CN111225299A

CN111225299A - ONU (optical network Unit) fault identification and repair method and device

Info

Publication number: CN111225299A
Application number: CN201811425957.5A
Authority: CN
Inventors: 汤斯鹏; 曹海山; 张培炜; 池鸿源; 赵佳琳; 陈楠; 张炳华; 陈梓珩
Original assignee: China Mobile Communications Group Co Ltd; China Mobile Group Guangdong Co Ltd
Current assignee: China Mobile Communications Group Co Ltd; China Mobile Group Guangdong Co Ltd
Priority date: 2018-11-27
Filing date: 2018-11-27
Publication date: 2020-06-02

Abstract

The embodiment of the invention provides an ONU fault identification and repair method and device, wherein the fault identification method comprises the following steps: acquiring an image to be identified; inputting an image to be identified into a target detection model, and acquiring an ONU image output by the target detection model; inputting the ONU image into a model classification model, and acquiring the ONU model output by the model classification model; and inputting the ONU image into a fault identification model corresponding to the ONU model, and acquiring the fault type output by the fault identification model. According to the method and the device provided by the embodiment of the invention, the user can obtain the fault type only by providing the picture of the fault ONU equipment without additionally mastering the use common knowledge of the related equipment, so that the fault identification threshold is reduced, the problems of inaccurate fault judgment and low efficiency caused by unsmooth communication during fault identification through text or voice interaction are solved, the efficient and accurate fault identification is realized, and the user experience is optimized.

Description

ONU (optical network Unit) fault identification and repair method and device

Technical Field

The embodiment of the invention relates to the technical field of machine learning, in particular to an ONU (optical network unit) fault identification and repair method and device.

Background

An ONU (Optical Network Unit ) is commonly called "Optical modem" and serves as a user-side device to provide various broadband services to connected users.

According to the current service statistics, the fault ratio of ONU in basic broadband and television services related to the home broadband service exceeds 50%, so that it is necessary to guide users to solve ONU faults by self-help. The existing methods for self-service identification and repair of ONU faults mainly comprise two methods: one is that label and two-dimensional code are pasted on ONU, guide user to scan two-dimensional code, and combine label description to repair itself according to image and text information provided by webpage; the other method is to issue common fault processing guide and function publicity through an electronic channel, so that a user can specify fault types and solution modes through an interactive mode, such as voice communication with customer service personnel.

According to the self-service identification and repair method, on one hand, a user needs to have certain ONU equipment use common knowledge, can distinguish a network cable interface, an optical fiber signal interface, a power supply interface and the like, and can distinguish the ONU status light condition, so that the user can perform self-service troubleshooting or fault reporting through image-text description, webpage interaction or voice interaction when the ONU fails; on the other hand, the voice customer service is used for assisting troubleshooting, customer service personnel are required to have fault discrimination capability on a variety of ONU devices, however, the types of the current mainstream ONU devices can reach dozens of types, and in addition, users cannot clearly express the ONU devices in the interaction process, so that the customer service personnel sometimes have failure understanding, and the fault repairing efficiency is influenced. Due to the various limitations, the self-service fault identification and repair rate for the ONU is difficult to improve, and when the self-service identification and repair method cannot solve the ONU fault, installation and maintenance personnel are often required to perform door-to-door service, so that the timeliness is insufficient and the cost is high.

Disclosure of Invention

The embodiment of the invention provides an ONU (optical network unit) fault identification and repair method and device, which are used for solving the problems of high threshold, low resolution and long time consumption of the existing ONU fault identification and repair method.

In a first aspect, an embodiment of the present invention provides an ONU fault identification method, including:

acquiring an image to be identified;

inputting the image to be recognized into a target detection model, and acquiring an ONU image output by the target detection model; the target detection model is obtained by training a sample image to be identified and a sample ONU image in the sample image to be identified;

inputting the ONU image into a model classification model, and acquiring the ONU model output by the model classification model; the model classification model is obtained by training a sample ONU image and a sample ONU model corresponding to the sample ONU image;

inputting the ONU image into a fault identification model corresponding to the ONU model, and acquiring the fault type output by the fault identification model; the fault identification model is obtained by training a sample ONU image and a sample fault type corresponding to the sample ONU image.

In a second aspect, an embodiment of the present invention provides an ONU fault repairing method, including:

acquiring the ONU model and the fault type of an image to be identified; the ONU model and the fault type are obtained based on the fault identification method provided by the first aspect;

calling a 3D model corresponding to the ONU model based on AR technology, and displaying a repair prompt corresponding to the fault type, so that a user can realize the fault repair of the ONU according to the repair prompt.

