Crane sole timber supporting leg detection method, device and equipment and medium

Abstract

The invention discloses a crane sole timber landing leg detection method, device and equipment and a medium, which are applied to the field of construction monitoring, and comprise the following steps: acquiring image data of a construction area, performing landing leg area identification and coordinate positioning processing on the image data, outputting positioning image data marked with landing leg position information, and cutting a landing leg area image. The method comprises the following steps: acquiring local image data of a support leg, performing sole timber existence state analysis processing on the local image data of the support leg based on texture-color joint features, outputting a sole timber state judgment result, and if the sole timber state judgment result is abnormal, starting early warning information and uploading a detection record. Through the automatic detection process, the time and the labor cost of manual inspection are reduced, meanwhile, missing detection and misjudgment caused by human factors are avoided, the detection efficiency is improved, stable operation in a complex construction environment can be achieved, the construction process is helped to be optimized, and the construction safety management level is improved.

Description

Technical Field

[0001] The present invention relates to the technical field of construction monitoring, and in particular to a method, device, equipment and medium for detecting crane skid legs. Background Art

[0002] With the rapid development of power grid construction, cranes are used more and more frequently in power grid construction, and their operational safety has attracted much attention.

[0003] Correct placement of these critical components, used to disperse outrigger pressure and prevent ground collapse, is crucial for crane stability and construction safety. Existing inspections of crane outrigger pads rely primarily on manual inspections, which are inefficient and prone to omissions, making real-time monitoring difficult, especially in complex construction environments. Furthermore, traditional computer vision methods have numerous limitations in power grid construction scenarios. For example, it's difficult to accurately locate small, easily obscured outrigger pads, resulting in a high rate of missed detections, which severely impacts construction safety and efficiency.

[0004] Therefore, how to improve the detection accuracy of crane leg pads has become a technical problem that needs to be solved urgently by those skilled in the art. Summary of the Invention

[0005] The present invention provides a crane leg pad detection method, device, equipment and medium for detecting the state of the leg pads of a working crane.

[0006] In order to solve the above technical problems, an embodiment of the present invention provides a method for detecting crane skid legs, comprising: The video stream of the construction area is acquired and preprocessed to obtain several standardized frame image data.

[0007] The leg region recognition and coordinate positioning processing are performed on each of the standardized frame image data, and positioning image data marked with the leg position information is output.

[0008] The leg area image is cropped based on the leg position information in the positioning image data to obtain leg local image data.

[0009] The local image data of the outrigger is subjected to a wood block presence status analysis based on the texture-color joint feature, and a wood block status determination result is output.

[0010] If the result of the said sleeper status is determined to be "abnormal", the early warning information is activated and the detection record is uploaded.

[0011] Furthermore, the construction area video stream is obtained and pre-processed to obtain a number of standardized frame image data, including: The collected video stream of the construction area is subjected to frame rate adjustment, resolution conversion, color correction and noise filtering to obtain a preprocessed video stream.

[0012] Sampling is performed from the pre-processed video stream according to a preset number of interval frames to obtain a plurality of sampling frame image data.

[0013] Each of the sampled frame image data is subjected to standardization processing to obtain standardized frame image data.

[0014] Furthermore, the performing of leg region recognition and coordinate positioning processing on each of the standardized frame image data and outputting positioning image data marked with leg position information includes: Each of the standardized frame image data is input into the pre-trained KSROI-Net model to perform leg area recognition to obtain the bounding box information of the leg area.

[0015] The key point coordinates of the leg are calculated based on the bounding box information of the leg area.

[0016] The bounding box information and key point coordinates of the leg area are marked on the corresponding standardized frame image data to obtain positioning image data.

[0017] Furthermore, the step of cropping the leg region image based on the leg position information in the positioning image data to obtain the leg local image data includes: The leg region image is cropped from the corresponding standardized frame image data according to the bounding box information in the positioning image data.

[0018] The ESRGAN algorithm is used to perform super-resolution reconstruction on the leg region image to obtain a first leg region image.

[0019] Grayscale processing is performed on the first leg area image, and its local binary pattern features are calculated to extract leg texture information.

