the feature parts to be collected and the image acquisition angle of the part to be identified.

[0132] In one possible implementation, when the part to be identified is a part of a first preset level, the part identification device determines the feature area to be collected and the image acquisition angle of the part to be identified.

[0133] In one example, the first preset level of parts is difficult parts, which are easily confused with similar parts and are difficult to distinguish.

[0134] Step 703: The part recognition device acquires an image of the feature area to be acquired based on the image acquisition angle.

[0135] In one possible implementation, the part identification device acquires an image of the feature area to be acquired based on the feature area to be acquired and the image acquisition angle of the part to be identified determined in step 702.

[0136] Step 704: When the part level of the part to be identified is the second preset level, the part identification device determines the feature parts to be collected of the part to be identified.

[0137] In one possible implementation, when the part to be identified is a part of the second preset level, the part identification device determines the feature area to be collected of the part to be identified.

[0138] In one example, the second preset level of parts is a large part that is difficult to see the whole picture of.

[0139] Step 705: The part recognition device acquires an image of the feature area to be acquired.

[0140] In one possible implementation, the part identification device acquires an image of the feature area to be acquired based on the feature area to be acquired and the image acquisition angle of the part to be identified determined in step 704.

[0141] Step 706: When the part to be identified is at the third preset level, the part identification device randomly acquires images of multiple feature parts of the part to be identified.

[0142] In one possible implementation, when the part to be identified is a part of the third preset level, the part identification device randomly acquires images of multiple feature parts of the part to be identified.

[0143] In one example, the third preset level of parts is a simple part that is easy to identify.

[0144] The above scheme brings at least the following beneficial effects: The part identification device acquires the part level of the part to be identified. When the part level is a first preset level, the part identification device determines the feature areas to be collected and the image acquisition angle of the part. The part identification device acquires images of the feature areas to be collected based on the image acquisition angle. When the part level is a second preset level, the part identification device determines the feature areas to be collected. The part identification device acquires images of the feature areas to be collected. When the part level is a third preset level, the part identification device randomly acquires images of multiple feature areas of the part. In this way, by classifying the part to determine the shooting angle and feature areas, the part identification device can improve the accuracy of part feature extraction.

[0145] The following, combined with Figure 8The overall process of the part identification device identifying parts is explained below:

[0146] Step 801: The part identification device scans the code on the container of the part to be identified.

[0147] One possible implementation is that the part identification device scans the label on the part container.

[0148] One example, such as Figure 9 As shown, the part identification device scans the labels on the part containers using, but not limited to, PDA devices, common industrial cameras, and mobile phones.

[0149] Step 802: The part identification device determines whether there is relevant information about the part to be identified in the database.

[0150] In one possible implementation, the part identification device determines in step 801 whether there is relevant information about the part to be identified in the database.

[0151] Specifically, the parts identification device scans the label on the parts container to determine if there is any information about the parts to be identified in the data.

[0152] Step 803: The part identification device retrieves relevant information about the part to be identified from the database.

[0153] In one possible implementation, if there is relevant information about the part to be identified in the database, the part identification device obtains the default relevant information about the part to be identified from the database.

[0154] Step 804: The part identification device generates a notification message.

[0155] In one possible implementation, if there is no relevant information about the part to be identified in the database, the part identification device notifies relevant personnel to collect and supplement the information about the part to be identified in the database.

[0156] One example, such as Figure 9 As shown, the parts identification device displays and analyzes the identified parts through a web management platform, providing a user interface for relevant personnel and statistically analyzing historical data. It also displays the processing and analysis of the identified parts.

[0157] Step 805: The part identification device determines the grade of the part to be identified.

[0158] The specific implementation of step 805 is similar to that of step 701 above. The specific implementation process can be referred to step 701, and will not be repeated here.

[0159] Step 806: When the part level of the part to be identified is the first preset level, the part identification device determines the feature parts to be collected and the image acquisition angle of the part to be identified.

[0160] The specific implementation of step 806 is similar to that of step 702 above. The specific implementation process can be referred to step 702, and will not be repeated here.

[0161] Step 807: When the part level of the part to be identified is the second preset level, the part identification device determines the feature parts to be collected of the part to be identified.

[0162] The specific implementation of step 807 is similar to that of step 704 above. The specific implementation process can be referred to step 704, and will not be repeated here.

[0163] Step 808: When the part to be identified is at the third preset level, the part identification device randomly acquires images of multiple feature parts of the part to be identified.

