Machine room equipment maintenance method and device, computer equipment, readable storage medium and program product

Abstract

The invention relates to a machine room equipment maintenance method and device, computer equipment, a readable storage medium and a program product. The method comprises the steps that an original image of a machine cabinet in a machine room at the current moment is acquired; performing correction processing on the original image to obtain a corrected image; performing target detection processing on the corrected image to obtain equipment information corresponding to equipment in the cabinet; the equipment information at least comprises a bounding box of the equipment; and comparing the equipment information at the current moment with the equipment information of the cabinet at the previous moment to obtain change information of the equipment in the cabinet, and recording the change information. By adopting the method, the equipment change condition can be effectively recorded and tracked, the management and maintenance efficiency of the equipment is improved, the original image is corrected, image distortion and deformation caused by factors such as a shooting angle and illumination can be eliminated, and thus the accuracy of subsequent target detection is improved.

Description

Technical Field

[0001] The present application relates to the field of computer vision technology, and in particular to a method and apparatus for maintaining equipment in a computer room, computer equipment, a computer-readable storage medium, and a computer program product. Background Art

[0002] In computer room management, the change, maintenance and upgrade of equipment are usually done manually by making change plans and recording the change process. However, this method is highly dependent on manual work, and the frequent loading and unloading of equipment in computer rooms makes management work cumbersome and time-consuming, with low efficiency, and it is difficult to meet the requirements of modern data centers for flexibility and responsiveness. Summary of the invention

[0003] Based on this, it is necessary to provide a computer room equipment maintenance method, device, computer equipment, computer-readable storage medium and computer program product to address the technical problems that the above-mentioned methods have low management efficiency and are difficult to meet the requirements of modern data centers for flexibility and response speed.

[0004] In a first aspect, the present application provides a method for maintaining equipment in a computer room. The method comprises:

[0005] Obtain the original image of the cabinet in the computer room at the current moment;

[0006] Performing correction processing on the original image to obtain a corrected image;

[0007] Performing target detection processing on the rectified image to obtain device information corresponding to the device in the cabinet; the device information at least includes a boundary box of the device;

[0008] The device information at the current moment is compared with the device information of the cabinet at the previous moment to obtain change information of the devices in the cabinet, and the change information is recorded.

[0009] In one embodiment, the correcting the original image to obtain the corrected image includes:

[0010] Performing semantic segmentation on the original image to obtain a cabinet area, and performing edge detection on the cabinet area to obtain a contour of the cabinet;

[0011] Extracting correction positioning points from the outline of the cabinet, and determining a perspective transformation matrix based on the correction positioning points;

[0012] The original image is corrected according to the perspective transformation matrix to obtain the corrected image.

[0013] In one embodiment, performing target detection processing on the corrected image to obtain device information corresponding to the device in the cabinet includes:

[0014] Performing target detection processing on the rectified image through a first target detection model to obtain first device information corresponding to the device in the cabinet; the first device information includes a bounding box and a device model of the device;

[0015] Performing target detection processing on the rectified image through a second target detection model to obtain second device information corresponding to the device in the cabinet; the second device information includes a bounding box of the device;

[0016] Deduplication processing is performed on the first device information and the second device information to obtain the device information.

[0017] In one embodiment, the performing deduplication processing on the first device information and the second device information to obtain the device information includes:

[0018] When the overlapping area ratio of the bounding box of any device in the second device information and the bounding box of any device in the first device information exceeds a predetermined threshold, deleting the device information of the any device in the second device information to obtain the second device information after deduplication;

[0019] The deduplicated second device information and the first device information are combined to form device information corresponding to the device in the cabinet.

[0020] In one embodiment, after the correction processing is performed on the original image to obtain the corrected image, the method further includes:

[0021] By using a third target detection model, target detection processing is performed on the corrected image to obtain label information and identification code information corresponding to the equipment in the cabinet; the label information includes a bounding box of the label, and the identification code information includes a bounding box of the identification code;

[0022] Extracting text information from a label corresponding to the device, and decoding an identification code of the device to obtain a device number of the device;

[0023] Verify the consistency among the device number, the text information, the label information and the device information, as well as the relevant information of the device recorded in the resource system of the computer room, and generate a prompt message if there is any inconsistency.

[0024] In one embodiment, the method further comprises:

[0025] Establishing a three-dimensional model of the cabinet according to the corrected image;

[0026] According to the device information corresponding to the device in the cabinet, the corresponding device in the three-dimensional model is marked to obtain the marked three-dimensional model of the cabinet;

[0027] A digital three-dimensional computer room corresponding to the computer room is generated according to the annotated three-dimensional models of all the cabinets in the computer room.

[0028] In a second aspect, the present application also provides a computer room equipment maintenance device. The device comprises:

[0029] An acquisition module is used to acquire the original image of the cabinet in the computer room at the current moment;

[0030] A correction module, used for performing correction processing on the original image to obtain a corrected image;

[0031] A detection module, configured to perform target detection processing on the rectified image to obtain device information corresponding to the device in the cabinet; the device information at least includes a boundary box of the device;

[0032] The comparison module is used to compare the device information at the current moment with the device information of the cabinet at the previous moment, obtain the change information of the device in the cabinet, and record the change information.

