Installation inspection method and device, electronic equipment and storage medium

Abstract

The application discloses an installation inspection method and device, electronic equipment and a storage medium. The method comprises the following steps: obtaining a to-be-processed image; the to-be-processed image can reflect at least an installation condition of a first optical network unit (ONU); based on the to-be-processed image, a first model is used to determine an optical fiber joint type of the first ONU, determine a tail fiber disc reservation condition of the first ONU, and identify a product serial number (SN) of the first ONU; and whether the installation of the first ONU is standard is determined by using the determined optical fiber joint type, the determined tail fiber disc reservation condition and the identified SN.

Description

Technical Field

[0001] This application relates to the field of computer technology, and in particular to an installation inspection method, apparatus, electronic device, and storage medium. Background Technology

[0002] An Optical Network Unit (ONU) can provide services such as data, IPTV, and integrated voice access devices (IADs), truly enabling triple-play applications. Using an ONU can effectively improve the uplink bandwidth utilization of the entire communication system. It also allows for the configuration of channel bandwidth according to the network application environment and the characteristics of applicable services, supporting as many end users as possible without affecting communication efficiency and quality, thereby improving network utilization and reducing user costs.

[0003] With the widespread adoption of internet TV, more and more households are installing ONUs (Online Units). To standardize the management of ONU installations and categorize them by resource, it's necessary to photograph the installed ONUs, check their installation accuracy based on these photos, and determine if the currently installed ONU was assigned to the corresponding user by the operator's management system. However, given the large volume of installations, manual verification is time-consuming, labor-intensive, and prone to errors. Therefore, how to automatically verify the standardization of ONU installations has become a pressing technical problem. Summary of the Invention

[0004] To address the related technical problems, embodiments of this application provide an installation inspection method, apparatus, electronic device, and storage medium.

[0005] The technical solution of this application embodiment is implemented as follows:

[0006] This application provides an installation inspection method, including:

[0007] Acquire an image to be processed; the image to be processed should at least reflect the installation status of the first ONU;

[0008] Based on the image to be processed, the fiber optic connector type of the first ONU is determined using the first model, the pigtail coiling status of the first ONU is determined, and the product serial number (SN) of the first ONU is identified.

[0009] Using the determined fiber optic connector type, the determined pigtail coiling status, and the identified serial number (SN), it is determined whether the installation of the first ONU is up to standard.

[0010] In the above scheme, determining the fiber optic connector type of the first ONU based on the image to be processed using the first model includes:

[0011] Using the first model, a first region is determined in the image to be processed; the first region contains the fiber optic connector of the first ONU.

[0012] Using the first model, the probability that the fiber optic connector in the first region is a hot-melt connector is determined, thus obtaining a first probability; and / or, using the first model, the probability that the fiber optic connector in the first region is a cold connector is determined, thus obtaining a second probability;

[0013] The fiber optic connector type of the first ONU is determined based on the first probability and / or the second probability.

[0014] In the above scheme, determining the fiber optic coiling status of the first ONU based on the image to be processed using the first model includes:

[0015] Using the first model, a second region is determined in the image to be processed; the second region contains the fiber tray of the first ONU;

[0016] Using the first model, the probability that there is a tail fiber coiled in the fiber spool in the second region is determined to obtain a third probability; and / or, using the first model, the probability that there is no tail fiber coiled in the fiber spool in the second region is determined to obtain a fourth probability.

[0017] The fiber optic cable retention status of the first ONU is determined based on the third probability and / or the fourth probability.

[0018] In the above scheme, the step of identifying the SN of the first ONU based on the image to be processed using the first model includes at least one of the following:

[0019] Using the first model, a third region is determined in the image to be processed; the third region contains the QR code of the first ONU; based on the third region, a QR code image is extracted from the image to be processed; the QR code image is parsed to obtain the first SN;

[0020] Using the first model, a fourth region is determined in the image to be processed; the fourth region contains the barcode of the first ONU; based on the fourth region, a barcode image is extracted from the image to be processed; the barcode image is parsed to obtain a second SN;

[0021] Using the first model, a fifth region is determined in the image to be processed; the fifth region contains the nameplate of the first ONU; based on the fifth region, a nameplate image is extracted from the image to be processed; text recognition is performed on the nameplate image to obtain a third SN; wherein,

[0022] If at least one of the identified SNs is the same as the target SN, the identified SN is determined to be consistent with the target SN; or, if at least one of the identified SNs is different from the target SN, the identified SN is determined to be inconsistent with the target SN.

[0023] In the above scheme, the step of extracting a QR code image from the image to be processed based on the third region includes:

[0024] The length of the third region is increased by a first distance, and the width of the third region is increased by a second distance to obtain a sixth region; the sixth region is cropped from the image to be processed to obtain the QR code image;

[0025] The parsing of the QR code image includes:

[0026] Based on the aspect ratio of the sixth region, the QR code image is scaled within the first pixel range, and the scaled QR code image is then parsed.

[0027] In the above scheme, the step of extracting the barcode image from the image to be processed based on the fourth region includes:

[0028] The length of the fourth region is increased by a third distance, and the width of the fourth region is increased by a fourth distance to obtain the seventh region; the seventh region is cropped from the image to be processed to obtain the barcode image;

[0029] The step of parsing the barcode image includes:

[0030] The barcode image is subjected to denoising and edge sharpening processing, and the denoised and edge-sharpened barcode image is then parsed.

[0031] In the above scheme, determining whether the installation of the first ONU is standardized by using the determined fiber optic connector type, the determined pigtail coiling status, and the identified serial number (SN) includes:

[0032] Given that the fiber optic connector type is a fusion splice, the pigtail coiling status is pigtail coiling, and the identified SN matches the target SN, determine the installation specifications of the first ONU.

[0033] The installation of the first ONU is determined to be non-standard if at least one of the following conditions is met:

[0034] The determined fiber optic connector type is a cold connector;

[0035] The confirmed situation regarding the retention of pigtail fibers is that there are no pigtail fibers retained.

[0036] The identified SN does not match the target SN.

[0037] This application also provides an installation inspection device, including:

[0038] An acquisition unit is used to acquire an image to be processed; the image to be processed can at least reflect the installation status of the first ONU;

[0039] The first processing unit is used to determine the fiber optic connector type of the first ONU, the pigtail coiling status of the first ONU, and the serial number (SN) of the first ONU based on the image to be processed and using the first model.

[0040] The second processing unit is used to determine whether the installation of the first ONU is up to standard by using the determined fiber optic connector type, the determined pigtail coiling status, and the identified serial number (SN).

[0041] This application also provides an electronic device, including: a processor and a memory for storing a computer program capable of running on the processor.

[0042] When the processor runs the computer program, it executes the steps of any of the above methods.

[0043] This application also provides a storage medium storing a computer program thereon, which, when executed by a processor, implements the steps of any of the above methods.

[0044] The installation inspection method, apparatus, electronic device, and storage medium provided in this application embodiment acquire an image to be processed; the image to be processed can at least reflect the installation status of a first ONU; based on the image to be processed, a first model is used to determine the fiber optic connector type of the first ONU, the pigtail coiling status of the first ONU, and identify the serial number (SN) of the first ONU; using the determined fiber optic connector type, the determined pigtail coiling status, and the identified SN, it is determined whether the installation of the first ONU is standardized. The solution in this application embodiment, based on an image to be processed that at least reflects the installation status of a first ONU, uses a first model to determine the fiber optic connector type of the first ONU, the pigtail coiling status of the first ONU, and identify the SN of the first ONU; using the determined fiber optic connector type, the determined pigtail coiling status, and the identified SN, it is determined whether the installation of the first ONU is standardized; thus, it can automatically inspect whether the ONU installation is standardized based on an image reflecting the ONU installation status; at the same time, through the first model, it can improve the accuracy and generalization ability of inspecting whether the ONU installation is standardized. Attached Figure Description

[0045] Figure 1 This is a schematic flowchart of the installation inspection method according to an embodiment of this application;

[0046] Figure 2 This is a schematic diagram illustrating the process of verifying whether the ONU installation is standardized in an application embodiment of this application;

[0047] Figure 3 These are schematic images illustrating the ONU installation status in the application embodiments of this application;

[0048] Figure 4 This is a schematic diagram of the structure of the inspection device installed in an embodiment of this application;

[0049] Figure 5 This is a schematic diagram of the structure of an electronic device according to an embodiment of this application. Detailed Implementation

[0050] The present application will now be described in further detail with reference to the accompanying drawings and embodiments.