[0164] The specific implementation of step 808 is similar to that of step 706 above. The specific implementation process can be referred to step 706, and will not be repeated here.

[0165] Step 809: The part recognition device uploads images of multiple feature parts of the part to be recognized to the server.

[0166] Step 8010: The part recognition device detects the image of the target feature part according to the preset image detection algorithm and determines the first part model set corresponding to the target feature part.

[0167] The specific implementation of step 8010 is similar to that of step 401 above. The specific implementation process can be referred to step 401, and will not be repeated here.

[0168] Step 8011: The part identification device extracts the image region of the feature part from the image of the feature part, processes the image region according to the preset image classification algorithm, and determines the second part model set corresponding to the target feature part.

[0169] The specific implementation of step 8011 is similar to that of step 402 above. The specific implementation process can be referred to step 402, and will not be repeated here.

[0170] Step 8012: The part identification device determines the set of part models corresponding to the target feature based on the first set of part models and the second set of part models.

[0171] The specific implementation of step 8012 is similar to that of step 403 above. The specific implementation process can be referred to step 403, and will not be repeated here.

[0172] Step 8013: The part recognition device analyzes the recognition results of the part to be recognized.

[0173] The specific implementation of step 8013 is similar to that of steps 601-604 above. The specific implementation process can be referred to steps 601-604, and will not be repeated here.

[0174] One example, such as Figure 9 As shown, the part recognition device uploads the acquired image of the part to be recognized to the server, uses a preset image detection algorithm to detect the image, and obtains the detection result corresponding to the part feature and the image region where it is located. The part recognition device cropped the image region where the part feature is located, and used a preset image classification algorithm to obtain the classification result corresponding to the image region. At the same time, the part recognition device combines the camera shooting perspective and, according to Formula 1 in step 505, fuses the part model set of the part to be recognized obtained by the preset image detection algorithm and the preset image classification algorithm to obtain the feature set of the part under multiple perspectives. This process continues until the image processing of all perspectives of the part to be recognized is completed.

[0175] Step 8014: The part recognition device obtains the image information of the part to be recognized from the database.

[0176] In one possible implementation, the template image information corresponding to the part to be identified is obtained from the data of the part identification device.

[0177] Step 8015: The part recognition device compares the image of the feature to be recognized with the image information of the part to be recognized in the database.

[0178] In one possible implementation, the part recognition device compares the image of the feature to be recognized with the template image information corresponding to the part to be recognized obtained from the database.

[0179] Step 8016: The part identification device determines whether the part model is consistent.

[0180] In one possible implementation, the part identification device determines whether the identification model of the part to be identified is consistent with the model obtained from the database based on the comparison result of step 8015.

[0181] Step 8017: The part identification device generates a warning message.

[0182] The warning message is used to remind you to add part model information to the database.

[0183] In one possible implementation, if the model number of the part to be identified does not match the model number obtained from the database, the part identification device will pop up a warning window to notify relevant personnel to handle the issue.

[0184] In one example, the parts identification device notifies relevant personnel to handle the issue. The personnel can log in with the appropriate permissions, such as... Figure 9 Check the recognition status in the web management system shown.

[0185] Step 8018: The part identification device uploads the results to the database.

[0186] In one possible implementation, after the part identification device processes the part to be identified, it uploads the identification result to the database.

[0187] In one example, the part identification device uploads the identification results to a database for storage, enabling statistical analysis and querying for subsequent identification work.

[0188] The functions of each device in the service transmission apparatus and the part identification apparatus involved in the embodiments of this disclosure, as well as the interaction between the devices, have been described in detail above.

[0189] As can be seen, the above mainly describes the technical solutions provided by the embodiments of this disclosure from a methodological perspective. To achieve the above functions, it includes corresponding hardware structures and / or software modules for executing each function. Those skilled in the art should readily recognize that, in conjunction with the modules and algorithm steps of the various examples described in the embodiments disclosed herein, the embodiments of this disclosure can be implemented in hardware or a combination of hardware and computer software. Whether a function is executed in hardware or by computer software driving hardware depends on the specific application and design constraints of the technical solution. Those skilled in the art can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this disclosure.

[0190] This disclosure embodiment can divide the part identification device into functional modules according to the above method example. For example, each function can be divided into a separate functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. Optionally, the module division in this disclosure embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0191] This disclosure embodiment can divide the part identification device into functional modules according to the above method example. For example, each function can be divided into a separate functional module, or two or more functions can be integrated into one processing module. The integrated module can be implemented in hardware or as a software functional module. Optionally, the module division in this disclosure embodiment is illustrative and only represents one logical functional division; other division methods may be used in actual implementation.