[0033] In a third aspect, the present application further provides a computer device. The computer device includes a memory and a processor, the memory stores a computer program, and the processor implements the following steps when executing the computer program:

[0034] Obtain the original image of the cabinet in the computer room at the current moment;

[0035] Performing correction processing on the original image to obtain a corrected image;

[0036] Performing target detection processing on the rectified image to obtain device information corresponding to the device in the cabinet; the device information at least includes a boundary box of the device;

[0037] The device information at the current moment is compared with the device information of the cabinet at the previous moment to obtain change information of the devices in the cabinet, and the change information is recorded.

[0038] In a fourth aspect, the present application further provides a computer-readable storage medium. The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0039] Obtain the original image of the cabinet in the computer room at the current moment;

[0040] Performing correction processing on the original image to obtain a corrected image;

[0041] Performing target detection processing on the rectified image to obtain device information corresponding to the device in the cabinet; the device information at least includes a boundary box of the device;

[0042] The device information at the current moment is compared with the device information of the cabinet at the previous moment to obtain change information of the devices in the cabinet, and the change information is recorded.

[0043] In a fifth aspect, the present application further provides a computer program product. The computer program product includes a computer program, and when the computer program is executed by a processor, the following steps are implemented:

[0044] Obtain the original image of the cabinet in the computer room at the current moment;

[0045] Performing correction processing on the original image to obtain a corrected image;

[0046] Performing target detection processing on the rectified image to obtain device information corresponding to the device in the cabinet; the device information at least includes a boundary box of the device;

[0047] The device information at the current moment is compared with the device information of the cabinet at the previous moment to obtain change information of the devices in the cabinet, and the change information is recorded.

[0048] The above-mentioned computer room equipment maintenance method, device, computer equipment, storage medium and computer program product obtain the corrected image after obtaining the original image of the cabinet in the computer room at the current moment; perform target detection processing on the corrected image to obtain the equipment information corresponding to the equipment in the cabinet; the equipment information at least includes the boundary box of the equipment; compare the equipment information at the current moment with the equipment information of the cabinet at the previous moment, obtain the change information of the equipment in the cabinet, and record the change information. The method realizes automatic recognition of equipment information by collecting cabinet images, performing target detection on the cabinet images, and further comparing the equipment information of the cabinet images collected at the current moment with the equipment information at the previous moment, which can realize effective recording and tracking of equipment changes, provide automated data support for computer room management, reduce the workload of manual inspection, and thus improve the management and maintenance efficiency of equipment. In addition, the correction processing of the original image can eliminate image distortion and deformation caused by factors such as shooting angle and lighting, thereby improving the accuracy of subsequent target detection. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 An application environment diagram of a computer room equipment maintenance method in an embodiment;

[0050] Figure 2 A schematic diagram of a process flow of a method for maintaining equipment in a computer room according to an embodiment;

[0051] Figure 3 is a schematic diagram of a correction process for correcting an original image of a cabinet in one embodiment;

[0052] Figure 4 A schematic diagram of a process for auditing labels and resource systems of equipment in a computer room in one embodiment;

[0053] Figure 5 A schematic diagram of a flow chart of a method for maintaining equipment in a computer room in another embodiment;

[0054] Figure 6 A structural block diagram of a computer room equipment maintenance device in one embodiment;

[0055] Figure 7 FIG. 4 is a diagram showing the internal structure of a computer device in one embodiment. DETAILED DESCRIPTION

[0056] In order to make the purpose, technical solution and advantages of the present application more clearly understood, the present application is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and are not used to limit the present application.

[0057] It should be noted that the terms "first", "second", etc. in the specification and claims of the present application and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the numbers used in this way can be interchanged where appropriate, so that the embodiments of the present application described herein can be implemented in an order other than those illustrated or described herein.

[0058] The computer room equipment maintenance method provided in the embodiment of the present application can be applied to Figure 1 In the application environment shown. Among them, the processing device 104 communicates with the image collector 102 through the network. The data storage system can store the data that the processing device 104 needs to process. The data storage system can be integrated on the processing device 104. In the application scenario of the present application, the user collects the original image of the cabinet in the computer room at the current moment through the image collector 102, and sends it to the processing device 104, so that the processing device 104 obtains the original image of the cabinet, and corrects the original image to obtain the corrected image; performs target detection processing on the corrected image to obtain the device information corresponding to the device in the cabinet; the device information at least includes the boundary box of the device; compares the device information at the current moment with the device information of the cabinet at the previous moment, obtains the change information of the device in the cabinet, and records the change information.

[0059] The image collector 102 is a device with an image acquisition function, for example, it can be but not limited to various cameras, smart phones and tablet computers, etc. The processing device 104 can be a terminal or a server, and the terminal can be but not limited to various personal computers, laptops, smart phones, tablet computers, etc. The server can be implemented as an independent server or a server cluster composed of multiple servers.

[0060] In one embodiment, Figure 2 As shown, a method for maintaining equipment in a computer room is provided. Figure 1 Taking the processing device 104 in the example as an example, the method includes the following steps:

[0061] Step S210, obtaining the original image of the cabinet in the computer room at the current moment.

[0062] In a specific implementation, the user can use the handheld image collector 102 to collect the image of the cabinet in the computer room at the current moment as the original image, and then send it to the processing device 104 for subsequent processing by the processing device 104.

[0063] There may be one or more cabinets in the computer room. When there are more than one cabinet, the original image of each cabinet is collected respectively, and the subsequent processing steps are performed.

[0064] Step S220, performing correction processing on the original image to obtain a corrected image.