In a third aspect, an embodiment of the present invention provides an ONU fault identifying apparatus, including:

the device comprises an acquisition unit, a recognition unit and a processing unit, wherein the acquisition unit is used for acquiring an image to be recognized;

the target detection unit is used for inputting the image to be identified into a target detection model and acquiring an ONU image output by the target detection model; the target detection model is obtained by training a sample image to be identified and a sample ONU image in the sample image to be identified;

the model classification unit is used for inputting the ONU image into a model classification model and acquiring the ONU model output by the model classification model; the model classification model is obtained by training a sample ONU image and a sample ONU model corresponding to the sample ONU image;

the fault identification unit is used for inputting the ONU image into a fault identification model corresponding to the ONU model and acquiring the fault type output by the fault identification model; the fault identification model is obtained by training a sample ONU image and a sample fault type corresponding to the sample ONU image.

In a fourth aspect, an embodiment of the present invention provides an ONU fault repairing apparatus, including:

the identification unit is used for acquiring the ONU model and the fault type of the image to be identified; the ONU model and the fault type are obtained based on the fault identification method provided by the first aspect;

and the repairing unit is used for calling the 3D model corresponding to the ONU model based on the AR technology and displaying a repairing prompt corresponding to the fault type so that a user can realize the fault repairing of the ONU according to the repairing prompt.

In a fifth aspect, an embodiment of the present invention provides an electronic device, including a processor, a communication interface, a memory, and a bus, where the processor and the communication interface, the memory complete communication with each other through the bus, and the processor may call logic instructions in the memory to perform the steps of the method as provided in the first aspect or the second aspect.

In a sixth aspect, embodiments of the present invention provide a non-transitory computer readable storage medium, on which a computer program is stored, which when executed by a processor, implements the steps of the method as provided in the first or second aspect.

According to the ONU fault identification and repair method and device provided by the embodiment of the invention, the fault type of the faulty ONU equipment in the image to be identified is obtained based on the target detection model, the model classification model and the fault identification model, a user can obtain the fault type only by providing the picture of the faulty ONU equipment without additionally mastering the use knowledge of the related equipment, the fault identification threshold is reduced, the problems of inaccurate fault judgment and low efficiency caused by unsmooth communication when fault identification is carried out through text or voice interaction are avoided, high-efficiency and accurate fault identification is realized, and the user experience is optimized. In addition, the artificial intelligence and the AR technology are combined, the 3D model and the repair prompt of the ONU equipment are directly displayed at the intelligent terminal, the user is guided to carry out ONU fault repair, the threshold of fault repair is reduced, and the cost input of home repair of maintenance personnel is saved.

Drawings

In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings used in the description of the embodiments or the prior art will be briefly described below, and it is obvious that the drawings in the following description are some embodiments of the present invention, and those skilled in the art can also obtain other drawings according to the drawings without creative efforts.

Fig. 1 is a schematic flowchart of an ONU fault identification method according to an embodiment of the present invention;

FIG. 2 is a schematic structural diagram of a target detection model according to an embodiment of the present invention;

fig. 3 is a schematic flowchart of an ONU fault repairing method according to an embodiment of the present invention;

fig. 4 is a schematic structural diagram of an ONU fault identifying device according to an embodiment of the present invention;

fig. 5 is a schematic structural diagram of an ONU fault repairing apparatus according to an embodiment of the present invention;

fig. 6 is a schematic structural diagram of an electronic device according to an embodiment of the present invention.

Detailed Description

In order to make the objects, technical solutions and advantages of the embodiments of the present invention clearer, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the drawings in the embodiments of the present invention, and it is obvious that the described embodiments are some, but not all, embodiments of the present invention. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.

The current self-service identification and repair method aiming at the ONU faults needs a user to know ONU equipment, so that the threshold of the self-service fault removal door method is high, the completion rate of the user in using pictures and texts or an H5 page to remove faults by self-service is low, and the problem solution rate of the self-service fault removal is low. In addition, due to the lack of a visualization method for a user side fault scene, even if the user is unable to express the fault by self-help troubleshooting through online customer service personnel, the situation that the customer service personnel cannot understand the fault in place occurs all the time, so that the self-help troubleshooting problem is long in solving time, low in solving rate and the like. In view of the above problems, embodiments of the present invention provide an ONU fault identification method to implement efficient and accurate ONU fault identification, thereby providing conditions for user self-help troubleshooting. Fig. 1 is a schematic flow chart of an ONU fault identification method according to an embodiment of the present invention, and as shown in fig. 1, an execution main body of the ONU fault identification method may be an intelligent terminal having a shooting function, or may be a processing unit such as a server having an information processing function, which is not limited in this embodiment of the present invention. The method comprises the following steps:

and 110, acquiring an image to be identified.

Here, the image to be recognized is an image including a faulty ONU device that needs to be recognized, the image to be recognized may be obtained by shooting the faulty ONU device through an intelligent terminal, where the intelligent terminal may be a smart phone, a smart wristwatch, a tablet computer, a notebook computer, or the like, and this is not particularly limited in the embodiment of the present invention.