[0051] In related technologies, given the diverse scenarios in which ONUs are used, it's difficult to apply a fixed installation inspection method to all input scenarios. For example, after adjusting parameters for an ONU with a black housing, if the input ONU housing becomes white, the original parameters become inapplicable. Therefore, considering differences in ONU orientation (i.e., photo shooting angle), coiling slot shape, and pigtail color, it's impossible to verify the correctness of ONU installation using a fixed parameter installation inspection method. Furthermore, when photographing installed ONUs, the nameplate label area is not specifically photographed in order to capture areas such as pigtail coiling and fiber optic connectors. Also, the nameplate area of ​​some ONUs is small, resulting in poor text clarity. If only text recognition is used to extract the serial number (SN) from the nameplate, both the recognition efficiency and accuracy are poor.

[0052] Based on this, in various embodiments of this application, based on the image to be processed that at least reflects the installation status of the first ONU, a first model is used to determine the fiber optic connector type of the first ONU, the pigtail coiling status of the first ONU, and the serial number (SN) of the first ONU. Using the determined fiber optic connector type, the determined pigtail coiling status, and the identified SN, it is determined whether the installation of the first ONU is up to standard. In this way, the installation status of the ONU can be automatically checked based on the image reflecting the installation status of the ONU. At the same time, the first model can improve the accuracy and generalization ability of checking whether the installation of the ONU is up to standard.

[0053] This application provides an installation inspection method, such as... Figure 1 As shown, the method includes:

[0054] Step 101: Obtain the image to be processed;

[0055] Here, the image to be processed can at least reflect the installation status of the first ONU;

[0056] Step 102: Based on the image to be processed, using the first model, determine the fiber optic connector type of the first ONU, determine the pigtail coiling status of the first ONU, and identify the SN of the first ONU;

[0057] Step 103: Using the determined fiber optic connector type, the determined pigtail coiling status, and the identified serial number (SN), determine whether the installation of the first ONU is up to standard.

[0058] It should be noted that the installation inspection method provided in this application embodiment can be applied to electronic devices, such as servers, shooting devices, robots, etc. The specific type of electronic device can be set according to requirements, and this application embodiment does not limit it in this regard.

[0059] In step 101, in practical applications, the electronic device can acquire the image to be processed from other devices; alternatively, the electronic device can also acquire the image to be processed locally. The specific method by which the electronic device acquires the image to be processed can be set according to requirements, and this embodiment does not limit this.

[0060] In practical applications, in order to reflect the installation status of the first ONU, the image to be processed must at least include the fiber optic cable tray area, the fiber optic connector area, and the nameplate label area of ​​the first ONU.

[0061] In practical applications, since the fiber tray area, fiber connector area, and nameplate label area of ​​the ONU are usually located on the back, the image to be processed may include the back area of ​​the first ONU.

[0062] In step 102, in practical applications, images reflecting the installation status of different types of ONUs can be pre-captured to construct a model training dataset. Using this model training dataset, the first model can be trained using a deep learning algorithm. Here, the specific deep learning algorithm used to train the first model can be set according to requirements, such as a convolutional neural network (CNN).

[0063] The images in the model training dataset may have at least one of the following distinct features:

[0064] Different resolutions;

[0065] Different levels of clarity;

[0066] The angle at which the ONU is photographed varies; that is, the orientation of the ONU in the image varies; the orientation can include four types: up, down, left, and right.

[0067] ONUs come in different types of fiber optic connectors; they can be either hot-melt connectors or cold-melt connectors.

[0068] ONUs come in different colors; for example, black, white, yellow, blue, etc.

[0069] ONUs come in different types of fiber trays; that is, the shape of the fiber trays is different, such as circular, elliptical, or approximately rectangular.

[0070] The condition of the pigtail coiling in the ONU varies; that is, there is pigtail coiling or no pigtail coiling in the coiling groove.

[0071] The ONU models are different;

[0072] The ONU box comes in different colors; for example, black, white, etc.

[0073] In practical applications, the images reflecting the installation conditions of various types of ONUs contained in the model training dataset need to be distributed proportionally. For example, the number of images of ONUs with thermal fusion connectors should be as equal as possible to the number of images of ONUs with cold connectors. In this way, the inaccuracy of the output results of the first model due to the imbalance of the model training dataset can be avoided, that is, the accuracy of the output results of the first model can be improved.

[0074] In practical applications, the first model can be used to determine the fiber optic connector area of ​​the first ONU in the image to be processed, and to determine whether the fiber optic connector type in the area is a hot-melt connector or a cold connector.

[0075] Based on this, in one embodiment, determining the fiber optic connector type of the first ONU using the first model based on the image to be processed may include:

[0076] Using the first model, a first region is determined in the image to be processed; the first region contains the fiber optic connector of the first ONU.

[0077] Using the first model, the probability that the fiber optic connector in the first region is a hot-melt connector is determined, thus obtaining a first probability; and / or, using the first model, the probability that the fiber optic connector in the first region is a cold connector is determined, thus obtaining a second probability;

[0078] The fiber optic connector type of the first ONU is determined based on the first probability and / or the second probability.

[0079] In practical applications, the image to be processed is input into the first model. The first model can output the normalized coordinates of two points of the rectangular outline of the first region, such as the normalized coordinates of the top left and bottom right vertices; and output the first probability and / or the second probability. When the first model only outputs the first probability, the fiber optic connector type of the first ONU can be determined to be a hot-melt connector. When the first model only outputs the second probability, the fiber optic connector type of the first ONU can be determined to be a cold connector. When the first model outputs both the first and second probabilities, and the first probability is greater than the second probability, the fiber optic connector type of the first ONU can be determined to be a hot-melt connector. When the first model outputs both the first and second probabilities, and the second probability is greater than the first probability, the fiber optic connector type of the first ONU can be determined to be a cold connector.

[0080] In practical applications, the first model can output first anomaly information even if the first region is not detected. The first anomaly information can indicate that the first model cannot identify (i.e., cannot determine) the fiber optic connector type of the first ONU. In this case, the fiber optic connector type of the first ONU can be determined manually, or a new image that can at least reflect the installation status of the first ONU can be acquired, and the first ONU can be re-inspected for installation.

[0081] In practical applications, if the fiber optic connector of the first ONU is a fusion splice, the installation specifications of the fiber optic connector of the first ONU can be determined; if the fiber optic connector of the first ONU is a cold splice, the installation of the fiber optic connector of the first ONU can be determined as non-standard.

[0082] In practical applications, the first model can also be used to determine the fiber tray area of ​​the first ONU in the image to be processed, and to determine whether there is a pigtail coil in the area.

[0083] Based on this, in one embodiment, determining the fiber optic cable retention status of the first ONU using the first model based on the image to be processed may include:

[0084] Using the first model, a second region is determined in the image to be processed; the second region contains the fiber tray of the first ONU;

[0085] Using the first model, the probability that there is a tail fiber coiled in the fiber spool in the second region is determined to obtain a third probability; and / or, using the first model, the probability that there is no tail fiber coiled in the fiber spool in the second region is determined to obtain a fourth probability.

[0086] The fiber optic cable retention status of the first ONU is determined based on the third probability and / or the fourth probability.

