Communication cable hanging-up identification method and system based on deep learning

Abstract

The invention discloses a communication cable hanging-up identification method and system based on deep learning, and the method comprises the following steps: obtaining the video stream data of a communication cable provided by an external camera; extracting two frames in the video stream data to obtain two images of the communication cable; performing communication cable identification, edge detection and bounding rectangle calculation on the two images in sequence to obtain bounding rectangle data of the two images; and comparing the external rectangular data of the two images, and judging whether the communication cable is hung up or not. According to the invention, whether the communication cable is hung up can be effectively identified. Compared with a traditional network ping test method and the like, the method has higher certainty and accuracy in the aspect of judging the hanging-up of the communication cable. The video stream data is processed and analyzed through the deep learning algorithm, and the abnormal change of the cable state can be accurately captured, so that the quick detection and response to the cable hanging-up event are realized.

Description

Technical Field

[0001] The present invention relates to the technical field of communication cable hang-up identification, and in particular to a communication cable hang-up identification method and system based on deep learning. Background Art

[0002] During the daily operation and maintenance of communication lines, overhead cables may be disconnected by large construction vehicles or other factors. After a cable is disconnected, network maintenance personnel on the network management side only see that the peer device cannot be pinged, and are unable to determine the specific cause. Some services mounted on overhead cables are not frequently used, making it difficult for network maintenance personnel to detect them in a timely manner. Therefore, a deep learning-based communication cable disconnection identification method and system are designed. By installing cameras to obtain real-time monitoring video of overhead cables at important nodes, the system can identify whether the communication cable is disconnected based on the video images. Summary of the Invention

[0003] According to an embodiment of the present invention, in a first aspect, a method for identifying a disconnected communication cable based on deep learning is provided, comprising the following steps: Obtain video stream data from a communication cable provided by an external camera; Extracting two frames from the video stream data to obtain two images of the communication cable; Perform communication cable recognition, edge detection, and bounding rectangle calculation on both images in sequence to obtain bounding rectangle data of the two images; Compare the bounding rectangle data of the two images to determine whether the communication cable is disconnected.

[0004] Furthermore, the identification of communication cables includes the following steps: Build a training dataset; Train the training data set through the U²Net model to obtain the trained model; The trained model is used to identify the two images and obtain the recognition results of the communication cables in the two images.

[0005] Furthermore, the data of the training dataset includes the following: Extract relevant data of communication cables from the collected public data sets; Download images related to communication cables from the Internet and manually label them; A simulated test environment was set up on site to record a video of the communication cable changing from normal to disconnected state, and a portion of images were extracted from the video for manual annotation.

[0006] Furthermore, edge detection includes the following steps: The recognition results of the communication cables in the two images are analyzed using the Canny operator of OpenCV to obtain the edge point sets of the communication cables in the two images.

[0007] Furthermore, the calculation of the bounding rectangle includes the following steps: The communication cable edge point sets of the two images are taken as input, and the minimum bounding rectangle data of the two images are obtained by quantitative calculation.

[0008] Furthermore, the quantitative algorithm includes the following steps: Use Graham's algorithm to solve the convex hull point set of two images; Determine the minimum bounding rectangle data of the two images by boundary rotation.

[0009] Furthermore, when extracting two frames from the video stream data, a time interval parameter may be added to the two frames.

[0010] According to a second aspect of an embodiment of the present invention, a communication cable hang-up identification system based on deep learning is provided, comprising: An acquisition module is used to acquire video stream data of a communication cable provided by an external camera; An extraction module, used for extracting two frames from the video stream data to obtain two images of the communication cable; A processing module, configured to perform communication cable recognition, edge detection, and bounding rectangle calculation on both images in sequence, to obtain bounding rectangle data of the two images; The comparison module is used to compare the circumscribed rectangle data of the two images to determine whether the communication cable is disconnected.