[0192] This disclosure provides a part identification device for executing the method required by any device in the above-described part identification system. The part identification device may be the part identification device described in this disclosure, or a module within a part identification device; it may also be a chip within a part identification device, or other devices for executing part identification methods; this disclosure does not limit the scope of the application.

[0193] like Figure 10 The diagram shown is a structural schematic of a part identification device provided in an embodiment of this disclosure. The part identification device includes a processing unit 1001 and a communication unit 1002.

[0194] The processing unit 1001 is used to acquire images of multiple feature parts of the part to be identified;

[0195] Optionally, the processing unit 1001 is also used to determine the set of part models corresponding to each of the multiple feature parts based on the images of the multiple feature parts;

[0196] Optionally, the processing unit 1001 is also used to determine the part model of the part to be identified based on the set of part models corresponding to each feature part.

[0197] Optionally, the processing unit 1001 is specifically configured to: perform the following target operation for each feature part based on the images of multiple feature parts, to determine the set of part models corresponding to each feature part; the target operation includes: detecting the image of the target feature part according to a preset image detection algorithm, and determining the first set of part models corresponding to the target feature part; the target feature part is one of the multiple feature parts; extracting the image region of the feature part from the image of the feature part, processing the image region according to a preset image classification algorithm, and determining the second set of part models corresponding to the target feature part; and determining the set of part models corresponding to the target feature part based on the first set of part models and the second set of part models.

[0198] Optionally, the image of the target feature region includes images of the target feature region acquired from multiple viewpoints; the first part model set includes the first part model of the target feature region from multiple viewpoints; the second part model set includes the second part model of the target feature region from multiple viewpoints; the second part model set includes the second part model of the target feature region from multiple viewpoints; the processing unit 1001 is specifically used to: determine the part model set corresponding to the target feature region, including: determining whether the first part model set and the second part model set are the same; if they are the same, then determining whether the part model set corresponding to the target feature region is the first part model set or the second part model set; if they are not the same, then determining the first confidence level corresponding to the first part model of the target feature region from the i-th viewpoint and the second confidence level corresponding to the second part model from the i-th viewpoint; i is a positive integer; the first part model is the part model in the first part model set; the second part model is the part model in the second part model set; comparing the first confidence level and the second confidence level, determining the target confidence level; determining the part model from the i-th viewpoint in the part model set corresponding to the target feature region as the part model corresponding to the target confidence level.

[0199] Optionally, the processing unit 1001 is specifically used to: when the first set of part models and the second set of part models are different, determine the part model of the target feature part from the i-th viewpoint in the set of part models corresponding to the target feature part. Satisfy the following formula:

[0200]

[0201] in, W represents the model number of the first part from the i-th perspective. d The first confidence level; Let W be the model number of the second part from the i-th perspective; c α is the second confidence level; β is the scaling factor of the first part model set; β is the scaling factor of the second part model set; and thres is the threshold for distinguishing between the first part model set and the second part model set.

[0202] Optionally, the processing unit 1001 is specifically used to: determine the intersection of the sets of part models corresponding to each feature part; determine whether the number of part models in the intersection is greater than a preset value; if it is not greater, determine the part models in the intersection as the part models of the part to be identified; if it is greater, determine the part model with the highest confidence in the part models in the intersection as the part model of the part to be identified.

[0203] Optionally, the processing unit 1001 is specifically used for: obtaining the part level of the part to be identified; when the part level of the part to be identified is a first preset level, determining the feature parts to be collected and the image acquisition angle of the part to be identified; acquiring the image of the feature parts to be collected according to the image acquisition angle; when the part level of the part to be identified is a second preset level, determining the feature parts to be collected of the part to be identified; acquiring the image of the feature parts to be collected; when the part level of the part to be identified is a third preset level, randomly acquiring images of multiple feature parts of the part to be identified.

[0204] This disclosure provides a part identification device for executing the method required by any device in the above-described part identification system. The part identification device may be the part identification device described in this disclosure, or a module within a part identification device; it may also be a chip within a part identification device, or other devices for executing part identification methods; this disclosure does not limit the scope of the application.

[0205] This disclosure also provides a computer-readable storage medium storing instructions that, when executed by a computer, perform each step of the method flow shown in the above method embodiments.