[0020] The leg texture information is fused with the grayscale image of the first leg region image to obtain a second leg region image.

[0021] Pixel value normalization processing is performed on each of the second leg area images to obtain standardized leg local image data.

[0022] Furthermore, the processing of analyzing the presence of a wooden block on the local image data of the outrigger based on the texture-color joint feature and outputting a result of the wooden block status determination includes: The local image data of the leg is input into the pre-trained MobileNetV3 network to extract the color features and texture features of the leg area.

[0023] The color features and the texture features at different scales are weightedly fused through a multi-scale feature fusion module to generate a joint feature representation.

[0024] The joint feature representation is input into a binary classifier for analysis to obtain a result of the state determination of the skid.

[0025] Furthermore, if the result of the sleeper status determination is "abnormal", the early warning information is activated and the detection record is uploaded, including: When the result of the skid status determination is "abnormal", a multi-level early warning mechanism is triggered.

[0026] The multi-level early warning mechanism includes activating sound and light alarm devices, recording violation information and remotely notifying security management personnel.

[0027] Furthermore, after the early warning information is started and the detection record is uploaded, the following is also included: The inspection records are analyzed through a cloud management platform to obtain the frequency and pattern of missing skids in each construction area.

[0028] Feedback the analysis results to the management end of the on-site detection equipment.

[0029] Another embodiment of the present invention provides a crane skid leg detection device, comprising: The data acquisition module is used to acquire the video stream of the construction area and perform preprocessing to obtain a number of standardized frame image data.

[0030] The data positioning module is used to perform leg area recognition and coordinate positioning processing on each of the standardized frame image data, and output positioning image data marked with leg position information.

[0031] The image cropping module is used to crop the leg area image based on the leg position information in the positioning image data to obtain leg local image data.

[0032] The result determination module is used to perform a block presence status analysis on the local image data of the support leg based on the texture-color joint feature, and output a block status determination result.

[0033] The information warning module is used to start the warning information and upload the detection record if the result of the said skid status is judged to be "abnormal".

[0034] Another embodiment of the present invention provides a computer device, including a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein the processor implements the crane pad leg detection method as described above when executing the computer program.

[0035] Yet another embodiment of the present invention provides a computer-readable storage medium storing a computer program, wherein when the device where the computer-readable storage medium is located executes the computer program, the crane pad leg detection method as described above is implemented.

[0036] Compared with the prior art, the embodiments of the present invention have the following advantages: The automated inspection process reduces the time and labor costs of manual inspections, while avoiding missed inspections and misjudgments caused by human factors, improving inspection efficiency. The system can operate stably in complex construction environments, accurately identifying even small or partially obscured outriggers, ensuring robust inspection results. The introduction of a multi-level early warning mechanism promptly notifies relevant personnel when missing outriggers are detected, effectively preventing potential safety incidents and ensuring construction safety. At the same time, big data analysis of inspection records provides data support for construction safety management, helping to optimize construction processes and improve construction safety management. BRIEF DESCRIPTION OF THE DRAWINGS

[0037] Figure 1 Flowchart of the steps of a method for detecting crane skid legs in one embodiment of the present invention; Figure 2 A schematic diagram of system deployment in one embodiment of the present invention; Figure 3 This is a KSROI-Net model architecture diagram of a crane skid leg detection method in one embodiment of the present invention; Figure 4 This is a structural diagram of the DefMixer module of a crane skid leg detection method in one embodiment of the present invention; Figure 5 This is a structural block diagram of a crane skid leg detection device in one embodiment of the present invention; Figure 6 A structural diagram of a computer device provided in an embodiment of the present invention. DETAILED DESCRIPTION

[0038] The following will be combined with the accompanying drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the embodiments described are only part of the embodiments of the present invention, not all of the embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without making creative efforts shall fall within the scope of protection of the present invention.

[0039] In the description of this application, the terms "first," "second," "third," etc. are used for descriptive purposes only and should not be understood to indicate or imply relative importance or implicitly specify the number of the technical features indicated. Therefore, a feature specified as "first," "second," "third," etc. may explicitly or implicitly include one or more of the features. In the description of this application, unless otherwise specified, "plurality" means two or more.