[0011] According to a third aspect of an embodiment of the present invention, there is provided an electronic device, including a processor and a memory; The memory is used to store one or more computer programs; when the one or more computer programs stored in the memory are executed by the processor, the electronic device can implement a communication cable hang-up identification method based on deep learning in the first aspect.

[0012] According to an embodiment of the present invention, in a fourth aspect, a computer storage medium, a computer-readable storage medium includes a computer program, which, when the computer program runs on an electronic device, enables the electronic device to execute a communication cable hang-up identification method based on deep learning in the first aspect.

[0013] According to an embodiment of the present invention, a method and system for identifying disconnected communication cables based on deep learning can effectively identify whether a communication cable has been disconnected. Compared with traditional methods such as network ping tests, this method has higher certainty and accuracy in determining whether a communication cable has been disconnected. By processing and analyzing video stream data using a deep learning algorithm, abnormal changes in cable status can be accurately captured, enabling rapid detection and response to cable disconnection events. Currently, no other similar technologies have been found to use deep learning to identify disconnected cables using video stream data, thus filling an industry gap.

[0014] It is to be understood that both the foregoing general description and the following detailed description are exemplary, and are intended to provide further explanation of the technology as claimed. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 The figure is a flowchart of a method for identifying a disconnected communication cable based on deep learning according to an embodiment of the present invention.

[0016] Figure 2 This is a process diagram of identifying a communication cable in a method for identifying a communication cable hang-up based on deep learning according to an embodiment of the present invention.

[0017] Figure 3 This figure shows an example of a data set for identifying communication cables in a method for identifying communication cable hang-up based on deep learning according to an embodiment of the present invention.

[0018] Figure 4 This is a comparison chart of communication cable recognition results obtained by two models in a communication cable hang-up recognition method based on deep learning according to an embodiment of the present invention.

[0019] Figure 5 This figure shows the loss curve of the U²Net model training in a communication cable hang-up recognition method based on deep learning according to an embodiment of the present invention.

[0020] Figure 6 This is a schematic diagram of a communication cable recognition result in a communication cable hang-up recognition method based on deep learning according to an embodiment of the present invention.

[0021] Figure 7 The figure is a schematic diagram of detection results of edge detection in a method for identifying a communication cable hang-up based on deep learning according to an embodiment of the present invention.

[0022] Figure 8 Schematic diagram of solving the convex hull point set of a target image in a communication cable hang-up recognition method based on deep learning according to an embodiment of the present invention.

[0023] Figure 9This is a schematic diagram of determining a minimum circumscribed rectangle by boundary rotation in a method for identifying a communication cable hang-up based on deep learning according to an embodiment of the present invention.

[0024] Figure 10 Schematic diagram of calculation results of a circumscribed rectangle in a communication cable hang-up recognition method based on deep learning according to an embodiment of the present invention.

[0025] Figure 11 The present invention is a schematic diagram of test results of a communication cable hang-up simulation scenario in a communication cable hang-up identification method based on deep learning according to an embodiment of the present invention.

[0026] Figure 12 The block diagram of a communication cable hang-up recognition system based on deep learning according to an embodiment of the present invention.

[0027] Figure 13 The figure is a principle block diagram of an electronic device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0028] The preferred embodiments of the present invention will be described in detail below with reference to the accompanying drawings to further illustrate the present invention.

[0029] First, combine Figures 1 to 11 A method for identifying a communication cable disconnection based on deep learning according to an embodiment of the present invention is described. The method is used for identifying a communication cable disconnection and has a wide range of application scenarios.

[0030] like Figures 1 to 11 As shown, a method for identifying a disconnected communication cable based on deep learning according to an embodiment of the present invention comprises the following steps: S1: Obtain video stream data of communication cables provided by external cameras. Install cameras to obtain real-time monitoring video of communication cables at important nodes.

[0031] S2: Extract two frames from the video stream data to obtain two images of the communication cable.