[0206] Embodiments of this disclosure provide a computer program product containing instructions that, when executed on a computer, cause the computer to perform the part identification method described in the above method embodiments.

[0207] Embodiments of this disclosure provide a chip including a processor and a communication interface, the communication interface and the processor being coupled together, the processor being used to run computer programs or instructions to implement the part identification method as described in the above method embodiments.

[0208] The computer-readable storage medium may be, for example, but not limited to, an electrical, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or any combination thereof. More specific examples of computer-readable storage media (a non-exhaustive list) include: an electrical connection having one or more wires, a portable computer disk, a hard disk, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), registers, hard disks, optical fibers, compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination of the foregoing, or any other form of computer-readable storage medium in the art. An exemplary storage medium is coupled to a processor, enabling the processor to read information from and write information to the storage medium. Of course, the storage medium may also be a component of the processor. The processor and the storage medium may reside in an application-specific integrated circuit (ASIC). In this embodiment of the disclosure, the computer-readable storage medium can be any tangible medium that contains or stores a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0209] Since the apparatus, devices, computer-readable storage media, and computer program products in the embodiments of this disclosure can be applied to the above methods, the technical effects they can achieve can also be referred to the above method embodiments. The embodiments of this disclosure will not be repeated here.

[0210] The above description is merely a specific embodiment of this disclosure, but the scope of protection of this disclosure is not limited thereto. Any changes or substitutions within the technical scope disclosed in this disclosure should be included within the scope of protection of this disclosure. Therefore, the scope of protection of this disclosure should be determined by the scope of the claims.

Claims

1. A method for identifying parts, characterized in that, The method includes: Acquire images of multiple feature areas of the part to be identified; Based on the images of the multiple feature regions, determine the set of part models corresponding to each feature region; Based on the set of part models corresponding to each feature part, determine the part model of the part to be identified; The step of determining the set of part models corresponding to each of the plurality of feature parts based on the images of the plurality of feature parts includes: Based on the images of the multiple feature regions, perform the following target operation for each feature region to determine the set of part models corresponding to each feature region; The target operation includes: The image of the target feature region is detected according to a preset image detection algorithm, and the first set of part models corresponding to the target feature region is determined; the target feature region is one of the multiple feature regions. Extract the image region of the feature region from the image of the feature region, process the image region according to a preset image classification algorithm, and determine the second part model set corresponding to the target feature region; Based on the first set of part models and the second set of part models, determine the set of part models corresponding to the target feature.

2. The method according to claim 1, characterized in that, The image of the target feature includes images of the target feature acquired from multiple viewpoints; the first part model set includes a first part model of the target feature from the multiple viewpoints; the second part model set includes a second part model of the target feature from the multiple viewpoints. Determining the set of part models corresponding to the target feature based on the first set of part models and the second set of part models includes: Determine whether the first set of part models and the second set of part models are the same; In response to the fact that the first part model set and the second part model set are the same, the part model set corresponding to the target feature part is determined to be either the first part model set or the second part model set; In response to the fact that the first set of part models and the second set of part models are different, a first confidence level corresponding to the first part model of the target feature part under the i-th view and a second confidence level corresponding to the second part model under the i-th view are determined; i is a positive integer; the first part model is the part model in the first set of part models; the second part model is the part model in the second set of part models. By comparing the first confidence level and the second confidence level, the target confidence level is determined; The part model in the i-th view of the part model set corresponding to the target feature part is determined as the part model corresponding to the target confidence level.

3. The method according to claim 2, characterized in that, When the first set of part types and the second set of part types are different, the part type from the i-th viewpoint in the set of part types corresponding to the target feature part. Satisfy the following formula: in, W represents the model number of the first part under the i-th viewpoint; d This represents the first confidence level; W represents the model number of the second part in the i-th viewpoint. c α is the second confidence level; β is the scaling factor of the first part model set; thres is the distinction threshold between the first part model set and the second part model set.

4. The method according to any one of claims 1-3, characterized in that, The step of determining the part model of the part to be identified based on the set of part models corresponding to each feature part includes: Determine the intersection of the sets of part models corresponding to each of the aforementioned feature parts; Determine whether the number of part models in the intersection is greater than a preset value; In response to the fact that the number of part models in the intersection is less than or equal to a preset value, the part models in the intersection are determined as the part models of the part to be identified; In response to the fact that the number of part models in the intersection is greater than a preset value, the part model with the highest confidence in the intersection is determined as the part model of the part to be identified.