[0040] In the description of this application, it should be noted that, unless otherwise expressly specified and limited, the terms "installed", "connected" and "connected" should be understood in a broad sense. For example, it can be a fixed connection, a detachable connection, or an integral connection; it can be a mechanical connection or an electrical connection; it can be a direct connection, or an indirect connection through an intermediate medium, or it can be a communication between the two components. The terms "vertical", "horizontal", "left", "right", "up", "down" and similar expressions used herein are for illustrative purposes only, and do not indicate or imply that the device or component referred to must have a specific orientation, be constructed and operated in a specific orientation, and therefore cannot be understood as a limitation on the present invention. The term "and / or" used herein includes any and all combinations of one or more related listed items. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood according to specific circumstances.

[0041] In the description of this application, it should be noted that, unless otherwise defined, all technical and scientific terms used in this application have the same meanings as those commonly understood by those skilled in the art. The terms used in the specification of the present invention are only for the purpose of describing specific embodiments and are not intended to limit the present invention. For those of ordinary skill in the art, the specific meanings of the above terms in this application can be understood by those skilled in the art in specific circumstances.

[0042] An embodiment of the present invention provides a method for detecting crane skid legs. For details, see Figure 1 , Figure 1 The flowchart of the method for detecting crane skid legs in one embodiment of the present invention includes steps S11 to S15: S11. Obtain a video stream of the construction area and perform preprocessing to obtain a number of standardized frame image data.

[0043] like Figure 2As shown in the figure, the system uses cameras deployed to capture video streams from the construction process. During the crane outrigger inspection process, video streams from the construction area are an important data source. Due to the complex construction site environment, video streams may have issues such as inconsistent frame rates, resolution mismatches, color deviations, and noise interference. To ensure the accuracy and efficiency of subsequent processing, the captured video streams must first be preprocessed.

[0044] Specifically, frame rate adjustment can ensure that the video stream is played at a moderate speed, avoiding image information loss or redundancy caused by too high or too low a frame rate. Resolution conversion can unify videos of different resolutions into a standard resolution, facilitating subsequent image processing and analysis. Color correction can correct color deviations caused by lighting conditions or equipment differences, making the image color more realistic and consistent. Noise filtering can remove random noise in the video and improve image clarity and quality.

[0045] After obtaining the preprocessed video stream, in order to further reduce the amount of data and improve processing efficiency, it is necessary to extract representative frame images from the video stream. The setting of the sampling interval needs to comprehensively consider the dynamic changes in the construction area and the requirements of the detection task. If the sampling interval is too short, it may lead to data redundancy and increase unnecessary computational burden; if the sampling interval is too long, important information may be missed. Therefore, this embodiment presets a reasonable sampling interval frame number based on the specific conditions of the construction area and the detection target, and extracts frame images from the preprocessed video stream according to this interval.

[0046] Although the sampled frame image data has been preprocessed, there may still be differences in size, brightness, contrast, etc. In order to ensure the consistency and accuracy of subsequent analysis, these sampled frame image data need to be standardized.

[0047] Normalization involves resizing the image to a fixed size for model input and adjusting the image's brightness and contrast to meet a preset standard range. Furthermore, images can be normalized to unify pixel values to [0, 1] or other standard ranges to facilitate subsequent feature extraction and analysis. This standardization step ensures greater consistency and comparability of the resulting standardized frame image data.

[0048] S12, performing leg region recognition and coordinate positioning processing on each of the standardized frame image data, and outputting positioning image data marked with leg position information.

[0049] In order to achieve efficient and accurate leg area recognition, this embodiment proposes the KSROI-Net (Keypoint-Status-ROI Network) model. Through the multi-head plus ROI (region of interest) cropping design, while ensuring computational efficiency, a set of backbone networks can be shared to complete global coarse positioning and status classification, and based on the key points output by the key point prediction branch, fine ROIs are dynamically cropped to provide high-precision local detection input for the next stage model.