[0087] In practical applications, the image to be processed is input into the first model. The first model can output the normalized coordinates of two points of the rectangular outline of the second region, such as the normalized coordinates of the two points of the upper right corner vertex and the lower left corner vertex; and output the third probability and / or the fourth probability. When the first model only outputs the third probability, it can be determined that the fiber optic cable of the first ONU is coiled. When the first model only outputs the fourth probability, it can be determined that the fiber optic cable of the first ONU is not coiled. When the first model outputs both the third and fourth probabilities, and the third probability is greater than the fourth probability, it can be determined that the fiber optic cable of the first ONU is coiled. When the first model outputs both the third and fourth probabilities, and the fourth probability is greater than the third probability, it can be determined that the fiber optic cable of the first ONU is not coiled.

[0088] In practical applications, the first model can output second anomaly information even if the second region is not detected. The second anomaly information indicates that the first model cannot identify (i.e., cannot determine) whether there is a coiled fiber in the fiber tray of the first ONU. In this case, the presence of a coiled fiber in the fiber tray of the first ONU can be manually determined, or a new image that at least reflects the installation status of the first ONU can be acquired, and the first ONU can be re-inspected for installation.

[0089] In practical applications, if the first ONU has a coiled fiber optic cable, the fiber optic cable installation of the first ONU can be determined to be up to standard; if the first ONU has no coiled fiber optic cable, the fiber optic cable installation of the first ONU can be determined to be non-standard.

[0090] In practical applications, since the QR code, barcode, and nameplate of the first ONU all contain the SN information of the first ONU, the first model can also be used to obtain at least one SN identification result based on at least one of the QR code, barcode, and nameplate of the first ONU.

[0091] Based on this, in one embodiment, identifying the SN of the first ONU using the first model based on the image to be processed may include at least one of the following:

[0092] Using the first model, a third region is determined in the image to be processed; the third region contains the QR code of the first ONU; based on the third region, a QR code image is extracted from the image to be processed; the QR code image is parsed to obtain the first SN;

[0093] Using the first model, a fourth region is determined in the image to be processed; the fourth region contains the barcode of the first ONU; based on the fourth region, a barcode image is extracted from the image to be processed; the barcode image is parsed to obtain a second SN;

[0094] Using the first model, a fifth region is determined in the image to be processed; the fifth region contains the nameplate of the first ONU; based on the fifth region, a nameplate image is extracted from the image to be processed; text recognition is performed on the nameplate image to obtain a third SN; wherein,

[0095] If at least one of the identified SNs is the same as the target SN, the identified SN is determined to be consistent with the target SN; or, if at least one of the identified SNs is different from the target SN, the identified SN is determined to be inconsistent with the target SN.

[0096] In practical applications, the image to be processed is input into the first model, and the first model can output the normalized coordinates of two points of the rectangular outline of at least one of the third region, the fourth region, and the fifth region, such as the normalized coordinates of the top left corner vertex and the bottom right corner vertex.

[0097] In practical applications, after obtaining the normalized coordinates of two points of the rectangular outline of at least one of the third, fourth, and fifth regions output by the first model, the electronic device can identify the serial number (SN) of the first ONU based on at least one of the third, fourth, and fifth regions, thus obtaining at least one SN among the first SN, second SN, and third SN. If the SN of the first ONU cannot be identified, the electronic device can output a third abnormality message to prompt the user that the SN of the first ONU cannot be identified. In this case, the SN of the first ONU can be manually identified, or a new image reflecting at least the installation status of the first ONU can be acquired, and the first ONU can be re-inspected for installation.

[0098] The situation where the SN of the first ONU cannot be identified may include at least one of the following:

[0099] Unable to parse the QR code image;

[0100] Unable to parse the barcode image;

[0101] Text recognition cannot be performed on the nameplate image; that is, no characters can be recognized in the nameplate image, and only an empty string is obtained.

[0102] In practical applications, the identified at least one SN can be understood as at least one of the first SN, the second SN, and the third SN. The target SN can be understood as the SN of the ONU assigned to the corresponding user by the operator's management system. In other words, the purpose of identifying the SN of the first ONU is to determine whether the first ONU is an ONU assigned to the corresponding user by the operator's management system.

[0103] In practical applications, the identified SN (i.e., the first SN, the second SN, or the third SN) can be matched with the target SN to determine whether the identified SN is the same as the target SN. If any one of the first SN, the second SN, or the third SN is the same as the target SN, it can be determined that the identified SN is consistent with the target SN, that is, the first ONU is determined to be an ONU assigned to the corresponding user by the operator's management system. If the first SN, the second SN, and the third SN are all different from the target SN, it can be determined that the identified SN is inconsistent with the target SN, that is, the first ONU is determined not to be an ONU assigned to the corresponding user by the operator's management system.

[0104] In practical applications, to further improve the accuracy of SN recognition, after determining the third region, it can be expanded outward based on a preset distance (i.e., length) to ensure that a complete QR code can be extracted from the image to be processed. Simultaneously, when parsing the QR code image, it can be scaled to improve the success rate of parsing.

[0105] Based on this, in one embodiment, the step of extracting a QR code image from the image to be processed based on the third region may include:

[0106] The length of the third region is increased by a first distance, and the width of the third region is increased by a second distance to obtain a sixth region; the sixth region is cropped from the image to be processed to obtain the QR code image;

[0107] The parsing of the QR code image may include:

[0108] Based on the aspect ratio of the sixth region, the QR code image is scaled within the first pixel range, and the scaled QR code image is then parsed.

[0109] In practical applications, the first distance, the second distance, and the first pixel range can be set according to requirements; the first distance and the second distance can be the same or different. For example, the first distance is 25% of the length of the third region, the second distance is 15% of the width of the third region, and the first pixel range is 100 pixels to 300 pixels.

[0110] In practical applications, a pre-defined QR code parsing library, such as Zbar, can be used to parse the QR code image after scaling.

[0111] In practical applications, since the parsing result of the QR code image may contain other fields besides the SN, the first SN can be extracted from the parsing result of the QR code image using regular expression matching.

[0112] In practical applications, to further improve the accuracy of SN recognition, after determining the fourth region, it can be expanded outward based on a preset distance (i.e., length) to ensure that the complete barcode can be extracted from the image to be processed. Simultaneously, when parsing the barcode image, denoising and edge sharpening processes can be performed to improve the success rate of parsing the barcode image.

[0113] Based on this, in one embodiment, the step of extracting a barcode image from the image to be processed based on the fourth region may include:

[0114] The length of the fourth region is increased by a third distance, and the width of the fourth region is increased by a fourth distance to obtain the seventh region; the seventh region is cropped from the image to be processed to obtain the barcode image;

[0115] The parsing of the barcode image may include:

[0116] The barcode image is subjected to denoising and edge sharpening processing, and the denoised and edge-sharpened barcode image is then parsed.

[0117] In practical applications, the third distance and the fourth distance can be set according to requirements; the third distance and the fourth distance can be the same or different.

[0118] In practical applications, the method for denoising the barcode image can be set according to requirements, such as Gaussian blur.

[0119] In practical applications, the method for edge sharpening of the barcode image can also be set according to requirements, such as the edge sharpening method based on the Laplacian operator.

[0120] In practical applications, a preset barcode parsing library, such as Zbar, can be used to parse the barcode image after denoising and edge sharpening.

[0121] In practical applications, during text recognition of the nameplate image, text regions can be detected first. The specific method for detecting text regions can be set according to requirements. For example, OpenCV's Deep Neural Networks (DNN) model can be used to detect text regions in the nameplate image. After determining the text regions, multiple characters to be recognized and their spatial relationships can be determined within the text regions. These spatial relationships are then used to sort the characters, obtaining a recognition order. An OpenCV Convolutional Recurrent Neural Network (CRNN) model is then used to recognize the characters according to their recognition order, yielding a text recognition result. The text recognition result includes at least a third serial number (SN). This avoids errors in the recognition order of the characters, thus improving the accuracy of the recognized SN code.