[0065] In a specific implementation, considering that the original image collected by the image collector 102 may have deformation problems, after obtaining the original image, the original image may be corrected first. The correction process may be implemented based on semantic segmentation and perspective transformation. First, the cabinet area is identified by semantic segmentation, and then the cabinet outline is extracted by edge detection. Then, the correction positioning points are extracted from the cabinet outline. The perspective transformation matrix is ​​determined according to the correction positioning points. The corresponding pixel points in the corrected image are obtained by mapping the perspective transformation matrix, and the vacancies in the corrected image are supplemented to achieve image correction.

[0066] Step S230, performing target detection processing on the corrected image to obtain device information corresponding to the device in the cabinet; the device information at least includes a boundary box of the device.

[0067] Among them, the bounding box (Bounding Box), also known as the bounding volume or bounding area, is a rectangular box used to describe the position and range of the target in the image. It is defined by the coordinates of the upper left corner and the lower right corner of the rectangular box, so it is used to mark and locate the target object, that is, the bounding box of the device can reflect the location of the device.

[0068] In a specific implementation, target detection on the rectified image can be performed by pre-training a target detection model, inputting the rectified image into the target detection model for detection, and obtaining the bounding box of the device in the cabinet. It can be understood that there can be multiple devices in the cabinet. If there are multiple devices, each device will detect the corresponding bounding box. Among them, the target detection model can be a Yolo model (You Only Look Once, a deep learning model for target detection), an R-CNN model (Region-based Convolutional Neural Networks, a detection model based on convolutional neural networks), etc., and this application does not make specific limitations on this.

[0069] Furthermore, the detection results obtained by performing target detection on the rectified image may also include the bounding box of the cabinet and the bounding box of the equipment outside the cabinet. Therefore, after obtaining the detection results of the rectified image, the detection results may be filtered to remove the detection results of the equipment outside the bounding box of the cabinet. In addition, each bounding box in the detection result has a corresponding confidence score, which is used to characterize the probability of containing the equipment in the bounding box. Filtering the detection results may also filter out the detection results with a confidence lower than a confidence threshold (e.g., 0.8), so as to improve the accuracy of the equipment information corresponding to the equipment in the cabinet obtained in the end.

[0070] Step S240, comparing the device information at the current moment with the device information of the cabinet at the previous moment, obtaining the change information of the devices in the cabinet, and recording the change information.

[0071] In the specific implementation, after obtaining the device information corresponding to the cabinet image collected at the current moment, the device information can be compared with the device information determined by the cabinet image collected at the previous moment, and the newly added and reduced equipment between the two inspections can be recorded to realize intelligent maintenance of the changes in the equipment in the cabinet.

[0072] In the above-mentioned computer room equipment maintenance method, after obtaining the original image of the cabinet in the computer room at the current moment, a corrected image is obtained; target detection processing is performed on the corrected image to obtain the device information corresponding to the device in the cabinet; the device information at least includes the bounding box of the device; the device information at the current moment is compared with the device information of the cabinet at the previous moment to obtain the change information of the device in the cabinet, and the change information is recorded. This method realizes automatic recognition of device information by collecting cabinet images and performing target detection on the cabinet images, and further compares the device information of the cabinet images collected at the current moment with the device information at the previous moment, so as to realize effective recording and tracking of equipment changes, provide automated data support for computer room management, reduce the workload of manual inspection, and thus improve the management and maintenance efficiency of equipment. In addition, correcting the original image can eliminate image distortion and deformation caused by factors such as shooting angle and lighting, thereby improving the accuracy of subsequent target detection.

[0073] In an exemplary embodiment, the above-mentioned step S220 performs correction processing on the original image to obtain a corrected image, specifically including: performing semantic segmentation on the original image to obtain a cabinet area, performing edge detection on the cabinet area to obtain the outline of the cabinet; extracting correction positioning points from the outline of the cabinet, and determining a perspective transformation matrix based on the correction positioning points; and performing correction processing on the original image based on the perspective transformation matrix to obtain a corrected image.

[0074] In the specific implementation, the original image can be semantically segmented to obtain the cabinet area in the original image, and then edge detection algorithms such as the Canny edge algorithm can be used to detect the edge of the cabinet area and extract the cabinet outline. Semantic segmentation is used to classify each pixel in the image and divide the image into several meaningful areas. Canny edge detection detects the edge pixels of the cabinet area to obtain the cabinet outline.

[0075] Perspective-based image correction requires locating the reference point of the transformed object. Therefore, after obtaining the outline of the cabinet through semantic segmentation and edge detection, the correction positioning points for correction can be extracted from the outline. The outline of the cabinet is generally a quadrilateral, so the four points closest to the image vertices can be extracted from the outline of the cabinet, namely the upper left, upper right, lower left and lower right points, as correction positioning points, and the coordinates of the four correction positioning points in the original image can be determined. Then define the ideal coordinates of the correction positioning points in the corrected image, for example, the lower left point (0,0), the upper left point (0,h), the lower right point (w,0), and the upper right point (w,h), where w and h represent the width and height of the cabinet, respectively. Substitute the coordinates of the correction positioning points in the original image and the corrected image into the perspective transformation formula to solve the perspective transformation matrix.

[0076] Suppose a plane coordinate point on the two-dimensional image is , the corresponding coordinate points on the rectified image are , then the perspective transformation relationship can be expressed as:

[0077] (1)

[0078] Wherein, T represents the perspective transformation matrix. After obtaining the perspective transformation matrix, the original image can be corrected based on the perspective transformation matrix, the pixel points in the original image can be mapped to the corrected image, and the gaps in the corrected image can be supplemented according to the difference method to complete the correction and obtain the corrected image.