[0032] S3: performing communication cable recognition, edge detection, and bounding rectangle calculation on both images in sequence to obtain bounding rectangle data of the two images; Furthermore, in the embodiment, S31: identifying the communication cable, comprises the following steps: S311: Build a training dataset.

[0033] Currently, there are no publicly available deep learning datasets specifically for communication cable identification. Existing datasets are mostly comprehensive identification datasets for various cables, including transmission lines and communication optical cables. Communication cables differ significantly from electrical cables in terms of surface gloss and thickness. Currently, cable identification is widely used. Therefore, the proportion of cable samples in datasets is relatively high, and publicly available datasets related to communication cables are very scarce. Therefore, in this embodiment, the data sources for the training dataset include the following: 1) Extract relevant data of communication cables from the collected public datasets; 2) Download images related to communication cables from the Internet and manually label them; 3) Set up a simulated test environment on site, record a video of the communication cable changing from normal to disconnected state, and extract a portion of the image from the video for manual annotation. Figure 3 As described above, part of the communication cable dataset is shown, with sample images on the left and sample labels on the right.

[0034] S312: Train the training dataset using the U²Net model to obtain a trained model.

[0035] Model selection: Since the UNet model is well-established in the industry, we initially attempted to use it to identify communication cables. However, this model would identify poles and other columns as cables. We then attempted to use the U²Net model, which showed better cable recognition and was able to accurately identify communication cables. The following table compares the recognition results of the two models for communication cables: Figure 4 shown.

[0036] from Figure 4 As can be seen, the UNet model misidentifies utility poles as communication cables, but tends to miss the areas of the communication cables with strong light reflection, and the segmentation mask is not smooth. This is primarily due to the lack of such examples in the dataset, resulting in poor generalization of the UNet model in this scenario. The U²Net model, on the other hand, has a stronger ability to recognize edge contours and more complete recognition of communication cables, able to distinguish between communication cables and utility poles. Furthermore, it can well identify the areas of the communication cables with strong light reflection.

[0037] Comparison results show that the UNet model has lower accuracy than the U²Net model in identifying communication cables, and secondly, its generalization performance is poor. The U²Net model outperforms the U²Net model primarily because it can integrate highly granular local information with global information, while UNet can achieve global perception but cannot take into account highly granular local information. Therefore, the U²Net model is more suitable for communication cable recognition. Other deep learning models can also be used here, as long as they can accurately identify communication cables from images.

[0038] Model training: During training, the Epoch is set to 400, the batch size is set to 1, and the Adam optimization function is used to participate in the training. It can adaptively select hyperparameters without relying on the initial parameter settings. The training time for each Epoch is about 17 seconds, and the training loss is as follows: Figure 5 As shown in the figure, the model gradually converges after 300 epochs, and the loss does not change after converging to about 0.08.

[0039] S313: Use the trained model to identify the two images and obtain the recognition results of the communication cables in the two images. Figure 6 As shown in the figure, the left side is the original image of the test scene, and the right side is the recognition result of the communication cable identified by the U²Net model.

[0040] Further, if Figure 7 As shown, in this embodiment, S32: edge detection includes the following steps: S321: The recognition results of the communication cables in the two images are analyzed using the Canny operator of OpenCV to obtain a set of edge points of the communication cables in the two images.

[0041] Since the set of communication cable edge points obtained in step S321 is an irregular polygon, in the process of determining whether the line is disconnected, the irregular polygon is not suitable for quantitative calculation. Therefore, based on the previous step, the minimum circumscribed rectangle of the polygon is calculated as the basis for subsequent judgment.

[0042] Furthermore, in this embodiment, S33: calculating the circumscribed rectangle includes the following steps: S331: Take the communication cable edge point set of the two images as input and calculate the minimum bounding rectangle data of the two images through quantitative algorithm. Figure 10 As shown, the left side is the calculated circumscribed rectangular frame, and the right side is the display result of superimposing the rectangular frame on the video image.