[0122] For step 103, in one embodiment, determining whether the installation of the first ONU is standard using the determined fiber optic connector type, the determined pigtail coiling status, and the identified serial number (SN) may include:

[0123] Given that the fiber optic connector type is a fusion splice, the pigtail coiling status is pigtail coiling, and the identified SN matches the target SN, determine the installation specifications of the first ONU.

[0124] The installation of the first ONU is determined to be non-standard if at least one of the following conditions is met:

[0125] The determined fiber optic connector type is a cold connector;

[0126] The confirmed situation regarding the retention of pigtail fibers is that there are no pigtail fibers retained.

[0127] The identified SN does not match the target SN.

[0128] The installation inspection method provided in this application embodiment acquires an image to be processed; the image to be processed at least reflects the installation status of a first ONU; based on the image to be processed, a first model is used to determine the fiber optic connector type of the first ONU, the pigtail coiling status of the first ONU, and identify the serial number (SN) of the first ONU; using the determined fiber optic connector type, the determined pigtail coiling status, and the identified SN, it is determined whether the installation of the first ONU is standardized. The solution in this application embodiment, based on an image to be processed that at least reflects the installation status of a first ONU, uses a first model to determine the fiber optic connector type of the first ONU, the pigtail coiling status of the first ONU, and identify the SN of the first ONU; using the determined fiber optic connector type, the determined pigtail coiling status, and the identified SN, it is determined whether the installation of the first ONU is standardized; thus, it can automatically inspect whether the ONU installation is standardized based on an image reflecting the ONU installation status; at the same time, the first model can improve the accuracy and generalization ability of inspecting whether the ONU installation is standardized.

[0129] Furthermore, the installation verification method provided in this application utilizes the first model to obtain at least one of the first SN, second SN, and third SN based on at least one of the QR code, barcode, and nameplate of the first ONU. If any one of the first SN, second SN, and third SN is the same as the target SN, it is determined that the identified SN is consistent with the target SN, i.e., the first ONU is determined to be an ONU assigned to the corresponding user by the operator's management system. If all three SNs are different from the target SN, it is determined that the identified SN is inconsistent with the target SN, i.e., the first ONU is determined not to be an ONU assigned to the corresponding user by the operator's management system. Thus, compared with using a single method to identify the SN, it can reduce misjudgments and improve the accuracy of SN identification.

[0130] The present application will be further described in detail below with reference to application examples.

[0131] In this application example, for the ONU installation scenario of broadband installation and maintenance by operators, in order to ensure that a large number of ONUs can be installed in accordance with the specifications and to confirm that the currently installed ONU is the ONU assigned to the corresponding user by the operator's management system, it is necessary to take pictures of the ONU installation and upload them to the ONU installation inspection system. The ONU installation pictures need to include the ONU's fiber optic connector area, fiber tray area, and nameplate label area.

[0132] In this application example, the ONU installation inspection system checks the following three items:

[0133] 1) The type of the fiber optic connector of the ONU, that is, the fiber optic connector of the ONU is a heat fusion connector or a cold connector;

[0134] 2) The situation of the coiled optical fiber pigtail of the ONU, that is, there is a coiled optical fiber pigtail or no coiled optical fiber pigtail in the fiber optic pigtail slot of the ONU;

[0135] 3) The consistency of the unique identifier of the ONU, that is, the SN of the ONU is consistent or inconsistent with the target SN.

[0136] In this application embodiment, the ONU installation inspection system combines the detection results of the above three detection items to judge whether the installation of the ONU is standard; specifically, when the fiber optic connector of the ONU is a heat fusion connector, there is a coiled optical fiber pigtail in the fiber optic pigtail slot of the ONU, and the SN of the ONU is consistent with the target SN, it can be determined that the installation of the ONU is standard, that is, it is determined that the installation of the ONU is qualified; in other cases, it is determined that the installation of the ONU is unqualified; thus, the ONU installation ability of the installation and maintenance personnel can be evaluated. Specifically, when one of the following conditions is met, it can be determined that the installation of the ONU is unqualified:

[0137] The fiber optic connector of the ONU is a cold connector;

[0138] There is no coiled optical fiber pigtail in the fiber optic pigtail slot of the ONU;

[0139] The SN of the ONU is inconsistent with the target SN.

[0140] In this application embodiment, the taking of the ONU installation picture can be manually completed by the installation and maintenance personnel or automatically taken by the intelligent robot.

[0141] In this application embodiment, the ONU installation inspection system is implemented based on the target detection model and the digital image processing method. Specifically, as Figure 2 shown, the process of the ONU installation inspection system for inspecting whether the installation of the ONU is standard can include the following steps:

[0142] Step 201: Input the ONU installation picture (that is, the above-mentioned image to be processed); then execute Step 202;

[0143] Step 202: Use the trained target detection model (that is, the above-mentioned first model) to perform seven types of target detections: cold connector, heat fusion connector, coiled optical fiber pigtail, no coiled optical fiber pigtail, two-dimensional code, bar code and nameplate; then execute Step 203;

[0144] Step 203: According to the cold connector detection result (that is, the above-mentioned second probability) and the heat fusion connector detection result (that is, the above-mentioned first probability), judge whether the fiber optic connector detection item of the ONU is qualified; then execute Step 204;

[0145] Step 204: Based on the results of the fiber optic coiling test (i.e., the third probability mentioned above) and the results of the fiber optic coiling test (i.e., the fourth probability mentioned above), determine whether the ONU's fiber optic coiling test is qualified; then proceed to step 205.

[0146] Step 205: Use the target detection model to locate and capture the QR code area (i.e., the third area mentioned above), the barcode area (i.e., the fourth area mentioned above), and the nameplate area (i.e., the fifth area mentioned above); perform image processing on the screenshot to identify the SN; output the three identified SNs. If any SN matches the target SN, the unique identifier detection item of the ONU is determined to be qualified; then proceed to step 206.

[0147] Step 206: Based on the test results of the three test items, determine whether the installation of the ONU in the current ONU installation picture is standardized.

[0148] In step 201, since the ONU's fiber tray area, fiber optic connector area, and nameplate label area are usually located on the back, the ONU installation picture can include the back area of ​​the ONU.

[0149] In step 202, the object detection model needs to be trained first. This training process may include the following steps:

[0150] Step 2021: Collect installation images of the ONU and build a model training dataset; then proceed to step 2022.

[0151] Step 2022: Construct a convolutional neural network to train an object detection model for seven object categories; then proceed to step 2023:

[0152] Step 2023: Training complete, output the optimal object detection model.

[0153] In step 2021, such as Figure 3As shown, we can try to filter ONU installation images from the collected images to identify different types of ONUs, such as those with different orientations (i.e., the angle at which the ONU is photographed), different pigtail colors, different fiber optic cable tray types, different ONU housing colors, and different ONU fiber optic connector types. Simultaneously, for multiple cases of the same feature, we should try to select the same number of images; for example, the number of images of thermal connectors should be as similar as the number of images of cold connectors. This can help avoid inaccurate detection results due to imbalanced training data. For the selected ONU installation images, we can use rectangles to mark the locations of cold connector areas, thermal connector areas, areas with pigtail coils, areas without pigtail coils, QR code areas, barcode areas, and nameplate areas in each image, generating corresponding image annotation files. For the multiple selected ONU installation images and their corresponding annotation files, we can divide them into model training datasets, model validation datasets, and model testing datasets according to a preset ratio (e.g., 8:1:1). The model training dataset is used to train the object detection model and needs to include installation images of the ONU and the corresponding annotation files for each image; the model validation dataset is used to validate the training results of the object detection model and needs to include installation images of the ONU and the corresponding annotation files for each image; the model test dataset is used to test the object detection model and may not include annotation files.