[0079] For example, refer to Figure 3 , is a schematic diagram of a correction process for correcting an original image of a cabinet according to an embodiment. For the collected original image of the cabinet, the cabinet area is first detected by semantic segmentation, and then the edge detection is performed on the cabinet area to extract the outline of the cabinet, and four corner points are extracted from the outline of the cabinet as correction positioning points. The perspective transformation matrix is ​​calculated through the four positioning points, and the perspective transformation matrix is ​​used to correct the original image to obtain a corrected image.

[0080] In this embodiment, the cabinet area is obtained by semantic segmentation of the original image, and the edge detection of the cabinet area is performed to obtain the outline of the cabinet. The correction positioning points are extracted from the outline of the cabinet, and the perspective transformation matrix is ​​determined according to the correction positioning points; the original image is corrected according to the perspective transformation matrix to obtain a corrected image. In this way, the correction of the original image of the cabinet is achieved, and the deformation problem of the original image caused by the shooting angle is solved, so that the subsequent target detection results based on the corrected image are more accurate.

[0081] In an exemplary embodiment, the above step S230, performing target detection processing on the corrected image to obtain device information corresponding to the device in the cabinet, includes: performing target detection processing on the corrected image through a first target detection model to obtain first device information corresponding to the device in the cabinet; the first device information includes a bounding box and a device model of the device; performing target detection processing on the corrected image through a second target detection model to obtain second device information corresponding to the device in the cabinet; the second device information includes a bounding box of the device; and deduplicating the first device information and the second device information to obtain the device information.

[0082] In the specific implementation, the first target detection model can detect the device model in addition to the device bounding box, compared to the second target detection model. That is, the first target detection model can identify cabinets and devices of different models, and the detection results are more precise. The second target detection model is mainly used to detect the device bounding box, but does not involve the detection of device models, so it can detect unknown models of devices and has stronger generalization ability.

[0083] Since the detection targets of the first target detection model and the second target detection model are different, the first target detection model and the second target detection model can be obtained by training respectively through different sample data. Specifically, the sample data for training the first target detection model may include the first sample cabinet image and the annotation of the device in the first sample cabinet image, and the annotation content may include the bounding box and the device model of the device. During training, the first sample cabinet image is used as the input variable, and the annotation of the bounding box and the device model is used as the supervision information to train the target detection model to obtain the first target detection model. The sample data for training the second target detection model may include the second sample cabinet image and the annotation of the device in the second sample cabinet image, and the annotation content may include the bounding box of the device, which is equivalent to uniformly annotating various types and specifications of devices as "devices" without distinguishing the device models, so as to improve the generalization ability of the second target detection model obtained by training, so that it can detect devices of unknown models. During training, the second sample cabinet image is used as the input variable, and the annotated device bounding box is used as the supervision information to train the target detection model to obtain the second target detection model. The model structures of the first target detection model and the second target detection model may be the same, that is, the same model may be used as the target detection model to be trained.

[0084] After the training of the two target detection models is completed, they can be further used to detect the rectified image of the cabinet. Specifically, the rectified image can be input into the first target detection model and the second target detection model respectively. The first target detection model detects the bounding box of the device in the cabinet and the specific device model, and the second target detection model detects the bounding box of the device in the cabinet. It can be understood that the cabinet can also be regarded as a device, and the two target detection models can also detect the bounding box of the cabinet to filter the detection results of the devices outside the bounding box of the cabinet.

[0085] Since the two target detection models perform target detection on the same rectified image, the detection results will be repeated, so it is also necessary to deduplicate the detection results of the two target detection models to avoid duplicate records. Since the second target detection model is a generalized detection and the first target detection model is a refined detection, when the two detection results are repeated, the detection results of the generalized detection should be removed, thereby achieving the ability to identify both the specific models of the equipment in the cabinet and the unknown models of the equipment. Furthermore, some of the equipment information obtained after deduplication has equipment models, and some are unknown models of equipment obtained through the generalized detection of the second target detection model. For equipment of unknown models, maintenance personnel can be prompted to supplement information in a timely manner to add the training data of the first target detection model for subsequent model update training.

[0086] In this embodiment, target detection is performed on the corrected image of the cabinet by using a first target detection model that can detect the device model and a second detection model for generalized detection of the device, which can identify both the specific model of the device in the cabinet and the device of unknown model, thereby avoiding missed detection of the device and ensuring the accuracy and reliability of the detection results. Further, by deduplicating the detection results of the two target detection models, repeated recording of the detection results of the same device can be avoided, reducing the workload of subsequent processing.

[0087] Furthermore, in an exemplary embodiment, the above-mentioned deduplication processing is performed on the first device information and the second device information to obtain the device information, and the step includes: when the ratio of the overlapping area of ​​the bounding box of any device in the second device information and the bounding box of any device in the first device information exceeds a predetermined threshold, deleting the device information of any device in the second device information to obtain the deduplicated second device information; and combining the deduplicated second device information and the first device information to form the device information corresponding to the device in the cabinet.