[0043] The process of calculating the bounding rectangle is mainly divided into two steps: (1) Using Graham's algorithm to solve the convex hull point set of the target image The schematic diagram of solving the convex hull point set of the target image is as follows Figure 8 As shown, the specific steps are: 1) First find the point P0 with the smallest ordinate in the point set, and connect it with other points to obtain the line segment set L P , calculate L P The length of each line segment and the angle with the positive direction of the X axis are used to number the points from small to large according to the angle and line segment length (P1, P2, ···P n ); 2) Determine the difference between P and i , P i+1 Are all points other than point distributed on line segment P? i P i+1 (i=0, 1, ···, n-1), if it meets the requirement, continue to select the next line segment for judgment, if not, abandon the end point P of the current line segment i+1 , the following point P i+j (j=2, 3, ···, n) as the new end point, connecting the line segment P i P i+j Repeat the judgment until it is satisfied. According to this rule, the final convex hull boundary and vertex set W = {P0, P1, P3, P6, P8, P9} can be obtained.

[0044] (2) Determine the minimum enclosing rectangle by boundary rotation The schematic diagram of determining the minimum enclosing rectangle by boundary rotation is as follows Figure 9 As shown, the specific steps are: 1) Based on the W point set, find the points with the maximum and minimum coordinates in the X-axis and Y-axis directions respectively, and then you can get the initial minimum enclosing rectangle Rect0. Record the area and the positions of the four vertices of Rect0.

[0045] 2) Assume that the coordinates of the K-th point K=1,...,n in the W point set are (X K , Y K ), the angle between the Kth boundary line and the positive direction of the X-axis is θ K , rotate the original image around the origin of the coordinate system by θ K Angle, the coordinates of the Kth point become ( , ), and get the rotated vertex set W1. According to the point set W1, find the maximum and minimum points in the coordinate axis direction respectively, so as to get the minimum circumscribed rectangle Rect1, and rotate Rect1 inversely around the origin of the coordinate system by θ K The angle is returned to the original image. According to this rule, all boundary rotations are completed and the minimum bounding rectangle set S can be obtained. R , take S RThe one with the smallest area is taken as the final minimum enclosing rectangle.

[0046] Among them, calculate θ K The angle and rotated coordinate formulas are as follows:

[0047] S4: Compare the circumscribed rectangle data of the two images to determine whether the communication cable is disconnected.

[0048] The process of comparison and difference mainly includes calculating the number and aspect ratio of the bounding rectangles. Generally speaking, when a cable is disconnected, if the communication cable is not disconnected in the previous frame, there is a unique bounding rectangle for the communication cable. If the communication cable is disconnected in the next frame, the bounding rectangle will undergo three changes: 1) The number of bounding rectangles decreases, that is, the cable disappears from the video field of view after being disconnected.

[0049] 2) The number of circumscribed rectangles remains unchanged, but their shapes change significantly, i.e., after the cable is disconnected, a portion of it disappears from the field of view.

[0050] 3) The number of circumscribed rectangles increases, which means that the cable is broken into two parts, and both parts appear in the field of view.

[0051] For the above three situations, when the number of bounding rectangles remains unchanged, the degree of change is determined by comparing the aspect ratios of the bounding rectangles in the previous and next frames; when the number of bounding rectangles changes, it can be directly determined that the cable is disconnected.

[0052] Since it is difficult to capture the hang-up scene of a communication cable in real life, we simulated several communication hang-up situations and recorded the communication cable hang-up scene video for algorithm testing. According to the above judgment method, experiments were conducted in the simulated scene, and the results are as follows Figure 11 As shown: Four cable disconnection situations are simulated in the scene, Figure 11 As can be seen from the figure, when the communication cable is disconnected, the algorithm successfully issues an alarm.

[0053] Furthermore, in this embodiment, when extracting two frames from a video stream, a time interval parameter can be added to the two frames. Since the video stream data volume is huge, real-time calculation would consume a significant amount of computing power. Adding the time interval parameter to the two frames is more suitable for practical engineering needs.