[0050] The model structure is as follows Figure 3 As shown in Figure 2, the model uses the DefMixer module (Deformable Mixer, deformable mixed convolution module) to enhance the representation of irregular targets such as crane legs and skids. The structure of DefMixer is as follows: Figure 4 As shown, Backbone ( Figure 3 The backbone network (enclosed in the dashed box) is the KSROI-Net backbone feature extraction network. It integrates basic feature extraction modules such as CBS (convolutional blocks) and MPConv (multi-path convolution) to extract low-level features such as edges, corners, and textures. It also includes a concat (feature concatenation) operation to combine and aggregate local features. The CBR (Convolution, BatchNormalization ReLU) module aligns features to a uniform dimension and feeds them into the DefMixer module for extracting leg features.

[0051] Due to the special shape of the crane legs (the legs are in an inverted "L" shape), the traditional square n*n convolution kernel design is not good at capturing the characteristics of the crane legs. Figure 3 The CBR→DefMixer transformation (within the dashed box) obtains three feature maps of different sizes. This transform uses channel adaptation and spatially deformable feature remixing to allow the network to more flexibly capture the behavior of slender, irregular objects like legs at different scales and angles. CBR is used for feature alignment, unifying the number of channels to C through downsampling.

[0052] For possible mixed inspection objects with similar shapes to outriggers, such as cable channels and hydraulic rods, ROI regional morphology estimation can be used to further determine whether they are crane pads based on texture and aspect ratio (generally maintained at 4:3).

[0053] The DefMixer module structure of this embodiment is as follows Figure 4As shown in the figure, after receiving the feature map from the CBR layer, DefMixer first flattens it into a C×N matrix according to the channel and spatial dimensions, and maps it to the new feature space through a fully connected Linear layer; then, a Channel aware Block is applied to the obtained features, and the channel importance weights are extracted through global average pooling and adaptive scaling of each channel is performed to highlight the key features; on this basis, the features are sent to the first round of Dynamic Tanh (DyT, dynamic hyperbolic tangent activation function) module, which replaces the traditional Layer Normalization with dynamic hyperbolic tangent and adjusts the activation curve with learnable parameters to achieve information aggregation across spatial positions; then, Spatial aware The Deformable module (spatial-aware deformable convolution module) performs a deformable convolution operation on the features according to the self-learned offset, as shown in formulas (1) and (2), so that the structure can capture irregular targets more flexibly. The features after this deformable convolution enter the second round of DyT again to complete deeper global dependency modeling. Finally, the output of the second round of DyT is added to the residual of the initial mapping feature, and the addition result is reshaped (feature transformation) back to a C×H×W feature map for subsequent network use.

[0054] (1) (2) Among them, X represents the input features, is a standard convolutional layer, They represent the predicted spatial position offset and the predicted modulation factor respectively. Indicates that the output feature map is The value at , k represents the kth sampling point or convolution kernel position index, represents the kth weight of the convolution kernel, Indicates the output location, Indicates the preset offset of the convolution kernel, express Dynamically learned spatial offsets, express Modulation factors for dynamic learning.

[0055] The feature maps output by DefMixer are spliced in the channel dimension, and the multi-head attention mechanism is used for feature interaction. CBR, DefMixer and Feature Interaction Block together form the Neck of the model. The KeyPointHead branch and the WorkStatusHead branch use the feature maps obtained after feature interaction. Predict the key points of the crane legs and the working status of the crane.

[0056] (3) in, It represents the feature map after three channels are spliced by DefMixer. Represents the multi-head attention mechanism, Q, K, V represent the Query, Key, and Value vectors in the attention respectively, Represents the fused features after multi-head attention.

[0057] In the actual application of crane skid leg detection, in order to ensure the efficiency and accuracy of detection, this embodiment first inputs the preprocessed and standardized frame image data into the pre-trained KSROI-Net model. With its unique multi-task learning architecture and key point-guided ROI cropping mechanism, the model can accurately identify the leg area and output the corresponding bounding box information.

[0058] After obtaining the bounding box information of the leg area, the coordinates of the key points of the leg, such as the center point and connection point of the leg, are further calculated based on this information.

[0059] Finally, to facilitate subsequent visual analysis and further processing, the system marks the bounding box information and key point coordinates of the leg area on the corresponding standardized frame image data, generates positioning image data, and intuitively displays the position of the leg and the specific position of the key points, making the detection results clear at a glance.