[0088] In a specific implementation, the second target detection model is a generalized detection, and the first target detection model is a refined detection. Therefore, when there are duplicates in the two detection results, the detection results of the generalized detection should be removed. More specifically, whether the two detection results are duplicated and need to be deleted can be determined by the overlapping area ratio of the bounding boxes in the two detection results. When the overlapping area ratio of the bounding boxes exceeds a predetermined threshold, such as 0.5, the two detection results are determined to be duplicates, and the corresponding information in the detection results of the second target detection model is deleted. After the deduplication identification is completed for the bounding boxes of all devices in the two target detection models, the deduplicated second device information is obtained, and the deduplicated second device information and the first device information are combined to obtain the device information corresponding to the device in the cabinet. The device information contains both the specific device model of the device that can identify the device model and the detection results of the unknown model device.

[0089] In this embodiment, by setting the overlapping area ratio threshold, duplicate records can be effectively identified and deleted to ensure the accuracy of the final device information; redundant data is removed by judging the overlap of the bounding boxes in the two detection results to ensure that each device information is unique and accurate, which can reduce the amount of device information data. At the same time, after deduplication, it can ensure that the device information from the two sources will not conflict, thereby improving the consistency of the overall data.

[0090] In addition to the management of the equipment in the computer room, the labels of the equipment in the computer room and the resource system of the computer room can also be maintained. Figure 4 As shown, after the original image is corrected in the above step S220 to obtain the corrected image, the following further includes:

[0091] Step S410, performing target detection processing on the corrected image through a third target detection model to obtain label information and identification code information corresponding to the equipment in the cabinet; the label information includes a bounding box of the label, and the identification code information includes a bounding box of the identification code.

[0092] The identification code may be a barcode, a QR code, etc. The label may be attached to the device.

[0093] Specifically, the model type of the third target detection model may be the same as or different from that of the first target detection model and the second target detection model, and the model type may be, for example, a Yolo model, an R-CNN model, etc. The third target detection model for locating the label and identification code of the device may be obtained in advance through training with the third sample data. The third sample data may include a third sample cabinet image and the annotation of the label and the annotation of the identification code in the third sample cabinet image. The annotation content of the label may be a bounding box of the label, and the annotation of the identification code may be a bounding box of the identification code. During training, the target detection model to be trained is trained with the third sample cabinet image as the input variable and the annotation of the label and the identification code as the supervision information to obtain the third target detection model. After the training is completed, the third target detection model can be used for positioning detection of device labels and identification codes. Specifically, the corrected image of the cabinet can be input into the third target detection model for detection processing to obtain the bounding box of the label and the bounding box of the identification code of each device in the cabinet.

[0094] Furthermore, after obtaining the detection result of the third target detection model for the rectified image, filtering processing can be performed. Specifically, the bounding boxes of the device labels and identification codes in the detection results have corresponding confidence scores, which are used to characterize the probability that the bounding boxes contain labels or identification codes. Filtering the detection results can also filter out detection results with confidences lower than a confidence threshold (for example, 0.8), so as to improve the accuracy of the labels and identification codes corresponding to the devices in the cabinet.

[0095] The identification code can be inside the label, that is, the label's bounding box contains the identification code's bounding box; or it can be outside the label, that is, the label and the identification code are attached to the device separately. When the identification code is inside the label, after obtaining the detection result of the third object detection model, the detection result can be further filtered to remove the detection result of the identification code outside the label boundary.

[0096] Step S420: extract the text information in the label corresponding to the device, and decode the identification code of the device to obtain the device number of the device.

[0097] Specifically, the extraction of text information in the label can be achieved through text recognition technology, such as optical character recognition (OCR) to extract text information in the label, such as using a lightweight text recognition network SVTR_LCNet, which combines the convolution-based algorithm PP-LCNet (responsible for text area detection and local feature extraction) with the Transformer-based algorithm SVTR (for global recognition and conversion of text). The network is suitable for running on mobile devices. At the same time, the text that exceeds the label boundary is removed.

[0098] The decoding process of the device identification code can be realized through the decoding library. First, the identification code is aligned, the data is extracted and restored to the original binary code, and it is converted into corresponding characters to complete the decoding and obtain the device number of the device. Among them, when the identification code of a device contains both a QR code and a barcode, each identification code can be decoded separately. Under normal circumstances, the device numbers identified by the two identification codes should be the same. In practical applications, the consistency of the device numbers proposed by the two identification codes can also be verified.

[0099] Step S430, verifying the consistency among the device number, text information, label information and device information, as well as the relevant information of the device recorded in the resource system of the computer room, and generating a prompt message if there is any inconsistency.

[0100] Specifically, consistency verification may include the following: (1) Comparing the label information corresponding to the device with the device information to determine whether the device label is missing. For the same device, if the device information is detected by the first target detection model and the second target detection model, but the label information is not detected by the third target detection model, it indicates that the label is missing. For example, if there is no bounding box containing the label within the bounding box of a certain device, it can be determined that the label of the device is missing, and a prompt message can be generated to prompt the maintenance personnel to add the label in time. (2) When the device information includes the device model, compare the device information corresponding to the device with the text information in the corresponding label to determine the matching between the label and the device. Specifically, because the text information extracted from the label will contain the device model, for the same device, the text information in the identified label and the device model in the device information of the device detected by the first target detection model and the second target detection model are compared to determine the consistency between the two. If the device model in the device information does not match the device model in the text information of the label, a prompt message can be generated to prompt the maintenance personnel to replace the label in time. (3) According to the equipment number of the equipment, query the relevant information of the equipment recorded in the resource system of the computer room. If the relevant information recorded in the resource system is inconsistent with the detected equipment information, label information and text information in the label, generate a prompt message to prompt the maintenance personnel to update the resource system in time.