[0054] As described above, a deep learning-based communication cable disconnection detection method according to an embodiment of the present invention can effectively identify whether a communication cable has been disconnected. Compared with traditional methods such as network ping tests, this method has higher certainty and accuracy in determining whether a communication cable has been disconnected. By processing and analyzing video stream data using a deep learning algorithm, abnormal changes in cable status can be accurately captured, enabling rapid detection and response to cable disconnection events. Currently, no other similar technologies have been found that utilize deep learning to detect cable disconnection in video stream data, thus filling an industry gap.

[0055] The above combined with the attached Figures 1 to 11 A method for identifying a communication cable disconnection based on deep learning according to an embodiment of the present invention is described. Furthermore, the present invention can also be applied to a communication cable disconnection identification system based on deep learning.

[0056] like Figure 12 As shown, according to an embodiment of the present invention, in a second aspect, a communication cable hang-up identification system based on deep learning is provided, comprising: An acquisition module 100 is used to acquire video stream data of a communication cable provided by an external camera; An extraction module 200 is used to extract two frames from the video stream data to obtain two images of the communication cable; The processing module 300 is used to sequentially perform communication cable recognition, edge detection, and bounding rectangle calculation on the two images to obtain bounding rectangle data of the two images; The comparison module 400 is used to compare the circumscribed rectangle data of the two images to determine whether the communication cable is disconnected.

[0057] The above combined with the attached Figure 12 A terrain classification system based on SAR altimeter delay Doppler images according to an embodiment of the present invention is described. Furthermore, the present invention can also be applied to an electronic device.

[0058] like Figure 13 As shown, according to an embodiment of the present invention, in a third aspect, an electronic device is provided. The electronic device may include: one or more processors 1401, a memory 1402, and one or more computer programs 1403. The above components may be connected via one or more communication buses. The one or more computer programs 1403 are stored in the memory 1402 and configured to be executed by the one or more processors 1401. The one or more computer programs 1403 include instructions that enable the electronic device to execute a method for identifying a communication cable hang-up based on deep learning according to the first aspect of the embodiment of the present invention.

[0059] The above combined with the attached Figure 13An electronic device according to an embodiment of the present invention is described. Furthermore, the present invention can also be applied to a computer storage medium.

[0060] According to an embodiment of the present invention, in a fourth aspect, a computer storage medium, the computer readable storage medium includes a computer program, when the computer program is run on an electronic device, the electronic device executes a communication cable hang-up recognition method based on deep learning in the first aspect. Through the description of the above embodiments, it will be clear to those skilled in the art that the embodiments of the present application can be implemented in hardware, firmware, or a combination thereof. When software is used for implementation, the above functions can be stored in a computer-readable medium or transmitted as one or more instructions or codes on a computer-readable medium. Computer-readable media include computer storage media and communication media, wherein the communication media include any medium that facilitates the transmission of a computer program from one place to another. The storage medium can be any available medium that a computer can access. For example, but not limited to: a computer-readable medium may include RAM, ROM, electrically erasable programmable read-only memory (EEPROM), a compact disc (CD-ROM) or other optical disc storage, a magnetic disk storage medium or other magnetic storage device, or any other medium that can be used to carry or store a desired program code in the form of an instruction or data structure and can be accessed by a computer. In addition, any connection can be appropriately a computer-readable medium. For example, if the software is transmitted from a website, server, or other remote source using a coaxial cable, fiber optic cable, twisted pair, digital subscriber line (DSL), or wireless technologies such as infrared, radio, and microwave, then the coaxial cable, fiber optic cable, twisted pair, DSL, or wireless technologies such as infrared, radio, and microwave are included in the definition of the medium. As used in the embodiments of the present application, disk and disc include compact disc (CD), laser disc, optical disc, digital versatile disc (DVD), floppy disk, and Blu-ray disc, where disks typically reproduce data magnetically and discs use lasers to reproduce data optically. Combinations of the above should also be included within the scope of protection of computer-readable media.