[0060] S13. Crop the leg area image based on the leg position information in the positioning image data to obtain leg local image data.

[0061] During crane skid leg detection, the bounding box information in the localization image data provides precise guidance for cropping the leg region. This bounding box information clearly defines the leg's position and extent within the image, enabling the system to accurately crop the leg region from the standardized frame image data. This precise cropping removes background information unrelated to the leg, reducing interference and providing clear image input for subsequent super-resolution reconstruction and texture feature extraction.

[0062] The cropped outrigger area image may have insufficient details due to the resolution limitation or compression loss of the original video stream. In order to enhance the details and clarity of the image, this embodiment uses the ESRGAN (Enhanced Super-Resolution Generative Adversarial Networks) algorithm to perform super-resolution reconstruction on the outrigger area image. The ESRGAN algorithm, through the GAN (Generative Adversarial Network) architecture, can effectively improve the resolution of the image while maintaining the texture and details of the image. The first outrigger area image obtained after super-resolution reconstruction has higher resolution and richer details. Specifically, the super-resolution reconstruction process of this embodiment is to magnify the ROI area by 4 times to obtain the first outrigger area image. The selection of this magnification is based on the actual demand analysis of the crane outrigger pad detection task. In the power grid construction scene, the size of the outrigger pad is relatively small and is easily blocked in a complex environment or difficult to identify due to insufficient light. After 4 times magnification, the resolution of the outrigger area image can be significantly improved, making the texture and shape characteristics of the pad more clearly visible.

[0063] After obtaining a high-resolution image of the first leg region, this embodiment further grayscales it. Grayscaling converts a color image into a grayscale image, simplifying the image representation while preserving the image's primary structure and texture information. Subsequently, this embodiment calculates the local binary pattern (LBP) features of the grayscale image. LBP features are an effective texture descriptor that can capture local leg texture information.

[0064] To further enhance the features of the leg region image, this embodiment fuses the extracted leg texture information with the grayscale image of the first leg region image. This combines the structural and texture information of the grayscale image to generate a second leg region image. This fused image not only retains the structural features of the original image but also enhances texture detail, making the features of the leg region more distinct and prominent.

[0065] After image fusion, the system normalizes the pixel values of the second leg region image. Pixel normalization adjusts the image's pixel value range to a standard range (such as [0, 1] or [-1, 1]) to eliminate differences in pixel value ranges between different images, ensuring greater consistency and comparability of the resulting standardized leg region image data.

[0066] S14. Performing a block presence status analysis on the local image data of the outrigger based on the texture-color joint feature, and outputting a block status determination result.

[0067] In order to accurately determine the status of the crane leg pads, this embodiment uses deep learning technology to extract and analyze the features of the local image data of the legs. Specifically, the preprocessed local image data of the legs is first input into the pre-trained MobileNetV3 network. MobileNetV3 is a lightweight and efficient convolutional neural network suitable for running on edge devices. In this embodiment, the MobileNetV3 network can efficiently extract color features and texture features from the local images of the legs. Color features help identify the material and surface condition of the pads, while texture features can reveal the degree of wear, cracks or other potential problems of the pads.

[0068] Because crane leg shims may be affected by a variety of factors in actual construction scenarios, such as lighting changes, occlusion, and different viewing angles, features at a single scale often fail to fully reflect the true condition of the shims. After extracting color and texture features, a multi-scale feature fusion module is used to perform a weighted fusion of color and texture features at different scales to generate a joint feature representation. Multi-scale feature fusion can integrate feature information at different scales to more accurately describe the condition of the shims. For example, at larger scales, texture features may better reflect the overall structure of the shims, while at smaller scales, color features may better reveal subtle changes on the shims' surface.

[0069] Finally, the generated joint feature representation is input into a binary classifier for analysis. The binary classifier's task is to determine whether the state of the sleepers is normal based on the joint feature representation. In this embodiment, the binary classifier, by learning from a large amount of labeled data, can accurately distinguish between the normal and abnormal states of the sleepers. For example, if the sleepers are worn, cracked, or otherwise damaged, the binary classifier will determine that they are in an abnormal state; conversely, if the sleepers are in good condition, they will be determined to be in a normal state.