[0154] In step 2022, a convolutional neural network can be constructed to train and optimize the target detection model for seven types of targets (i.e., cold connectors, hot-melt connectors, coiled fiber optic cables, coiled fiber optic cables, QR codes, barcodes, and nameplates). Specifically, this application embodiment uses the Faster Region-based Convolutional Neural Network (FASTER-RCNN) algorithm. The Region Proposal Network (RPN) branch of this algorithm can extract and merge candidate boxes into the deep network, achieving end-to-end fast target detection. During the training and optimization of the target detection model, parameters such as the number of training iterations and the learning rate can be set. The model training dataset from step 2021 is used to train the parameters of the target detection model, and the model validation dataset and model test dataset are used to verify the training effect of the target detection model.

[0155] In step 202, the ONU installation image of the installation inspection system currently input to the ONU is input into the trained target detection model. The target detection model can locate seven types of target areas in the image: cold joint area, hot melt head area, area with pigtail coil, area without pigtail coil, QR code area, barcode area, and nameplate area. The output of the target detection model includes the normalized coordinate information (x1, y1) and (x2, y2) of the upper left and lower right corner vertices of the rectangular border (i.e., contour) of each target area.

[0156] In step 203, the output of the target detection model also includes cold joint detection results and / or hot fusion head detection results; the cold joint detection result is the probability value corresponding to the cold joint area (i.e., the second probability), that is, the probability that the ONU's fiber optic connector is a cold joint; the hot fusion head detection result is the probability value corresponding to the hot fusion head area (i.e., the first probability), that is, the probability that the ONU's fiber optic connector is a hot fusion head. When the target detection model only outputs the cold connector detection result, the ONU installation inspection system can determine that the ONU's fiber optic connector is a cold connector; when the target detection model only outputs the hot fusion connector detection result, the ONU installation inspection system can determine that the ONU's fiber optic connector is a hot fusion connector; when the target detection model outputs both the cold connector and hot fusion connector detection results, the ONU installation inspection system takes the detection result with the highest probability value among the two detection results as the ONU's fiber optic connector type; when the ONU's fiber optic connector is a cold connector, the ONU installation inspection system determines that the ONU's fiber optic connector detection item is unqualified; when the ONU's fiber optic connector is a hot fusion connector, the ONU installation inspection system determines that the ONU's fiber optic connector detection item is qualified.

[0157] In practical applications, if the target detection model does not detect the cold joint area and the hot fusion joint area, the ONU installation inspection system can end the current ONU installation inspection process and return information that the fiber optic connector type of the ONU cannot be identified (i.e., the first abnormal information mentioned above).

[0158] In step 204, the output of the target detection model also includes a detection result of fiber optic coiling and / or a detection result of no fiber optic coiling; the detection result of fiber optic coiling is the probability value corresponding to the area with fiber optic coiling (i.e., the third probability), which is the probability that there is fiber optic coiling in the ONU's coiling slot; the detection result of no fiber optic coiling is the probability value corresponding to the area without fiber optic coiling (i.e., the fourth probability), which is the probability that there is no fiber optic coiling in the ONU's coiling slot. When the target detection model only outputs a result indicating that there is a coiled fiber, the ONU installation inspection system can determine that there is a coiled fiber in the ONU's fiber tray. When the target detection model only outputs a result indicating that there is no coiled fiber, the ONU installation inspection system can determine that there is no coiled fiber in the ONU's fiber tray. When the target detection model outputs both results indicating that there is a coiled fiber and results indicating that there is no coiled fiber, the ONU installation inspection system takes the result with the highest probability value as the ONU's fiber coil status. When there is a coiled fiber in the ONU's fiber tray, the ONU installation inspection system determines that the ONU's fiber coil inspection item is qualified. When there is no coiled fiber in the ONU's fiber tray, the ONU installation inspection system determines that the ONU's fiber coil inspection item is unqualified.

[0159] In practical applications, if the target detection model does not detect a fiber optic coiled area or a non-fiber optic coiled area, the ONU installation inspection system can end the current ONU installation inspection process and return information that the fiber optic coiled status of the ONU cannot be identified (i.e., the second abnormal information mentioned above).

[0160] In step 205, the target detection model is used to locate and capture the QR code area, barcode area, and nameplate area to obtain QR code screenshots (i.e., the QR code image mentioned above), barcode screenshots (i.e., the barcode image mentioned above), and nameplate screenshots (i.e., the nameplate image mentioned above), which serve as input images for the ONU installation inspection system to perform SN identification in the future.

[0161] Specifically, the ONU installation inspection system can use the QR code area located by the target detection model to extract a screenshot of the QR code from the ONU installation image. Since the rectangular outline of the QR code area located by the target detection model may not completely contain the ONU's QR code, it is necessary to appropriately expand the QR code area located by the target detection model. Here, according to the length-to-width ratio of the QR code area located by the target detection model, combined with a preset expansion ratio parameter α (the value of α can be set according to requirements, such as 15%, 25%, etc.), the QR code area can be expanded outward by a corresponding distance in all directions (i.e., the first distance and the second distance mentioned above; the method of determining the expansion distance using α is described later in this application embodiment) to obtain a new QR code area (i.e., the sixth area mentioned above). Subsequently, the ONU installation verification system can extract a QR code screenshot from the ONU installation image based on the new QR code area, fix the aspect ratio of the QR code screenshot, and scale the short side of the QR code screenshot within a preset pixel range (i.e., the aforementioned first pixel range). The scaled QR code screenshot is then sequentially input into the parsing library for parsing until a QR code parsing result is successfully obtained. At this point, the traversal ends and the QR code parsing result is recorded (i.e., stored). Otherwise, the traversal of all scaling dimensions continues.

[0162] The ONU installation inspection system can also use the barcode area located by the target detection model to extract a barcode screenshot from the ONU installation image. Since the rectangular outline of the barcode area located by the target detection model may not completely encompass the ONU barcode, it is necessary to appropriately expand the barcode area located by the target detection model. Here, the barcode area can be expanded outwards vertically and horizontally by a corresponding distance (i.e., the third and fourth distances mentioned above; the method for determining the expansion distance using α is described later in this application embodiment) according to the length-to-width ratio of the barcode area located by the target detection model, to obtain a new barcode area (i.e., the seventh area mentioned above). Afterwards, the ONU installation inspection system can extract a barcode screenshot from the ONU installation image based on the new barcode area.

[0163] In practical applications, due to the elongated shape and column-based pixel distribution of barcodes, filtering and edge sharpening are necessary to improve the success rate of barcode screenshot parsing. Here, Gaussian blur is chosen for denoising the barcode screenshot because it preserves image contours and key features well. Simultaneously, the Laplacian operator, as a second-order differential operator, is more effective at refining fine image details and can enhance the edges of the barcode screenshot; therefore, this application uses a Laplacian-based edge sharpening method for edge sharpening. Afterwards, the ONU's installation verification system can input the processed barcode screenshot into the parsing library for parsing and record (i.e., store) the barcode parsing results.

[0164] In this application embodiment, the expansion distance of the QR code screenshot and the barcode screenshot can be determined by using the expansion ratio parameter α through the following steps: Assuming the length of the ONU installation image is H and the width is W, and the percentage coordinates of the upper left corner vertex and the lower right corner vertex of the rectangular border of the corresponding target area (i.e., the QR code area and the barcode area) output by the target detection model are (x1, y1) and (x2, y2), the width w0 of the corresponding target area can be calculated by the following formula:

[0165] w0=W×(x2-x1) (1)

[0166] The length h0 of the corresponding target region is calculated using the following formula:

[0167] h0=H×(y2-y1) (2)

[0168] Based on formulas (1) and (2), and combined with α, the coordinates of the upper left corner vertex of the expanded screenshot can be determined as (W·x1-α·w0, H·y1-α·h0) and the coordinates of the lower right corner vertex as (W·x2+α·w0, H·y2+α·h0). The ONU installation inspection system can capture QR code or barcode screenshots based on the expanded screenshot border. The length of the screenshot can be represented as h1 and the width can be represented as w1.