[0061] It should be noted that, in this specification, the terms "comprises," "includes," or any other variations thereof are intended to encompass non-exclusive inclusion, such that a process, method, article, or apparatus comprising a series of elements includes not only those elements but also other elements not explicitly listed, or elements inherent to such process, method, article, or apparatus. In the absence of further limitations, elements defined by the phrase "comprising..." do not exclude the presence of other identical elements in the process, method, article, or apparatus comprising the elements.

[0062] Although the present invention has been described in detail through the above preferred embodiments, it should be understood that the above description is not intended to limit the present invention. After reading the above description, various modifications and substitutions of the present invention will become apparent to those skilled in the art. Therefore, the scope of protection of the present invention should be defined by the appended claims.

Claims

1. A communication cable hang-up recognition method based on deep learning, characterized in that: The following steps are included: Obtain video stream data from a communication cable provided by an external camera; Extracting two frames from the video stream data to obtain two images of the communication cable; Perform communication cable recognition, edge detection, and bounding rectangle calculation on both images in sequence to obtain bounding rectangle data of the two images; The circumscribed rectangle data of the two images are compared to determine whether the communication cable is disconnected.

2. The method for identifying a disconnected communication cable based on deep learning according to claim 1, wherein: The identification of the communication cable comprises the following steps: Build a training dataset; Train the training data set through the U²Net model to obtain the trained model; The trained model is used to identify the two images and obtain the recognition results of the communication cables in the two images.

3. The method for identifying a disconnected communication cable based on deep learning according to claim 2, wherein: The data of the training dataset includes the following: Extract relevant data of communication cables from the collected public data sets; Download images related to communication cables from the Internet and manually label them; A simulated test environment was set up on site to record a video of the communication cable changing from normal to disconnected state, and a portion of images were extracted from the video for manual annotation.

4. The method for identifying a disconnected communication cable based on deep learning according to claim 2, wherein: The edge detection comprises the following steps: The recognition results of the communication cables in the two images are analyzed using the Canny operator of OpenCV to obtain the edge point sets of the communication cables in the two images.

5. The method for identifying a disconnected communication cable based on deep learning according to claim 4, wherein: The calculation of the bounding rectangle includes the following steps: The communication cable edge point sets of the two images are taken as input, and the minimum bounding rectangle data of the two images are obtained by quantitative calculation.

6. The method for identifying a disconnected communication cable based on deep learning according to claim 5, wherein: The quantitative algorithm comprises the following steps: Use Graham's algorithm to solve the convex hull point set of two images; Determine the minimum bounding rectangle data of the two images by boundary rotation.

7. The method for identifying a disconnected communication cable based on deep learning according to claim 1, wherein: When extracting two frames from the video stream data, a time interval parameter may be added to the two frames.

8. A communication cable hang-up recognition system based on deep learning, characterized in that: Include: An acquisition module is used to acquire video stream data of a communication cable provided by an external camera; an extraction module, configured to extract two frames from the video stream data to obtain two images of the communication cable; A processing module, configured to perform communication cable recognition, edge detection, and bounding rectangle calculation on both images in sequence, to obtain bounding rectangle data of the two images; The comparison module is used to compare the circumscribed rectangle data of the two images to determine whether the communication cable is disconnected.

9. An electronic device, characterized in that: including processor and memory; Wherein, the memory is used to store one or more computer programs; when the one or more computer programs stored in the memory are executed by the processor, the electronic device is able to implement the communication cable hang-up identification method based on deep learning as described in any one of claims 1 to 7.

10. A computer storage medium, characterized in that The computer-readable storage medium includes a computer program, which, when executed on an electronic device, enables the electronic device to execute the communication cable hang-up identification method based on deep learning as described in any one of claims 1 to 7.

Images