[0070] Specifically, when analyzing through a binary classifier, if the detection result is "no pads", the morphological verification is triggered to verify the presence or absence of the leg pads. If the verification result is no pads, it is determined to be truly missing. If the verification result is that there are pads, the posture of the pads is identified. If an abnormality is found in the posture, it is determined to be a displacement of the pad position.

[0071] S15. If the result of the sleeper status determination is "abnormal", an early warning message is activated and a detection record is uploaded.

[0072] The aforementioned actual missing and positional displacement are both abnormal conditions. When the result of the block status determination is "abnormal," it indicates that the crane leg blocks may pose a safety hazard, such as being missing, damaged, or improperly placed. To promptly alert on-site workers and enable them to take appropriate measures, this embodiment triggers a multi-level early warning mechanism to ensure construction safety.

[0073] The multi-level warning mechanism of this embodiment includes at least the following steps: Activate the sound and light alarm device: Once the system detects an abnormal state of the skid, it will immediately activate the sound and light alarm device on site. The sound and light alarm device will quickly attract the attention of the on-site staff by emitting a high-decibel buzzer and flashing light signals.

[0074] Recording Violation Information: In addition to on-site alarms, the system automatically records violation information. This information includes the timestamp, specific location, crane number, detailed description of the skid condition, and relevant image or video evidence, providing important evidence for subsequent safety incident investigations.

[0075] Remote notification of safety managers: To ensure management is promptly informed of site safety conditions and can take appropriate management measures, the system remotely sends violation information to safety managers via the network. Notifications can be sent via text message, email, or instant messaging. Upon receiving the notification, safety managers can decide whether to take immediate action based on the severity and specific circumstances of the violation, such as dispatching maintenance personnel to address the skidding issue or adjusting construction plans to avoid further safety risks.

[0076] After initiating the warning information and uploading the inspection records, this embodiment further utilizes the cloud management platform to conduct in-depth big data analysis of the inspection records. By collecting and integrating inspection records from multiple construction areas, the cloud management platform can collect statistics and analyze the frequency and patterns of missing skids.

[0077] Inspection records include a detailed description of the time and location of the violation, the crane number, the condition of the skids, and any relevant image or video evidence. Data analysis reveals patterns in the distribution of missing skids across different construction areas and time periods. For example, skids may be more likely to be missing in certain areas due to frequent crane operation, or skids may be more susceptible to damage or displacement under specific weather conditions.

[0078] After analysis, the analysis results are fed back to the management end of the on-site detection equipment. The management end of the on-site detection equipment can receive the analysis results sent by the cloud management platform in real time. These results are presented to the on-site management personnel in an intuitive manner through visual charts, reports, etc. If the frequency of missing skids in a certain construction area is high, the management personnel can increase the inspection frequency of the area, or conduct targeted safety training for crane operators to reduce the occurrence of missing skids.

[0079] The crane leg detection method of the present invention reduces the time and labor costs of manual inspections through an automated detection process, while avoiding missed detections and misjudgments caused by human factors, improving detection efficiency, and being able to operate stably in complex construction environments. Even when the leg pads are small or partially obscured, they can be accurately identified, ensuring high robustness of the detection results. The introduction of a multi-level early warning mechanism can promptly notify relevant personnel when missing pads are detected, effectively preventing potential safety accidents and ensuring construction safety. At the same time, through big data analysis of detection records, data support is provided for construction safety management, helping to optimize the construction process and improving the level of construction safety management.

[0080] The embodiment of the present invention further provides a crane skid leg detection device, which is used to execute the crane skid leg detection method described above. Figure 5 This is a structural block diagram of a crane skid leg detection device according to an embodiment of the present invention, the device comprising: The data acquisition module 21 is used to acquire the video stream of the construction area and perform preprocessing to obtain a number of standardized frame image data; The data positioning module 22 is used to perform leg region recognition and coordinate positioning processing on each of the standardized frame image data, and output positioning image data marked with leg position information; An image cropping module 23 is configured to crop the image of the leg region based on the leg position information in the positioning image data to obtain local image data of the leg; A result determination module 24 is configured to perform a wood-bearing state analysis on the leg local image data based on a texture-color joint feature and output a wood-bearing state determination result; The information warning module 25 is used to start the warning information and upload the detection record if the result of the judgment result of the state of the skid is "abnormal".