5. The method according to any one of claims 1-3, characterized in that, Before acquiring images of multiple feature regions of the part to be identified, the method further includes: Obtain the part grade of the part to be identified; When the part level of the part to be identified is a first preset level, the feature parts to be collected and the image collection angle of the part to be identified are determined. According to the image acquisition perspective, the image of the feature region to be acquired is acquired; If the part level of the part to be identified is the second preset level, determine the feature parts of the part to be identified to be collected; Acquire images of the feature regions to be acquired; When the part to be identified is at the third preset level, images of multiple feature parts of the part to be identified are randomly acquired.

6. A parts identification device, characterized in that, The device includes: a processing unit; The processing unit is used to acquire images of multiple feature parts of the part to be identified; The processing unit is further configured to determine, based on the images of the plurality of feature parts, a set of part models corresponding to each feature part among the plurality of feature parts; The processing unit is further configured to determine the part model of the part to be identified based on the set of part models corresponding to each feature part; The processing unit is specifically used for: Based on the images of the multiple feature regions, perform the following target operation for each feature region to determine the set of part models corresponding to each feature region; The target operation includes: The image of the target feature region is detected according to a preset image detection algorithm, and the first set of part models corresponding to the target feature region is determined; the target feature region is one of the multiple feature regions. Extract the image region of the feature region from the image of the feature region, process the image region according to a preset image classification algorithm, and determine the second part model set corresponding to the target feature region; Based on the first set of part models and the second set of part models, determine the set of part models corresponding to the target feature.

7. The apparatus according to claim 6, characterized in that, The image of the target feature includes images of the target feature acquired from multiple viewpoints; the first part model set includes a first part model of the target feature from the multiple viewpoints; the second part model set includes a second part model of the target feature from the multiple viewpoints. The processing unit is specifically used to: determine whether the first set of part models and the second set of part models are the same; In response to the fact that the first part model set and the second part model set are the same, the part model set corresponding to the target feature part is determined to be either the first part model set or the second part model set; In response to the fact that the first set of part models and the second set of part models are different, a first confidence level corresponding to the first part model of the target feature part under the i-th view and a second confidence level corresponding to the second part model under the i-th view are determined; i is a positive integer; the first part model is the part model in the first set of part models; the second part model is the part model in the second set of part models. By comparing the first confidence level and the second confidence level, the target confidence level is determined; The part model in the i-th view of the part model set corresponding to the target feature part is determined as the part model corresponding to the target confidence level.

8. The apparatus according to claim 7, characterized in that, When the first set of part types and the second set of part types are different, the part type from the i-th viewpoint in the set of part types corresponding to the target feature part. Satisfy the following formula: in, W represents the model number of the first part under the i-th viewpoint; d This represents the first confidence level; W represents the model number of the second part in the i-th viewpoint. c α is the second confidence level; β is the scaling factor of the first part model set; thres is the distinction threshold between the first part model set and the second part model set.

9. The apparatus according to any one of claims 6-8, characterized in that, The processing unit is specifically used for: Determine the intersection of the sets of part models corresponding to each of the aforementioned feature parts; Determine whether the number of part models in the intersection is greater than a preset value; In response to the fact that the number of part models in the intersection is less than or equal to a preset value, the part models in the intersection are determined as the part models of the part to be identified; In response to the fact that the number of part models in the intersection is greater than a preset value, the part model with the highest confidence in the intersection is determined as the part model of the part to be identified.

10. The apparatus according to any one of claims 6-8, characterized in that, The processing unit is specifically used for: Obtain the part grade of the part to be identified; When the part level of the part to be identified is a first preset level, the feature parts to be collected and the image collection angle of the part to be identified are determined. According to the image acquisition perspective, the image of the feature region to be acquired is acquired; If the part level of the part to be identified is the second preset level, determine the feature parts of the part to be identified to be collected; Acquire images of the feature regions to be acquired; When the part to be identified is at the third preset level, images of multiple feature parts of the part to be identified are randomly acquired.

11. A parts identification device, characterized in that, include: A processor and a communication interface; the communication interface is coupled to the processor, the processor being used to run computer programs or instructions to implement the part identification method as described in any one of claims 1-5.

12. A computer-readable storage medium storing instructions, characterized in that, When the computer executes the instruction, the computer performs the part identification method as described in any one of claims 1-5.

Images