[0101] In this embodiment, by comparing the consistency between the label information corresponding to the device and the device information, the missing label of the device can be determined, thereby prompting the maintenance personnel to add the label in time; by comparing the matching between the device information corresponding to the device and the text information in the corresponding label, the problem of wrong label affixation can be detected, thereby prompting the maintenance personnel to make corrections in time; by querying the relevant information of the device recorded in the resource system through the detected device number of a certain device, and comparing the consistency between the information recorded in the resource system and the information detected by each target detection model, the problem of untimely update of the resource system can be detected, thereby prompting the maintenance personnel to update in time. This method compares and corrects the identification result of the device model with the decoding result of the identification code and the text recognition result of the label, thereby realizing automatic update inspection of the label and resource system.

[0102] In an exemplary embodiment, the method further includes: establishing a three-dimensional model of the cabinet based on the corrected image; labeling the corresponding equipment in the three-dimensional model based on the equipment information corresponding to the equipment in the cabinet to obtain a labeled three-dimensional model of the cabinet; and generating a digital three-dimensional computer room corresponding to the computer room based on the labeled three-dimensional models of all cabinets in the computer room.

[0103] Specifically, the rectified image of each cabinet in the computer room can be used to construct a corresponding three-dimensional model of each cabinet. Based on the three-dimensional model of each cabinet, a digital three-dimensional computer room can be generated, so that the computer room maintenance personnel can more intuitively understand the internal structure and equipment layout of the computer room, thereby improving the efficiency and accuracy of maintenance work. For example, the DUST3R model (Dense Unconstrained Stereo 3D Reconstruction) for three-dimensional reconstruction can be used to construct the three-dimensional model of each cabinet based on the rectified image of each cabinet. This method can perform dense three-dimensional reconstruction on non-calibrated images without pose information, and does not require any camera calibration or prior information of viewpoint pose, that is, a single image or a pair of images can be used to perform 3D reconstruction in a very short time. Therefore, for each cabinet, the construction of the three-dimensional model of the cabinet can be completed based only on the rectified image of the cabinet. After completing the construction of the three-dimensional model of the cabinet, the three-dimensional model of each cabinet can be labeled with information such as the equipment boundary box and equipment model of each cabinet detected by the target detection model, thereby obtaining the labeled three-dimensional model corresponding to each cabinet in the computer room. Based on the labeled three-dimensional models of all cabinets in the computer room, the digital three-dimensional computer room corresponding to the computer room is generated, realizing accurate mapping of the real world to the digital computer room three-dimensional model.

[0104] This embodiment generates a 3D model based on an image, automatically generates a cabinet model by taking a photo with one click, and uses the detected device model and 3D modeling connection annotation to ensure that the device position and model information in the 3D model are consistent with the actual cabinet.

[0105] In one embodiment, in order to facilitate those skilled in the art to understand the embodiments of the present application, the following will be described with reference to specific examples of the accompanying drawings. Figure 5 , shows a schematic diagram of the overall process of a computer room equipment maintenance method, which integrates image processing technology and computer vision algorithms to achieve comprehensive perception of the computer room environment and assist in automated operation and maintenance. It mainly includes the following contents:

[0106] First, in order to solve the image deformation problem of cabinets photographed by handheld devices, automatic image correction is implemented based on semantic segmentation and perspective transformation to improve the accuracy of subsequent image processing. Secondly, for the corrected images, the target detection algorithm is used to automatically detect the equipment in the computer room. On the one hand, the equipment model and location information are obtained to supplement unknown model equipment, and the equipment is recorded by comparing the upstream and downstream detection results in time. On the other hand, the location information of the equipment label and equipment identification code is obtained through target detection, and the equipment number in the identification code is obtained through decoding technology, and the equipment model in the label is obtained through text recognition. By comparing the detection results of the two tests, the audit of the label and the audit of the resource system are realized. Finally, through the three-dimensional model generation technology, the 3D model of each cabinet is generated based on the corrected image of each cabinet, and the digital three-dimensional computer room of the computer room is generated based on the 3D model of each cabinet, realizing the three-dimensional visualization of the computer room.

[0107] Based on image technology, this method realizes comprehensive perception and intelligent management of the computer room environment, including the comprehensive acquisition and processing of multi-dimensional information such as equipment model, location, QR code, label text, etc., to achieve accurate positioning and management of computer room equipment. Through multi-dimensional perception, the accuracy of equipment detection is ensured, and finally the management of equipment on and off the shelf, multi-party comparison of equipment model and identification code decoding information and label text, label error correction and resource system error correction are realized, and the precise mapping of equipment and its virtual space three-dimensional model is realized by combining 3D modeling. The integrated application of this computer room multi-dimensional perception and intelligent management system provides a full range of intelligent solutions for computer room management, greatly simplifying the process of equipment management and maintenance. In addition, this method has a high degree of automation. It only needs to take pictures with a handheld smart terminal to realize the management of equipment on and off the shelf, update and check the identification code and resource system, and three-dimensional visualization of the cabinet in one step, which significantly reduces the workload of manual work and improves the efficiency and accuracy of computer room maintenance.