120, inputting the image to be identified into a target detection model, and acquiring an ONU image output by the target detection model; the target detection model is obtained by training a sample image to be identified and a sample ONU image in the sample image to be identified.

Specifically, after an image to be recognized is obtained, the image to be recognized is input into a target detection model, the target detection model performs feature extraction on the image to be recognized, the position of a fault ONU device in the image to be recognized is determined, the fault ONU device in the image to be recognized is intercepted, and an ONU image is obtained and output. Here, the ONU image is an image of a faulty ONU device intercepted from the image to be recognized, which is output by the object detection model.

In addition, before step 120 is executed, the target detection model may be obtained by training in advance, and specifically, the target detection model may be obtained by training in the following manner: firstly, collecting a large number of images to be identified of samples and ONU images of the samples; the image to be identified of the sample is an image containing ONU equipment, and the image of the ONU equipment of the sample is cut from the image to be identified of the sample. And training the initial model based on the image to be identified of the sample and the sample ONU image so as to obtain a target detection model. The initial model may be a single neural network model or a combination of a plurality of neural network models, and the embodiment of the present invention does not specifically limit the type and structure of the initial model.

Inputting the ONU image into the model classification model to obtain the ONU model output by the model classification model 130; the model classification model is obtained by training a sample ONU model corresponding to the sample ONU image and the sample ONU image.

Specifically, after the ONU image is acquired, the ONU image is input into a model classification model, and the model classification model outputs a model of a faulty ONU device in the ONU image, that is, an ONU model. Before step 130 is executed, the model classification model may also be obtained by training in advance, and specifically, the model classification model may be obtained by training in the following manner: firstly, collecting a large number of sample ONU images and sample ONU models corresponding to the sample ONU images; the sample ONU image is an image of the ONU device, the sample ONU model of the sample ONU image is predetermined, and the sample ONU model is the model of the ONU device in the sample ONU image. And training the initial model based on the sample ONU image and the sample ONU so as to obtain a model classification model. The initial model may be a single neural network model or a combination of a plurality of neural network models, and the embodiment of the present invention does not specifically limit the type and structure of the initial model.

140, inputting the ONU image into a fault identification model corresponding to the ONU model, and acquiring the fault type output by the fault identification model; the fault identification model is obtained by training a sample ONU image and a sample fault type corresponding to the sample ONU image.

Specifically, there is a correspondence between the ONU models and the fault identification model, and one ONU model may correspond to one fault identification model, or multiple ONU models may correspond to one fault identification model. After the ONU model corresponding to the ONU image is obtained, the ONU image is input into a fault identification model corresponding to the ONU model, the fault identification model determines the state of the fault ONU equipment based on the state of an indicator lamp of the fault ONU equipment in the ONU image, the fault type of the fault ONU equipment is further obtained, and the fault type is output.

Before step 140 is executed, the fault identification models corresponding to the respective ONU models may also be obtained by training in advance, and specifically may be obtained by training in the following manner: taking a fault identification model corresponding to any ONU model as an example, firstly, collecting a large number of sample ONU images and fault types corresponding to the sample ONU images; the sample ONU image is an image of the ONU equipment of the ONU model, the sample fault type of the sample ONU image is predetermined, and the sample fault type is determined according to the indicator lamp state of the ONU equipment of the ONU model in the sample ONU image. And training the initial model based on the sample ONU image and the sample fault type corresponding to the ONU model, thereby obtaining a fault identification model. The initial model may be a single neural network model or a combination of a plurality of neural network models, and the embodiment of the present invention does not specifically limit the type and structure of the initial model.

According to the method provided by the embodiment of the invention, the fault type of the fault ONU equipment in the image to be recognized is obtained based on the target detection model, the model classification model and the fault recognition model, a user can obtain the fault type only by providing the picture of the fault ONU equipment without additionally mastering the use knowledge of the related equipment, the threshold of fault recognition is reduced, the problems of inaccurate fault judgment and low efficiency caused by unsmooth communication when fault recognition is carried out through text or voice interaction are avoided, efficient and accurate fault recognition is realized, and the user experience is optimized.

Based on any of the above embodiments, fig. 2 is a schematic structural diagram of a target detection model provided in an embodiment of the present invention, and referring to fig. 2, step 120 specifically includes:

and 121, inputting the image to be recognized into a first detection model in the target detection model, and acquiring the candidate ONU image output by the first detection model and the confidence of the candidate ONU image.