[0169] In this application embodiment, Gaussian blur is used to filter and denoise the barcode screenshot. Specifically, Gaussian blur obtains the values ​​of several pixels surrounding the pixel to be processed in the barcode screenshot, and calculates a new value for the pixel by adding them with different weights using the following formula:

[0170] O(i,j)=∑ X,Y I(i+x,j+y)·G(x,y) (3)

[0171] Where I(i,j) represents the pixel before filtering; O(i,j) represents the pixel after filtering; G represents the weight matrix of size X×Y; the weights are calculated using a Gaussian function, the expression of which is as follows:

[0172]

[0173] Where G(x, y) represents the weights of the pixels surrounding pixel I(i, j); x and y represent the relative coordinates of the surrounding pixels to the center pixel; and σ represents the blur radius.

[0174] In this application embodiment, an edge sharpening method based on the Laplacian operator is used to sharpen the edges of a barcode screenshot. Specifically, the sharpening filter uses the derivative of the neighborhood as an operator to increase the difference between pixels in the neighborhood, making the abrupt changes in the barcode screenshot more obvious. Here, the sharpening effect is to enhance the edges and contours of the image. The Laplacian operator is the simplest isotropic differential operator, which can be understood as an isotropic filter; the response of this filter is independent of the direction of abrupt changes in the image. In other words, an isotropic filter is rotation-invariant, meaning that rotating the barcode screenshot before filtering yields the same result as filtering the barcode screenshot first and then rotating it.

[0175] In practical applications, the Laplace transform of a binary graph function f(x, y) can be defined by the following formula:

[0176]

[0177] Since derivatives of any order are linear operations, the Laplace transform is also a linear operator. To better suit digital image processing, equation (5) needs to be expressed in discrete form. The second-order partial derivative in the x-direction is defined by the following formula:

[0178]

[0179] Accordingly, the second-order partial differential in the y-direction is defined by the following formula:

[0180]

[0181] Adding the two components of formula (6) and formula (7), we get the following formula:

[0182]

[0183] To achieve the image sharpening effect, the changes in formula (8) can be superimposed onto the barcode screenshot using the following formula:

[0184]

[0185] In step 205, the ONU installation inspection system can also utilize the nameplate area located by the target detection model to extract a screenshot of the nameplate from the ONU installation image and recognize the text in the screenshot. Specifically, firstly, the ONU installation inspection system can use OpenCV's DNN model to detect text regions in the nameplate screenshot. For the returned text detection boxes (i.e., the detection boxes corresponding to the text to be recognized mentioned above), they are sorted according to their spatial position to determine the recognition order, avoiding errors in the output text recognition results due to incorrect order of different text detection boxes. Then, the ONU installation inspection system can call OpenCV's CRNN model to perform text recognition sequentially, combining the text recognition results corresponding to multiple text detection boxes (i.e., splicing them according to their spatial position) to obtain the text recognition result of the nameplate screenshot.

[0186] In practical applications, to obtain the SN (Signal Serial Number), the ONU installation verification system can perform different processing on the QR code parsing results, barcode parsing results, and text recognition results from the nameplate screenshot. Specifically, since the QR code parsing results contain many different fields, and this application embodiment only needs to extract the SN field to verify the consistency of the ONU's unique identifier, the ONU installation verification system can use regular expression matching to extract the SN-related string from the QR code parsing results. Since the barcode parsing results and the text recognition results from the nameplate screenshot only contain the SN, the ONU installation verification system does not need to perform further processing on these results.

[0187] In practical applications, if the serial number (SN) cannot be obtained, the ONU installation verification system can terminate the current ONU installation verification process and return information indicating that the unique identifier consistency of the ONU cannot be detected (i.e., the third abnormal information mentioned above). Here, the situation where the SN cannot be obtained can include at least one of the following:

[0188] Unable to parse the QR code;

[0189] Unable to parse the barcode;

[0190] Unable to recognize text in nameplate screenshots, for example, the text recognition result of the nameplate screenshot is an empty string.

[0191] In practical applications, for the three SNs identified based on the text recognition results of the QR code parsing results, barcode parsing results, and nameplate screenshots, the identified SNs can be matched with the target SNs. If any of the identified SNs can be successfully matched with the target SN (i.e., the identified SNs are the same as the target SNs), the unique identifier detection item of the ONU can be determined to be qualified.

[0192] The solution provided in this application embodiment has the following advantages:

[0193] 1) The ONU installation inspection system in this application embodiment utilizes a deep learning-based target detection network model to perform batch inspections on static ONU installation images. It classifies, extracts, and abstracts features from seemingly disparate and complex ONU installation scenarios caused by features such as ONU orientation and appearance. This allows for the rapid and accurate determination of the ONU's fiber optic connector type and pigtail coiling status, as well as whether the ONU's serial number (SN) in the image matches the target SN, thereby verifying whether the ONU installation is up to standard. In other words, the deep learning-based target detection network model improves the accuracy and generalization ability of verifying whether the ONU installation is up to standard.

[0194] 2) The QR code and barcode areas were appropriately expanded to ensure the integrity of their structures. Simultaneously, the QR code screenshots were subjected to a fixed aspect ratio scaling transformation, and the barcode screenshots were subjected to Gaussian filtering and Laplacian sharpening. This improves the success rate of parsing the QR code and barcode screenshots, thereby increasing the accuracy of the identified serial number (SN).

[0195] 3) SN identification is supported by three identification methods: QR code parsing, barcode parsing, and nameplate text detection and recognition. This completes the consistency detection between the ONU unique identifier and the target SN. Compared with the use of a single SN identification method, it can greatly improve the accuracy of comparison (i.e. consistency detection) and reduce the occurrence of misjudgment and wrong judgment.

[0196] To implement the method of the embodiments of this application, the embodiments of this application also provide an installation inspection device, such as... Figure 4 As shown, the device includes:

[0197] The acquisition unit 401 is used to acquire an image to be processed; the image to be processed can at least reflect the installation status of the first ONU;

[0198] The first processing unit 402 is used to determine the fiber optic connector type of the first ONU, the pigtail coiling status of the first ONU, and identify the SN of the first ONU based on the image to be processed and using the first model.

[0199] The second processing unit 403 is used to determine whether the installation of the first ONU is up to standard by using the determined fiber optic connector type, the determined pigtail coiling status and the identified SN.

[0200] In one embodiment, the first processing unit 402 is specifically used for:

[0201] Using the first model, a first region is determined in the image to be processed; the first region contains the fiber optic connector of the first ONU.

[0202] Using the first model, the probability that the fiber optic connector in the first region is a hot-melt connector is determined, thus obtaining a first probability; and / or, using the first model, the probability that the fiber optic connector in the first region is a cold connector is determined, thus obtaining a second probability;

[0203] The fiber optic connector type of the first ONU is determined based on the first probability and / or the second probability.

[0204] In one embodiment, the first processing unit 402 is further specifically used for:

[0205] Using the first model, a second region is determined in the image to be processed; the second region contains the fiber tray of the first ONU;

[0206] Using the first model, the probability that there is a tail fiber coiled in the fiber spool in the second region is determined to obtain a third probability; and / or, using the first model, the probability that there is no tail fiber coiled in the fiber spool in the second region is determined to obtain a fourth probability.

[0207] The fiber optic cable retention status of the first ONU is determined based on the third probability and / or the fourth probability.