[0081] The technical features and technical effects of the device proposed in the embodiment of the present invention are the same as those of the method proposed in the embodiment of the present invention and are not described in detail here. Each module in the above-mentioned device can be implemented in whole or in part by software, hardware, or a combination thereof. Each of the above-mentioned modules can be embedded in or independent of the processor in the computer device in the form of hardware, or can be stored in the memory of the computer device in the form of software so that the processor can call and execute the operations corresponding to the above modules.

[0082] See also Figure 6 , which is a structural block diagram of a computer device provided by an embodiment of the present invention. The computer device provided by an embodiment of the present invention includes a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor. When the processor executes the computer program, the steps in the above-mentioned crane skid leg detection method embodiment are implemented, for example Figure 1 or, when the processor executes the computer program, the functions of the modules in the above-mentioned device embodiments are realized, such as module 21 to module 25 of the crane pad leg detection device.

[0083] For example, the computer program may be divided into one or more modules, which are stored in the memory and executed by the processor to implement the present invention. The one or more modules may be a series of computer program instruction segments capable of implementing specific functions, and the instruction segments are used to describe the execution process of the computer program in the computer device.

[0084] The computer device may include, but is not limited to, a processor and a memory. Those skilled in the art will appreciate that the schematic diagram is merely an example of a computer device and does not limit the computer device. The computer device may include more or fewer components than shown, or a combination of certain components, or different components. For example, the computer device may also include input and output devices, network access devices, buses, etc.

[0085] The processor may be a central processing unit (CPU), other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc. A general-purpose processor may be a microprocessor or any conventional processor, etc. The processor is the control center of the computer device, connecting various parts of the entire computer device using various interfaces and lines.

[0086] The memory can be used to store the computer programs and / or modules. The processor implements the various functions of the computer device by running or executing the computer programs and / or modules stored in the memory and accessing the data stored in the memory. The memory may primarily include a program storage area and a data storage area. The program storage area may store an operating system and at least one application required for a function (such as a sound playback function or an image playback function); the data storage area may store data generated based on the use of the mobile phone (such as audio data, a phone book, etc.). Furthermore, the memory may include high-speed random access memory and non-volatile memory, such as a hard disk, internal memory, a plug-in hard disk, a smart media card (SMC), a secure digital (SD) card, a flash card, at least one disk storage device, a flash memory device, or other volatile solid-state storage device.

[0087] If the module integrated into the computer device is implemented as a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention can also implement all or part of the process steps in the above-mentioned method embodiments by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include: any entity or device capable of carrying the computer program code, recording medium, USB flash drive, mobile hard drive, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electrical carrier signal, telecommunication signal, and software distribution medium.

[0088] Those skilled in the art will appreciate that all or part of the processes in the above-described method embodiments can be implemented by instructing the relevant hardware through a computer program. The program can be stored in a computer-readable storage medium, and when executed, the program can include the processes in the above-described method embodiments. The storage medium can be a magnetic disk, an optical disk, a read-only memory (ROM), or a random access memory (RAM).

[0089] Accordingly, an embodiment of the present invention provides a computer-readable storage medium, wherein the computer-readable storage medium includes a stored computer program, wherein when the computer program is executed, the device where the computer-readable storage medium is located is controlled to execute the steps of the crane skid leg detection method as described in the above embodiment, for example Figure 1 Steps S11 to S15 described in .

[0090] The above-described embodiments merely illustrate several implementations of the present invention, and while their descriptions are relatively specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that a person skilled in the art would be able to make numerous variations and improvements without departing from the spirit of the present invention, all of which fall within the scope of protection of the present invention. Therefore, the scope of protection of the present invention shall be determined by the appended claims.

Claims

1. A method for detecting crane skid legs, characterized in that: include: Obtain the video stream of the construction area and perform preprocessing to obtain several standardized frame image data; Performing leg region recognition and coordinate positioning processing on each of the standardized frame image data, and outputting positioning image data marked with leg position information; Cropping the leg area image based on the leg position information in the positioning image data to obtain leg local image data; Performing a wood block presence analysis on the local image data of the outrigger based on a texture-color joint feature, and outputting a wood block status determination result; If the result of the said skid status is judged as "abnormal", the early warning information is activated and the detection record is uploaded.