Specifically, the target detection model includes a first detection model and a second detection model, and the first detection model and the second detection model may be a single neural network model or a combination of multiple neural network models. The first detection model is used for preliminarily detecting whether the image to be identified contains the defective ONU equipment or not, and outputting the candidate ONU image intercepted from the image to be identified and the confidence coefficient of the candidate ONU image. Here, the candidate ONU image is an image that is cut out by the first detection model from the image to be identified and may include a faulty ONU device, and the confidence of the candidate ONU image refers to a probability that the candidate ONU image includes the faulty ONU image.

122, if the confidence coefficient of the candidate ONU image is greater than or equal to the preset confidence coefficient, taking the candidate ONU image as the ONU image; otherwise, inputting the candidate ONU image into a second detection model in the target detection model, and acquiring a detection result output by the second detection model; and if the detection result shows that the ONU exists, taking the candidate ONU image as the ONU image.

Specifically, after acquiring the candidate ONU image and the confidence of the candidate ONU image, the confidence of the candidate ONU image is compared with a preset confidence, where the preset confidence is a preset confidence threshold. And if the confidence coefficient of the candidate ONU image is greater than or equal to the preset confidence coefficient, confirming that the candidate ONU image contains the faulty ONU equipment, and directly outputting the candidate ONU image as the ONU image from the target detection model. If the confidence coefficient of the candidate ONU image is smaller than the preset confidence coefficient, whether the candidate ONU image contains the fault ONU equipment or not can not be directly judged through the output result of the first detection model, and further judgment needs to be carried out through the second detection model. And inputting the candidate ONU image into a second detection model, wherein the second detection model is used for detecting whether the candidate ONU image contains the fault ONU equipment again and outputting a detection result. Here, the detection result indicates that there is an ONU, that is, the candidate ONU image includes a faulty ONU device, or that there is no ONU, that is, there is no faulty ONU device in the candidate ONU image. And if the detection result shows that the ONU exists, outputting the candidate ONU image as the ONU image from the target detection model. And if the detection result is that the ONU does not exist, the target detection model outputs the detection result.

The first detection model may also be trained in advance before step 121 is performed, and the second detection model may also be trained in advance before step 122 is performed. In addition, the trained first detection model and the trained second detection model may also be obtained by training the whole target detection model, which is not specifically limited in the embodiment of the present invention.

According to the method provided by the embodiment of the invention, in the target detection model, the ONU image of the image to be identified is obtained through the first detection model and the second detection model, so that the accurate positioning of the ONU image is realized, and conditions are provided for candidate type identification and fault detection based on the ONU image.

Based on any of the above embodiments, step 122 further includes: and if the detection result is that the ONU does not exist, sending a prompt for acquiring the image to be identified again based on the detection result output by the target detection model.

Specifically, in the target detection model, if the detection result output by the second detection model is that no ONU exists, the detection result is output from the target detection model. And sending a prompt for re-acquiring the image to be identified after the detection result is obtained, so that a user can re-shoot the image to be identified containing the ONU equipment with the fault according to the prompt, re-executing the acquisition of the image to be identified, inputting the image to be identified into the target detection model and the like.

For example, the execution subject of the method provided by the above embodiment is the intelligent terminal, and the user takes the image to be recognized through the intelligent terminal, and then the intelligent terminal inputs the image to be recognized into the target detection model, and obtains the detection result output by the target detection model, that is, there is no ONU. After the detection result is obtained, the intelligent terminal sends a prompt for reacquiring the image to be identified, the prompt is displayed on a screen of the intelligent terminal, and the user can shoot the image to be identified again through the intelligent terminal after observing the prompt.

For another example, the execution subject of the method provided by the above embodiment is the server, and the user takes the image to be recognized through the intelligent terminal, and then the intelligent terminal transmits the image to be recognized to the server. And after the server acquires the image to be recognized, inputting the image to be recognized into the target detection model, and acquiring a detection result output by the target detection model, namely that no ONU exists. After the detection result is obtained, the server sends a prompt of reacquiring the image to be recognized to the intelligent terminal, the intelligent terminal displays the prompt on the screen after receiving the prompt of reacquiring the image to be recognized, and after the user observes the prompt, the user can shoot the image to be recognized again through the intelligent terminal and transmits the image to be recognized to the server for processing through the intelligent terminal.

Based on any of the above embodiments, the first detection model is a Tensorflow object detection API, and the second detection model is an inclusion V4 neural network.

Specifically, the tensrflow is a symbolic mathematical system based on data flow programming, and is widely applied to programming of various machine learning algorithms. The Object Detection API (application programming Interface) is an Object Detection Interface issued by google, and a first Detection model can be obtained by training the Object Detection API.

The inclusion V4 Neural Network is a fourth generation version of the inclusion Network, which is an important milestone in the development history of Convolutional Neural Network (CNN) classifiers. The inclusion V4 neural network greatly accelerates the training speed and greatly improves the performance by combining the ResNet (Residual neural network) on the basis of the inclusion network. The second detection model can be obtained by training the IncepotionV 4 neural network.