[0208] In one embodiment, the first processing unit 402 is further configured to perform at least one of the following:

[0209] Using the first model, a third region is determined in the image to be processed; the third region contains the QR code of the first ONU; based on the third region, a QR code image is extracted from the image to be processed; the QR code image is parsed to obtain the first SN;

[0210] Using the first model, a fourth region is determined in the image to be processed; the fourth region contains the barcode of the first ONU; based on the fourth region, a barcode image is extracted from the image to be processed; the barcode image is parsed to obtain a second SN;

[0211] Using the first model, a fifth region is determined in the image to be processed; the fifth region contains the nameplate of the first ONU; based on the fifth region, a nameplate image is extracted from the image to be processed; text recognition is performed on the nameplate image to obtain a third SN; wherein,

[0212] If at least one of the identified SNs is the same as the target SN, the identified SN is determined to be consistent with the target SN; or, if at least one of the identified SNs is different from the target SN, the identified SN is determined to be inconsistent with the target SN.

[0213] In one embodiment, the first processing unit 402 is further configured to:

[0214] The length of the third region is increased by a first distance, and the width of the third region is increased by a second distance to obtain a sixth region; the sixth region is cropped from the image to be processed to obtain the QR code image;

[0215] Based on the aspect ratio of the sixth region, the QR code image is scaled within the first pixel range, and the scaled QR code image is then parsed.

[0216] In one embodiment, the first processing unit 402 is further configured to:

[0217] The length of the fourth region is increased by a third distance, and the width of the fourth region is increased by a fourth distance to obtain the seventh region; the seventh region is cropped from the image to be processed to obtain the barcode image;

[0218] The barcode image is subjected to denoising and edge sharpening processing, and the denoised and edge-sharpened barcode image is then parsed.

[0219] In one embodiment, the second processing unit 403 is specifically used for:

[0220] Given that the fiber optic connector type is a fusion splice, the pigtail coiling status is pigtail coiling, and the identified SN matches the target SN, determine the installation specifications of the first ONU.

[0221] The installation of the first ONU is determined to be non-standard if at least one of the following conditions is met:

[0222] The determined fiber optic connector type is a cold connector;

[0223] The confirmed situation regarding the retention of pigtail fibers is that there are no pigtail fibers retained.

[0224] The identified SN does not match the target SN.

[0225] In practical applications, the acquisition unit 401 can be implemented by the processor in the installation inspection device in conjunction with the communication interface; the first processing unit 402 and the second processing unit 403 can be implemented by the processor in the installation inspection device.

[0226] It should be noted that the installation inspection device provided in the above embodiments, when inspecting whether the ONU installation is standardized, is only illustrated by the division of the above-described program modules. In actual applications, the above processing can be assigned to different program modules as needed, that is, the internal structure of the device can be divided into different program modules to complete all or part of the processing described above. Furthermore, the installation inspection device and installation inspection method embodiments provided in the above embodiments belong to the same concept, and their specific implementation process is detailed in the method embodiments, which will not be repeated here.

[0227] Based on the hardware implementation of the above program modules, and in order to implement the method of the embodiments of this application, the embodiments of this application also provide an electronic device, such as... Figure 5 As shown, the electronic device 500 includes:

[0228] The communication interface 501 enables information exchange with other electronic devices;

[0229] The processor 502 is connected to the communication interface 501 to enable information interaction with other electronic devices and to execute the methods provided by one or more of the above-mentioned technical solutions when running computer programs;

[0230] The memory 503 stores computer programs that can run on the processor 502.

[0231] Specifically, the processor 502 is used for:

[0232] Acquire an image to be processed; the image to be processed should at least reflect the installation status of the first ONU;

[0233] Based on the image to be processed, the fiber optic connector type of the first ONU is determined using the first model, the pigtail coiling status of the first ONU is determined, and the serial number (SN) of the first ONU is identified.

[0234] Using the determined fiber optic connector type, the determined pigtail coiling status, and the identified serial number (SN), it is determined whether the installation of the first ONU is up to standard.

[0235] In one embodiment, the processor 502 is further configured to:

[0236] Using the first model, a first region is determined in the image to be processed; the first region contains the fiber optic connector of the first ONU.

[0237] Using the first model, the probability that the fiber optic connector in the first region is a hot-melt connector is determined, thus obtaining a first probability; and / or, using the first model, the probability that the fiber optic connector in the first region is a cold connector is determined, thus obtaining a second probability;

[0238] The fiber optic connector type of the first ONU is determined based on the first probability and / or the second probability.

[0239] In one embodiment, the processor 502 is further configured to:

[0240] Using the first model, a second region is determined in the image to be processed; the second region contains the fiber tray of the first ONU;

[0241] Using the first model, the probability that there is a tail fiber coiled in the fiber spool in the second region is determined to obtain a third probability; and / or, using the first model, the probability that there is no tail fiber coiled in the fiber spool in the second region is determined to obtain a fourth probability.

[0242] The fiber optic cable retention status of the first ONU is determined based on the third probability and / or the fourth probability.

[0243] In one embodiment, the processor 502 is further configured to perform at least one of the following:

[0244] Using the first model, a third region is determined in the image to be processed; the third region contains the QR code of the first ONU; based on the third region, a QR code image is extracted from the image to be processed; the QR code image is parsed to obtain the first SN;

[0245] Using the first model, a fourth region is determined in the image to be processed; the fourth region contains the barcode of the first ONU; based on the fourth region, a barcode image is extracted from the image to be processed; the barcode image is parsed to obtain a second SN;

[0246] Using the first model, a fifth region is determined in the image to be processed; the fifth region contains the nameplate of the first ONU; based on the fifth region, a nameplate image is extracted from the image to be processed; text recognition is performed on the nameplate image to obtain a third SN; wherein,

[0247] If at least one of the identified SNs is the same as the target SN, the identified SN is determined to be consistent with the target SN; or, if at least one of the identified SNs is different from the target SN, the identified SN is determined to be inconsistent with the target SN.

[0248] In one embodiment, the processor 502 is further configured to:

[0249] The length of the third region is increased by a first distance, and the width of the third region is increased by a second distance to obtain a sixth region; the sixth region is cropped from the image to be processed to obtain the QR code image;

[0250] Based on the aspect ratio of the sixth region, the QR code image is scaled within the first pixel range, and the scaled QR code image is then parsed.

[0251] In one embodiment, the processor 502 is further configured to:

[0252] The length of the fourth region is increased by a third distance, and the width of the fourth region is increased by a fourth distance to obtain the seventh region; the seventh region is cropped from the image to be processed to obtain the barcode image;

[0253] The barcode image is subjected to denoising and edge sharpening processing, and the denoised and edge-sharpened barcode image is then parsed.

[0254] In one embodiment, the processor 502 is further configured to:

[0255] Given that the fiber optic connector type is a fusion splice, the pigtail coiling status is pigtail coiling, and the identified SN matches the target SN, determine the installation specifications of the first ONU.

[0256] The installation of the first ONU is determined to be non-standard if at least one of the following conditions is met:

[0257] The determined fiber optic connector type is a cold connector;

[0258] The confirmed situation regarding the retention of pigtail fibers is that there are no pigtail fibers retained.

[0259] The identified SN does not match the target SN.

[0260] It should be noted that the specific process by which the processor 502 performs the above operations is detailed in the method embodiment, and will not be repeated here.

[0261] Of course, in practical applications, the various components in electronic device 500 are coupled together through bus system 504. It can be understood that bus system 504 is used to realize the connection and communication between these components. In addition to a data bus, bus system 504 also includes a power bus, a control bus, and a status signal bus. However, for the sake of clarity, in... Figure 5 The general designated all buses as Bus System 504.

[0262] The memory 503 in this embodiment is used to store various types of data to support the operation of the electronic device 500. Examples of such data include any computer program used to operate on the electronic device 500.