[0108] It should be understood that, although the various steps in the flowcharts involved in the above-mentioned embodiments are displayed in sequence according to the indication of the arrows, these steps are not necessarily executed in sequence according to the order indicated by the arrows. Unless there is a clear explanation in this article, the execution of these steps is not strictly limited in order, and these steps can be executed in other orders. Moreover, at least a part of the steps in the flowcharts involved in the above-mentioned embodiments can include multiple steps or multiple stages, and these steps or stages are not necessarily executed at the same time, but can be executed at different times, and the execution order of these steps or stages is not necessarily carried out in sequence, but can be executed in turn or alternately with other steps or at least a part of the steps or stages in other steps.

[0109] Based on the same inventive concept, the embodiment of the present application also provides a computer room equipment maintenance device for implementing the computer room equipment maintenance method involved above. The implementation scheme for solving the problem provided by the device is similar to the implementation scheme recorded in the above method, so the specific limitations in one or more computer room equipment maintenance device embodiments provided below can refer to the limitations on the computer room equipment maintenance method above, and will not be repeated here.

[0110] In one embodiment, Figure 6 As shown, a device for maintaining equipment in a computer room is provided, comprising:

[0111] An acquisition module 610 is used to acquire an original image of a cabinet in the computer room at a current moment;

[0112] The correction module 620 is used to perform correction processing on the original image to obtain a corrected image;

[0113] The detection module 630 is used to perform target detection processing on the corrected image to obtain device information corresponding to the device in the cabinet; the device information at least includes a bounding box of the device;

[0114] The comparison module 640 is used to compare the device information at the current moment with the device information of the cabinet at the previous moment, obtain the change information of the device in the cabinet, and record the change information.

[0115] In one embodiment, the correction module 620 is specifically used to perform semantic segmentation on the original image to obtain the cabinet area, perform edge detection on the cabinet area to obtain the outline of the cabinet; extract the correction positioning points from the outline of the cabinet, and determine the perspective transformation matrix based on the correction positioning points; according to the perspective transformation matrix, correct the original image to obtain a corrected image.

[0116] In one of the embodiments, the detection module 630 is specifically used to perform target detection processing on the corrected image through a first target detection model to obtain first device information corresponding to the device in the cabinet; the first device information includes the bounding box and device model of the device; perform target detection processing on the corrected image through a second target detection model to obtain second device information corresponding to the device in the cabinet; the second device information includes the bounding box of the device; and deduplicate the first device information and the second device information to obtain device information.

[0117] In one of the embodiments, the detection module 630 is also used to delete the device information of any device in the second device information when the ratio of the overlapping area between the boundary box of any device in the second device information and the boundary box of any device in the first device information exceeds a predetermined threshold, so as to obtain the deduplicated second device information; and combine the deduplicated second device information and the first device information to form the device information corresponding to the device in the cabinet.

[0118] In one embodiment, the device also includes an information consistency verification module, which is used to perform target detection processing on the corrected image through a third target detection model to obtain label information and identification code information corresponding to the equipment in the cabinet; the label information includes a bounding box of the label, and the identification code information includes a bounding box of the identification code; extract the text information in the label corresponding to the equipment, and decode the identification code of the equipment to obtain the equipment number of the equipment; verify the consistency between the equipment number, text information, label information and equipment information, as well as the relevant information of the equipment recorded in the resource system of the computer room, and generate a prompt message if there is any inconsistency.

[0119] In one embodiment, the device also includes a model building module, which is used to establish a three-dimensional model of the cabinet based on the corrected image; mark the corresponding equipment in the three-dimensional model according to the equipment information corresponding to the equipment in the cabinet to obtain the marked three-dimensional model of the cabinet; and generate a digital three-dimensional computer room corresponding to the computer room based on the marked three-dimensional models of all cabinets in the computer room.

[0120] Each module in the above-mentioned computer room equipment maintenance device can be implemented in whole or in part by software, hardware or a combination thereof. Each module can be stored in a memory in a computer device in the form of software, so that the processor can call and execute the operations corresponding to each module.

[0121] In one embodiment, a computer device is provided. The computer device may be a terminal, and its internal structure diagram may be as follows: Figure 7 As shown. The computer device includes a processor, a memory, a communication interface, a display screen and an input device connected through a system bus. Among them, the processor of the computer device is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system and a computer program. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The communication interface of the computer device is used to communicate with an external terminal in a wired or wireless manner, and the wireless manner can be achieved through WIFI, a mobile cellular network, NFC (near field communication) or other technologies. When the computer program is executed by the processor, a method for maintaining equipment in a computer room is implemented. The display screen of the computer device can be a liquid crystal display screen or an electronic ink display screen, and the input device of the computer device can be a touch layer covered on the display screen, or a key, trackball or touchpad set on the computer device shell, or an external keyboard, touchpad or mouse, etc.

[0122] Those skilled in the art will understand that Figure 7The structure shown in the figure is only a block diagram of a part of the structure related to the solution of the present application, and does not constitute a limitation on the computer device to which the solution of the present application is applied. The specific computer device may include more or fewer components than those shown in the figure, or combine certain components, or have a different arrangement of components.

[0123] In one embodiment, a computer device is further provided, including a memory and a processor, wherein a computer program is stored in the memory, and the processor implements the steps in the above method embodiments when executing the computer program.

[0124] In one embodiment, a computer-readable storage medium is provided, on which a computer program is stored. When the computer program is executed by a processor, the steps in the above-mentioned method embodiments are implemented.