Based on any of the above embodiments, step 130 specifically includes: 131, inputting the ONU image into a model rough classification model in the model classification model, and acquiring a preliminary classification result output by the model rough classification model; and 132, inputting the ONU image into a model subdivision model corresponding to the preliminary classification result in the model classification model, and acquiring the ONU model output by the model subdivision model.

Specifically, the model classification model comprises a model rough classification model and a plurality of model subdivision models, the model rough classification model and the model subdivision models can be a single neural network model or a combination of a plurality of neural network models, and the embodiment of the invention does not specifically limit the types and the structures of the model rough classification model and the model subdivision models.

The model rough classification model is used for carrying out preliminary model classification on the fault ONU equipment contained in the ONU image and outputting a preliminary classification result of the fault ONU equipment in the ONU image. Here, the preliminary classification results are used to represent the coarse classification of the defective ONU device models included in the ONU image, and each of the preliminary classification results corresponds to a plurality of subdivided ONU models. For example, the preliminary classification result is the ONU brand, and the segmentation result corresponding to the preliminary classification result is the specific model under the brand. For another example, the preliminary classification result is a type of ONU device with high appearance similarity, and a specific model of the ONU device with high appearance similarity is obtained by further subdivision.

The corresponding relationship between the preliminary classification result and the model subdivision model may be that one preliminary classification result corresponds to one model subdivision model, or that a plurality of preliminary classification results correspond to one model subdivision model, which is not specifically limited in the embodiment of the present invention. After a preliminary classification result corresponding to the ONU image is obtained, the ONU image is input into a model subdivision model corresponding to the preliminary classification result, and the model subdivision model outputs an ONU model corresponding to the faulty ONU equipment contained in the ONU image.

According to the method provided by the embodiment of the invention, the ONU models corresponding to the ONU images are accurately classified by the connection application of the model rough classification model and the model subdivision model.

Based on any one of the embodiments, the model rough classification model is an inclusion V4 neural network, and the model subdivision model is an AlexNet neural network.

Specifically, the AlexNet neural network is a model for introducing the convolutional neural network into the computer vision field for the first time and achieving breakthrough performance. The AlexNet neural network enables the model to be deeper and wider by stacking convolution layers, meanwhile, training is enabled to obtain results within an acceptable time range by means of the GPU, and development of the convolution neural network and even deep learning is promoted. Compared with an Incepton V4 neural network, the AlexNet neural network has a simpler structure and is more beneficial to reducing the operation overhead of the flow.

Based on any of the above embodiments, fig. 3 is a schematic flowchart of a method for repairing an ONU fault according to an embodiment of the present invention, and as shown in fig. 3, an execution subject of the method is an intelligent terminal, and the method includes:

and 310, acquiring the ONU model and the fault type of the image to be identified. The ONU model and the fault type are obtained based on the fault identification method provided in any of the above embodiments.

Here, the method for acquiring the ONU model and the fault type of the faulty ONU device in the image to be identified according to the image to be identified is as described in any of the above embodiments. It should be noted that the method for executing fault identification may be an intelligent terminal, or may be a processing unit such as a server and having an information processing function, which is not specifically limited in this embodiment of the present invention. When the execution main body of the fault identification method is an intelligent terminal, the intelligent terminal directly executes the fault identification method and obtains the ONU model and the fault type of the image to be identified; when the executing main body of the fault identification method is the processing unit, the processing unit executes the fault identification method, acquires the ONU model and the fault type of the image to be identified, and sends the ONU model and the fault type to the intelligent terminal.

And 320, calling a 3D model corresponding to the ONU model based on the AR technology, and displaying a repair prompt corresponding to the fault type, so that a user can realize the fault repair of the ONU according to the repair prompt.

Specifically, the AR (Augmented Reality) technology is a technology for calculating the position and angle of a camera image in real time and adding corresponding images, videos and 3D models, and the purpose of the technology is to overlap a virtual world on a screen in the real world and perform interaction. After the PNU model and the fault type of the fault ONU equipment are obtained, the corresponding ONU equipment 3D model is selected based on the ONU model, the corresponding repair prompt is matched based on the fault type, and the 3D model is called through the AR technology, so that the intelligent terminal can be used for repeating the 3D model on the position of the fault ONU equipment in the picture obtained by shooting through the shooting unit. In addition, a repair prompt is synchronously displayed on the intelligent terminal based on the 3D model, so that a user is instructed to repair the fault ONU equipment according to the 3D model and the repair prompt displayed by the intelligent terminal.

According to the method provided by the embodiment of the invention, the 3D model and the repair prompt of the ONU equipment are directly displayed on the intelligent terminal through the combination of artificial intelligence and AR technology, so that the user is guided to carry out ONU fault repair, the threshold of fault repair is reduced, the user experience is optimized, and the cost input of home repair of maintenance personnel is saved.