[0263] The methods disclosed in the embodiments of this application can be applied to or implemented by processor 502. Processor 502 may be an integrated circuit chip with signal processing capabilities. During implementation, each step of the above method can be completed by integrated logic circuits in the hardware of processor 502 or by instructions in software form. Processor 502 may be a general-purpose processor, a digital signal processor (DSP), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. Processor 502 can implement or execute the methods, steps, and logic block diagrams disclosed in the embodiments of this application. A general-purpose processor may be a microprocessor or any conventional processor. The steps of the methods disclosed in the embodiments of this application can be directly manifested as execution by a hardware decoding processor, or execution by a combination of hardware and software modules in the decoding processor. The software modules may be located in a storage medium, specifically memory 503. Processor 502 reads information from memory 503 and, in conjunction with its hardware, completes the steps of the aforementioned method.

[0264] In an exemplary embodiment, the electronic device 500 may be implemented by one or more application-specific integrated circuits (ASICs), DSPs, programmable logic devices (PLDs), complex programmable logic devices (CPLDs), field-programmable gate arrays (FPGAs), general-purpose processors, controllers, microcontrollers (MCUs), microprocessors, or other electronic components to perform the aforementioned method.

[0265] It is understood that the memory 503 in this embodiment can be volatile memory or non-volatile memory, or both. Non-volatile memory can be read-only memory (ROM), programmable read-only memory (PROM), erasable programmable read-only memory (EPROM), electrically erasable programmable read-only memory (EEPROM), ferromagnetic random access memory (FRAM), flash memory, magnetic surface memory, optical disc, or compact disc read-only memory (CD-ROM); magnetic surface memory can be disk storage or magnetic tape storage. Volatile memory can be random access memory (RAM), which is used as an external cache. By way of example, but not limitation, many forms of RAM are available, such as Static Random Access Memory (SRAM), Synchronous Static Random Access Memory (SSRAM), Dynamic Random Access Memory (DRAM), Synchronous Dynamic Random Access Memory (SDRAM), Double Data Rate Synchronous Dynamic Random Access Memory (DDRSDRAM), Enhanced Synchronous Dynamic Random Access Memory (ESDRAM), SyncLink Dynamic Random Access Memory (SLDRAM), and Direct Rambus Random Access Memory (DRRAM).The memories described in the embodiments of this application are intended to include, but are not limited to, these and any other suitable types of memories.

[0266] In an exemplary embodiment, this application also provides a storage medium, namely a computer storage medium, specifically a computer-readable storage medium, such as a memory 503 storing a computer program, which can be executed by the processor 502 of the electronic device 500 to complete the steps described in the aforementioned method. The computer-readable storage medium may be a memory such as FRAM, ROM, PROM, EPROM, EEPROM, Flash Memory, magnetic surface memory, optical disc, or CD-ROM.

[0267] It should be noted that terms such as "first" and "second" are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence.

[0268] Furthermore, the technical solutions described in the embodiments of this application can be combined arbitrarily without conflict.

[0269] The above description is merely a preferred embodiment of this application and is not intended to limit the scope of protection of this application.

Claims

1. An installation inspection method, characterized in that, include: Obtain the image to be processed; The image to be processed can at least reflect the installation status of the first optical network unit (ONU); Based on the image to be processed, the fiber optic connector type of the first ONU is determined using the first model, the pigtail coiling status of the first ONU is determined, and the product serial number (SN) of the first ONU is identified. Using the determined fiber optic connector type, the determined pigtail coiling status, and the identified serial number (SN), it is determined whether the installation of the first ONU is up to standard.

2. The method according to claim 1, characterized in that, The step of determining the fiber optic connector type of the first ONU based on the image to be processed and using the first model includes: Using the first model, a first region is determined in the image to be processed; the first region contains the fiber optic connector of the first ONU. Using the first model, the probability that the fiber optic connector in the first region is a hot-melt connector is determined, thus obtaining a first probability; and / or, using the first model, the probability that the fiber optic connector in the first region is a cold connector is determined, thus obtaining a second probability; The fiber optic connector type of the first ONU is determined based on the first probability and / or the second probability.

3. The method according to claim 1, characterized in that, The step of determining the fiber optic coiling status of the first ONU based on the image to be processed, using the first model, includes: Using the first model, a second region is determined in the image to be processed; the second region contains the fiber tray of the first ONU; Using the first model, the probability that there is a tail fiber coiled in the fiber spool in the second region is determined to obtain a third probability; and / or, using the first model, the probability that there is no tail fiber coiled in the fiber spool in the second region is determined to obtain a fourth probability. The fiber optic cable retention status of the first ONU is determined based on the third probability and / or the fourth probability.

4. The method according to claim 1, characterized in that, The step of identifying the SN of the first ONU based on the image to be processed using the first model includes at least one of the following: Using the first model, a third region is determined in the image to be processed; the third region contains the QR code of the first ONU; based on the third region, a QR code image is extracted from the image to be processed; the QR code image is parsed to obtain the first SN; Using the first model, a fourth region is determined in the image to be processed; the fourth region contains the barcode of the first ONU; based on the fourth region, a barcode image is extracted from the image to be processed. The barcode image is parsed to obtain the second SN; Using the first model, a fifth region is determined in the image to be processed; the fifth region contains the nameplate of the first ONU; based on the fifth region, a nameplate image is extracted from the image to be processed; text recognition is performed on the nameplate image to obtain a third SN; wherein, If at least one of the identified SNs is the same as the target SN, the identified SN is determined to be consistent with the target SN; or, if at least one of the identified SNs is different from the target SN, the identified SN is determined to be inconsistent with the target SN.

5. The method according to claim 4, characterized in that, The step of extracting a QR code image from the image to be processed based on the third region includes: The length of the third region is increased by a first distance, and the width of the third region is increased by a second distance to obtain a sixth region; the sixth region is cropped from the image to be processed to obtain the QR code image; The parsing of the QR code image includes: Based on the aspect ratio of the sixth region, the QR code image is scaled within the first pixel range, and the scaled QR code image is then parsed.

6. The method according to claim 4, characterized in that, The step of extracting a barcode image from the image to be processed based on the fourth region includes: The length of the fourth region is increased by a third distance, and the width of the fourth region is increased by a fourth distance to obtain the seventh region; the seventh region is cropped from the image to be processed to obtain the barcode image; The step of parsing the barcode image includes: The barcode image is subjected to denoising and edge sharpening processing, and the denoised and edge-sharpened barcode image is then parsed.

7. The method according to any one of claims 1 to 6, characterized in that, The process of determining whether the installation of the first ONU is up to standard by using the determined fiber optic connector type, the determined pigtail coiling status, and the identified serial number (SN) includes: Given that the fiber optic connector type is a fusion splice, the pigtail coiling status is pigtail coiling, and the identified SN matches the target SN, determine the installation specifications of the first ONU. The installation of the first ONU is determined to be non-standard if at least one of the following conditions is met: The determined fiber optic connector type is a cold connector; The confirmed situation regarding the retention of pigtail fibers is that there are no pigtail fibers retained. The identified SN does not match the target SN.

8. An installation inspection device, characterized in that, include: The acquisition unit is used to acquire the image to be processed; The image to be processed can at least reflect the installation status of the first ONU; The first processing unit is used to determine the fiber optic connector type of the first ONU, the pigtail coiling status of the first ONU, and the serial number (SN) of the first ONU based on the image to be processed and using the first model. The second processing unit is used to determine whether the installation of the first ONU is up to standard by using the determined fiber optic connector type, the determined pigtail coiling status, and the identified serial number (SN).

9. An electronic device, characterized in that, include: The processor and the memory used to store computer programs that can run on the processor. When the processor is used to run the computer program, it performs the steps of the method according to any one of claims 1 to 7.

10. A storage medium having a computer program stored thereon, characterized in that, When the computer program is executed by a processor, it implements the steps of the method according to any one of claims 1 to 7.

Images