2. The crane skid leg detection method according to claim 1, wherein: The video stream of the construction area is obtained and pre-processed to obtain a number of standardized frame image data, including: The collected video stream of the construction area is subjected to frame rate adjustment, resolution conversion, color correction and noise filtering to obtain a preprocessed video stream; Sampling from the pre-processed video stream according to a preset number of interval frames to obtain a plurality of sampling frame image data; Each of the sampled frame image data is subjected to standardization processing to obtain standardized frame image data.

3. The crane skid leg detection method according to claim 1, wherein: The performing of leg region recognition and coordinate positioning processing on each of the standardized frame image data, and outputting positioning image data marked with leg position information, includes: Inputting each of the standardized frame image data into the pre-trained KSROI-Net model to perform leg area recognition to obtain bounding box information of the leg area; Calculating the key point coordinates of the leg based on the bounding box information of the leg area; The bounding box information and key point coordinates of the leg area are marked on the corresponding standardized frame image data to obtain positioning image data.

4. The crane skid leg detection method according to claim 3, wherein: The step of cropping the leg region image based on the leg position information in the positioning image data to obtain the leg local image data includes: cropping a leg region image from the corresponding standardized frame image data according to the bounding box information in the positioning image data; Performing super-resolution reconstruction on the leg region image using the ESRGAN algorithm to obtain a first leg region image; grayscale the first leg region image and calculate its local binary pattern features to extract leg texture information; Fusing the leg texture information with the grayscale image of the first leg region image to obtain a second leg region image; Pixel value normalization processing is performed on each of the second leg area images to obtain standardized leg local image data.

5. The crane skid leg detection method according to claim 1, wherein: The step of performing a wood block presence analysis on the local image data of the outrigger based on the texture-color joint feature and outputting a wood block state determination result includes: Input the local image data of the leg into the pre-trained MobileNetV3 network to extract the color features and texture features of the leg area; Performing weighted fusion of the color features and the texture features at different scales through a multi-scale feature fusion module to generate a joint feature representation; The joint feature representation is input into a binary classifier for analysis to obtain a result of the state determination of the skid.

6. The crane skid leg detection method according to claim 1, wherein: If the result of the skid status determination is "abnormal", the warning information is activated and the detection record is uploaded, including: When the result of the skid status determination is "abnormal", a multi-level early warning mechanism is triggered; The multi-level early warning mechanism includes activating sound and light alarm devices, recording violation information and remotely notifying security management personnel.

7. The crane skid leg detection method according to claim 1, wherein: After the warning information is activated and the detection record is uploaded, the following steps are also included: The inspection records are analyzed through a cloud management platform to obtain the frequency and pattern of missing skids in each construction area; Feedback the analysis results to the management end of the on-site detection equipment.

8. A crane skid leg detection device, characterized in that: include: The data acquisition module is used to acquire the video stream of the construction area and perform preprocessing to obtain a number of standardized frame image data; A data positioning module is used to perform leg area recognition and coordinate positioning processing on each of the standardized frame image data, and output positioning image data marked with leg position information; An image cropping module, configured to crop the image of the leg region based on the leg position information in the positioning image data to obtain local image data of the leg; A result determination module is used to perform a wood-block presence analysis on the local image data of the outrigger based on a texture-color joint feature, and output a wood-block state determination result; The information warning module is used to start the warning information and upload the detection record if the result of the said skid status is judged as "abnormal".

9. A computer device, characterized in that: The method comprises a processor, a memory, and a computer program stored in the memory and configured to be executed by the processor, wherein when the processor executes the computer program, the method for detecting crane pad legs according to any one of claims 1 to 7 is implemented.

10. A computer-readable storage medium, characterized in that The computer-readable storage medium stores a computer program, wherein when the device where the computer-readable storage medium is located executes the computer program, the crane pad leg detection method according to any one of claims 1 to 7 is implemented.

Images