Based on any of the above embodiments, in step 320, invoking a 3D model corresponding to the ONU model based on the AR technology specifically includes: calling a 3D model corresponding to the ONU model based on the AR technology, and setting the 3D model based on the fault type, so that the 3D model can restore the state of the ONU in the image to be identified.

Specifically, the fault type is used for indicating the state of the faulty ONU device, and the state of the indicator in the 3D model can be set based on the fault type, so that the state of the indicator in the 3D model can be kept consistent with the state of the indicator of the faulty ONU device in the state corresponding to the fault type, and the state of the faulty ONU device shot in the image to be identified is effectively restored. After the 3D model is set based on the fault type, a user can obtain correct repair measures more intuitively through the 3D model and the repair prompt displayed by the intelligent terminal, and then the fault ONU equipment can be repaired.

Based on any of the above embodiments, the ONU fault repairing method includes the following steps:

firstly, a user opens a camera through a smart phone to shoot a picture of a fault ONU device, namely an image to be identified.

Secondly, the smart phone transmits the image to be recognized to an image recognition server. And after receiving the image to be recognized, the image recognition server inputs the image to be recognized into the target detection model. Further, the image to be recognized is input into a first detection model in the target detection model, and the candidate ONU image and the confidence coefficient of the candidate ONU image output by the first detection model are obtained. And if the confidence coefficient of the candidate ONU image is smaller than the preset confidence coefficient, inputting the candidate ONU image into a second detection model in the target detection model to obtain a detection result output by the second detection model. And outputting the candidate ONU image as the ONU image from the target detection model because the detection result is that the ONU exists. Here, the ONU image is a minimum rectangular region including a faulty ONU device, which is cut from the image to be identified, and the cut ONU image can greatly reduce interference caused by a background.

Subsequently, the image recognition server inputs the ONU image into the model classification model. And further, inputting the ONU image into a model rough classification model of the model classification model, and acquiring a preliminary classification result output by the model rough classification model. In order to improve the classification effect, ONU models with similar shapes are further distinguished, ONU images are input into a model subdivision model corresponding to the preliminary classification result, and the ONU models output by the model subdivision model are obtained.

And then, the image recognition server inputs the ONU image into a fault recognition model corresponding to the ONU signal model. The status of the ONU is mainly distinguished by indicator lights on the surface of the ONU device, including power, pon, los, and lan lights. The main indicator light combination status can be classified into 7 types according to statistics. The fault identification model classifies the combination state of the indicator lamps based on the state of the indicator lamps of the faulty ONU equipment in the ONU image, further determines the state of the faulty ONU equipment, obtains the fault type of the faulty ONU equipment, and outputs the fault type.

And then, the image recognition server returns the ONU model and the fault type of the image to be recognized obtained in the step to the smart phone. After the smart phone receives the ONU model and the fault type, the smart phone calls a 3D model corresponding to the ONU model, sets the 3D model based on the fault type, synchronously displays the current state of the fault ONU equipment on the smart phone in a WYSIWYG mode, and simultaneously displays a repair prompt corresponding to the fault type on the smart phone to guide a user to repair according to the prompt.

According to the method provided by the embodiment of the invention, the fault type of the fault ONU equipment in the image to be recognized is obtained based on the target detection model, the model classification model and the fault recognition model, a user can obtain the fault type only by providing the picture of the fault ONU equipment without additionally mastering the use knowledge of the related equipment, the threshold of fault recognition is reduced, the problems of inaccurate fault judgment and low efficiency caused by unsmooth communication when fault recognition is carried out through text or voice interaction are avoided, efficient and accurate fault recognition is realized, and the user experience is optimized. In addition, the artificial intelligence and the AR technology are combined, the 3D model and the repair prompt of the ONU equipment are directly displayed at the intelligent terminal, the user is guided to carry out ONU fault repair, the threshold of fault repair is reduced, and the cost input of home repair of maintenance personnel is saved.

Based on any of the above embodiments, fig. 4 is a schematic structural diagram of an ONU fault identifying device according to an embodiment of the present invention, and as shown in fig. 4, the ONU fault identifying device includes an obtaining unit 410, an object detecting unit 420, a model classifying unit 430, and a fault identifying unit 440;

the acquiring unit 410 is used for acquiring an image to be identified;

the target detection unit 420 is configured to input an image to be identified into a target detection model, and acquire an ONU image output by the target detection model; the target detection model is obtained by training a sample image to be identified and a sample ONU image in the sample image to be identified;

the model classification unit 430 is configured to input the ONU image into a model classification model, and obtain an ONU model output by the model classification model; the model classification model is obtained by training a sample ONU model corresponding to the sample ONU image and the sample ONU image;

the fault identification unit 440 is configured to input an ONU image into a fault identification model corresponding to the ONU model, and obtain a fault type output by the fault identification model; the fault identification model is obtained by training a sample ONU image and a sample fault type corresponding to the sample ONU image.