[0125] In one embodiment, a computer program product is provided, including a computer program, which implements the steps in the above method embodiments when executed by a processor.

[0126] It should be noted that the user information (including but not limited to user device information, user personal information, etc.) and data (including but not limited to data used for analysis, stored data, displayed data, etc.) involved in this application are all information and data authorized by the user or fully authorized by all parties, and the collection, use and processing of relevant data must comply with relevant laws, regulations and standards of relevant countries and regions.

[0127] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to the memory, database or other medium used in the embodiments provided in this application can include at least one of non-volatile and volatile memory. Non-volatile memory can include read-only memory (ROM), magnetic tape, floppy disk, flash memory, optical memory, high-density embedded non-volatile memory, resistive random access memory (ReRAM), magnetoresistive random access memory (MRAM), ferroelectric random access memory (FRAM), phase change memory (PCM), graphene memory, etc. Volatile memory can include random access memory (RAM) or external cache memory, etc. As an illustration and not limitation, RAM can be in various forms, such as static random access memory (SRAM) or dynamic random access memory (DRAM). The database involved in each embodiment provided in this application may include at least one of a relational database and a non-relational database. Non-relational databases may include distributed databases based on blockchains, etc., but are not limited to this. The processor involved in each embodiment provided in this application may be a general-purpose processor, a central processing unit, a graphics processor, a data processing logic based on quantum computing, etc., but are not limited to this.

[0128] The technical features of the above embodiments may be combined arbitrarily. To make the description concise, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0129] The above-described embodiments only express several implementation methods of the present application, and the descriptions thereof are relatively specific and detailed, but they cannot be understood as limiting the scope of the present application. It should be pointed out that, for a person of ordinary skill in the art, several variations and improvements can be made without departing from the concept of the present application, and these all belong to the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the attached claims.

Claims

1. A method for maintaining equipment in a computer room, characterized in that: The method comprises: Obtain the original image of the cabinet in the computer room at the current moment; Performing correction processing on the original image to obtain a corrected image; Performing target detection processing on the rectified image to obtain device information corresponding to the device in the cabinet; the device information at least includes a boundary box of the device; The device information at the current moment is compared with the device information of the cabinet at the previous moment to obtain change information of the devices in the cabinet, and the change information is recorded.

2. The method according to claim 1, characterized in that The correcting the original image to obtain a corrected image comprises: Performing semantic segmentation on the original image to obtain a cabinet area, and performing edge detection on the cabinet area to obtain a contour of the cabinet; Extracting correction positioning points from the outline of the cabinet, and determining a perspective transformation matrix based on the correction positioning points; The original image is corrected according to the perspective transformation matrix to obtain the corrected image.

3. The method according to claim 1, characterized in that The performing target detection processing on the corrected image to obtain device information corresponding to the device in the cabinet includes: Performing target detection processing on the rectified image through a first target detection model to obtain first device information corresponding to the device in the cabinet; the first device information includes a bounding box and a device model of the device; Performing target detection processing on the rectified image through a second target detection model to obtain second device information corresponding to the device in the cabinet; the second device information includes a bounding box of the device; Deduplication processing is performed on the first device information and the second device information to obtain the device information.

4. The method according to claim 3, characterized in that The performing deduplication processing on the first device information and the second device information to obtain the device information includes: When the overlapping area ratio of the bounding box of any device in the second device information and the bounding box of any device in the first device information exceeds a predetermined threshold, deleting the device information of the any device in the second device information to obtain the second device information after deduplication; The deduplicated second device information and the first device information are combined to form device information corresponding to the device in the cabinet.

5. The method according to claim 1, characterized in that After the correction processing is performed on the original image to obtain the corrected image, the method further includes: By using a third target detection model, target detection processing is performed on the corrected image to obtain label information and identification code information corresponding to the equipment in the cabinet; the label information includes a bounding box of the label, and the identification code information includes a bounding box of the identification code; Extracting text information from a label corresponding to the device, and decoding an identification code of the device to obtain a device number of the device; Verify the consistency among the device number, the text information, the label information and the device information, as well as the relevant information of the device recorded in the resource system of the computer room, and generate a prompt message if there is any inconsistency.

6. The method according to any one of claims 1 to 5, characterized in that: The method further comprises: Establishing a three-dimensional model of the cabinet according to the corrected image; According to the device information corresponding to the device in the cabinet, the corresponding device in the three-dimensional model is marked to obtain the marked three-dimensional model of the cabinet; A digital three-dimensional computer room corresponding to the computer room is generated according to the annotated three-dimensional models of all the cabinets in the computer room.

7. A computer room equipment maintenance device, characterized in that: The device comprises: An acquisition module is used to acquire the original image of the cabinet in the computer room at the current moment; A correction module, used for performing correction processing on the original image to obtain a corrected image; A detection module, configured to perform target detection processing on the rectified image to obtain device information corresponding to the device in the cabinet; the device information at least includes a boundary box of the device; The comparison module is used to compare the device information at the current moment with the device information of the cabinet at the previous moment, obtain the change information of the device in the cabinet, and record the change information.

8. A computer device comprising a memory and a processor, wherein the memory stores a computer program, wherein: When the processor executes the computer program, the steps of the computer room equipment maintenance method described in any one of claims 1 to 6 are implemented.

9. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the computer room equipment maintenance method described in any one of claims 1 to 6 are implemented.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the steps of the computer room equipment maintenance method described in any one of claims 1 to 6 are implemented.

Images