According to the device provided by the embodiment of the invention, the fault type of the fault ONU equipment in the image to be recognized is obtained based on the target detection model, the model classification model and the fault recognition model, a user can obtain the fault type only by providing a picture of the fault ONU equipment without additionally mastering the use knowledge of related equipment, the threshold of fault recognition is reduced, the problems of inaccurate fault judgment and low efficiency caused by unsmooth communication when fault recognition is carried out through text or voice interaction are avoided, efficient and accurate fault recognition is realized, and the user experience is optimized.

Based on any of the above embodiments, the target detection unit 420 is specifically configured to:

inputting an image to be recognized into a first detection model in a target detection model, and acquiring a candidate ONU image and confidence of the candidate ONU image output by the first detection model;

if the confidence coefficient of the candidate ONU image is greater than or equal to the preset confidence coefficient, the judgment subunit takes the candidate ONU image as the ONU image;

otherwise, inputting the candidate ONU image into a second detection model in the target detection model, and acquiring a detection result output by the second detection model;

and if the detection result shows that the ONU exists, taking the candidate ONU image as the ONU image.

Based on any of the above embodiments, the target detection unit 420 further includes:

and if the detection result is that the ONU does not exist, sending a prompt for acquiring the image to be identified again based on the detection result output by the target detection model.

Based on any of the above embodiments, the model classification unit 430 is specifically configured to:

inputting the ONU image into a model rough classification model in the model classification model, and acquiring a preliminary classification result output by the model rough classification model;

and inputting the ONU image into a model subdivision model corresponding to the preliminary classification result in the model classification model, and obtaining the ONU model output by the model subdivision model.

Based on any of the above embodiments, fig. 5 is a schematic structural diagram of an ONU fault repairing apparatus according to an embodiment of the present invention, and as shown in fig. 5, the ONU fault repairing apparatus includes an identification unit 510 and a repairing unit 520;

the identification unit 510 is configured to obtain an ONU model and a fault type of an image to be identified; the ONU model and the fault type are obtained based on the fault identification method provided in any of the above embodiments.

The repair unit 520 is configured to call a 3D model corresponding to the ONU model based on the AR technology, and display a repair prompt corresponding to the fault type, so that a user can implement fault repair of the ONU according to the repair prompt.

Based on any of the above embodiments, the repair unit 520 is specifically configured to: calling a 3D model corresponding to the ONU model based on the AR technology, and setting the 3D model based on the fault type, so that the 3D model can restore the state of the ONU in the image to be identified.

Fig. 6 is a schematic entity structure diagram of an electronic device according to an embodiment of the present invention, and as shown in fig. 6, the electronic device may include: a processor (processor)601, a communication Interface (Communications Interface)602, a memory (memory)603 and a communication bus 604, wherein the processor 601, the communication Interface 602 and the memory 603 complete communication with each other through the communication bus 604. The processor 601 may call a computer program stored on the memory 603 and executable on the processor 601 to execute the ONU fault identifying method provided by the above embodiments, for example, including: acquiring an image to be identified; inputting an image to be identified into a target detection model, and acquiring an ONU image output by the target detection model; the target detection model is obtained by training a sample image to be identified and a sample ONU image in the sample image to be identified; inputting the ONU image into a model classification model, and acquiring the ONU model output by the model classification model; the model classification model is obtained by training a sample ONU model corresponding to the sample ONU image and the sample ONU image; inputting an ONU image into a fault identification model corresponding to the ONU model, and acquiring the fault type output by the fault identification model; the fault identification model is obtained by training a sample ONU image and a sample fault type corresponding to the sample ONU image.

Furthermore, the processor 601 may also call a computer program stored on the memory 603 and executable on the processor 601 to execute the ONU failure repairing method provided by the foregoing embodiments, for example, including: acquiring the ONU model and the fault type of an image to be identified; the ONU model and the fault type are obtained based on a fault identification method; calling a 3D model corresponding to the ONU model based on the AR technology, and displaying a repair prompt corresponding to the fault type, so that a user can realize the fault repair of the ONU according to the repair prompt.

In addition, the logic instructions in the memory 603 may be implemented in the form of software functional units and stored in a computer readable storage medium when the logic instructions are sold or used as independent products. Based on such understanding, the technical solutions of the embodiments of the present invention may be essentially implemented or make a contribution to the prior art, or may be implemented in the form of a software product stored in a storage medium and including instructions for causing a computer device (which may be a personal computer, a server, or a network device) to execute all or part of the steps of the methods described in the embodiments of the present invention. And the aforementioned storage medium includes: a U-disk, a removable hard disk, a Read-Only Memory (ROM), a Random Access Memory (RAM), a magnetic disk or an optical disk, and other various media capable of storing program codes.

An embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program is implemented to perform the ONU fault identification method provided in each of the above embodiments when executed by a processor, and the ONU fault identification method includes: acquiring an image to be identified; inputting an image to be identified into a target detection model, and acquiring an ONU image output by the target detection model; the target detection model is obtained by training a sample image to be identified and a sample ONU image in the sample image to be identified; inputting the ONU image into a model classification model, and acquiring the ONU model output by the model classification model; the model classification model is obtained by training a sample ONU model corresponding to the sample ONU image and the sample ONU image; inputting an ONU image into a fault identification model corresponding to the ONU model, and acquiring the fault type output by the fault identification model; the fault identification model is obtained by training a sample ONU image and a sample fault type corresponding to the sample ONU image.

Furthermore, an embodiment of the present invention further provides a non-transitory computer-readable storage medium, on which a computer program is stored, where the computer program is implemented to perform the ONU fault repairing method provided in each of the above embodiments when executed by a processor, and for example, the ONU fault repairing method includes: acquiring the ONU model and the fault type of an image to be identified; the ONU model and the fault type are obtained based on a fault identification method; calling a 3D model corresponding to the ONU model based on the AR technology, and displaying a repair prompt corresponding to the fault type, so that a user can realize the fault repair of the ONU according to the repair prompt.

The above-described embodiments of the apparatus are merely illustrative, and the units described as separate parts may or may not be physically separate, and parts displayed as units may or may not be physical units, may be located in one place, or may be distributed on a plurality of network units. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of the present embodiment. One of ordinary skill in the art can understand and implement it without inventive effort.

Through the above description of the embodiments, those skilled in the art will clearly understand that each embodiment can be implemented by software plus a necessary general hardware platform, and certainly can also be implemented by hardware. With this understanding in mind, the above-described technical solutions may be embodied in the form of a software product, which can be stored in a computer-readable storage medium such as ROM/RAM, magnetic disk, optical disk, etc., and includes instructions for causing a computer device (which may be a personal computer, a server, or a network device, etc.) to execute the methods described in the embodiments or some parts of the embodiments.

Finally, it should be noted that: the above examples are only intended to illustrate the technical solution of the present invention, but not to limit it; although the present invention has been described in detail with reference to the foregoing embodiments, it will be understood by those of ordinary skill in the art that: the technical solutions described in the foregoing embodiments may still be modified, or some technical features may be equivalently replaced; and such modifications or substitutions do not depart from the spirit and scope of the corresponding technical solutions of the embodiments of the present invention.

Claims

1. An ONU fault identification method is characterized by comprising the following steps:

acquiring an image to be identified;

2. The method according to claim 1, wherein the inputting the image to be recognized into a target detection model and obtaining the ONU image output by the target detection model specifically comprises:

inputting the image to be identified into a first detection model in the target detection model, and acquiring a candidate ONU image output by the first detection model and a confidence coefficient of the candidate ONU image;

if the confidence coefficient of the candidate ONU image is greater than or equal to the preset confidence coefficient, taking the candidate ONU image as the ONU image;

and if the detection result is that the ONU exists, taking the candidate ONU image as the ONU image.

3. The method according to claim 2, wherein the inputting the candidate ONU image into a second detection model in the target detection models, and obtaining the detection result output by the second detection model, further comprises:

and if the detection result is that the ONU does not exist, sending a prompt for reacquiring the image to be identified based on the detection result output by the target detection model.

4. The fault identification method according to claim 2, wherein the first detection model is a Tensorflow object detection API and the second detection model is an inclusion V4 neural network.

5. The method according to claim 1, wherein the inputting the ONU image into a model classification model to obtain the ONU model output by the model classification model specifically comprises:

6. The fault identification method according to claim 5, wherein the model coarse classification model is an IncepotionV 4 neural network, and the model fine classification model is an AlexNet neural network.

7. An ONU fault repairing method is characterized by comprising the following steps:

acquiring the ONU model and the fault type of an image to be identified; the ONU model and the fault type are obtained based on the fault identification method of any one of claims 1 to 6;

8. The recovery method according to claim 7, wherein the calling the 3D model corresponding to the ONU model based on the AR technology specifically includes:

calling a 3D model corresponding to the ONU model based on AR technology, and setting the 3D model based on the fault type, so that the 3D model can restore the state of the ONU in the image to be identified.

9. An electronic device, comprising a processor, a communication interface, a memory and a bus, wherein the processor, the communication interface and the memory communicate with each other via the bus, and the processor can call logic instructions in the memory to execute the method according to any one of claims 1 to 8.

10. A non-transitory computer-readable storage medium, on which a computer program is stored, which, when being executed by a processor, implements the method according to any one of claims 1 to 8.