A method, related methods and apparatus for detecting dangerous behaviors based on target tracking

Abstract

This invention discloses a method, related methods, and apparatus for detecting dangerous behavior based on target tracking. The method includes: mapping and transforming pedestrian video based on an improved Gaussian model to obtain a transformed image sequence; extracting a frame from the pedestrian video as an initial background image; dividing the pedestrian video and the transformed image sequence into multiple video segments and multiple transformed image segments; analyzing the multiple video segments and multiple transformed image segments based on the initial background image and a target tracking algorithm, and then concatenating the paths to obtain a first target tracking path and a second target tracking path; determining whether the distance between the first target tracking path and the second target tracking path is greater than a preset threshold: if yes, updating the background image to obtain a new first target tracking path, and re-executing the above determination step; if no, using the first target tracking path as the target path; and obtaining the dangerous behavior detection result based on a preset safe action path and the target path.

Description

Technical Field

[0001] This invention relates to a method, related methods and apparatus for detecting dangerous behaviors based on target tracking. Background Technology

[0002] Construction projects are a high-risk industry. While awareness of the importance of safety has deepened and safety management levels have generally improved, the overall safety situation remains challenging. The ever-expanding scale of construction projects, increasingly complex construction processes, and higher technical requirements often present unprecedented difficulties and hidden dangers for safety management. Furthermore, large-scale projects frequently involve multiple layers of subcontracting, from general contractors to subcontractors, specialized subcontractors, and labor subcontractors, further complicating safety supervision. In this context, risk assessment of the project area is based on the risk values ​​determined by work permits to analyze and evaluate the HSE (Health, Safety, and Environment) status and provide real-time risk warnings. Through real-time collection of HSE data from work permits, an evaluation model corresponding to high and low risks in the area is established. Dynamic programming methods are used to monitor the dynamic monitoring and warning indicator data, currently primarily based on target tracking algorithms for hazardous behavior detection. Summary of the Invention

[0003] To obtain more accurate results in detecting dangerous behaviors, embodiments of the present invention provide a method, related methods, and apparatus for detecting dangerous behaviors based on target tracking.

[0004] In a first aspect, embodiments of the present invention provide a method for detecting dangerous behavior based on target tracking, the method comprising:

[0005] The acquired pedestrian video is mapped and transformed based on the improved Gaussian model to obtain a transformed image sequence;

[0006] Extract one frame from the pedestrian video as the initial background image;

[0007] Based on a preset window length and a preset step size, the pedestrian video and the transformed image sequence are respectively divided into multiple video segments and multiple transformed image segments; the preset window length is equal to the preset step size plus one.

[0008] Based on the initial background image and the target tracking algorithm, the multiple video segments and multiple transformed image sequences are analyzed to obtain multiple video segment paths and multiple transformed image sequences.

[0009] Based on the multiple video fragment paths and the multiple transformed image fragment paths, the first target tracking path and the second target tracking path are respectively spliced ​​together;

[0010] Determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold:

[0011] If so, then based on the initial background image, new background images corresponding to the multiple video segments are obtained;

[0012] Based on each video segment and the corresponding new background image, a new video segment path is obtained, and the multiple new video segment paths are spliced ​​together to obtain a new first target tracking path. The above judgment steps are then re-executed.

[0013] If not, the first target tracking path shall be taken as the target path;

[0014] Based on the acquired preset safe action path and the target path, the dangerous behavior detection results are obtained.

[0015] In one or more optional embodiments of this application, obtaining new background images corresponding to the plurality of video segments based on the initial background image includes:

[0016] For each pair of adjacent video segments, the average is calculated to obtain the corresponding average image;

[0017] Perform background subtraction on the two average images and the initial background image respectively to obtain two background difference images;

[0018] Determine whether the similarity between the two background difference images is less than a second preset threshold:

[0019] If so, the average image corresponding to the later video segment in the two adjacent video segments is fused with the initial background image to obtain a new background image corresponding to that video segment.

[0020] If not, the initial background image will be used as the new background image corresponding to that video segment.

[0021] In one or more optional embodiments of this application, the step of mapping and transforming the acquired pedestrian video based on an improved Gaussian model to obtain a transformed image sequence includes:

[0022] For each frame of the pedestrian video, the Gaussian kernel of each pixel is calculated;

[0023] Normalize the Gaussian kernel for each pixel;

[0024] The transformed value of the pixel is obtained by multiplying the Gaussian kernel of each pixel element by the pixel value of the corresponding region and summing the results.

[0025] The transformed image sequence is obtained based on the transformed values ​​of each pixel in each frame of the pedestrian video.

[0026] In one or more optional embodiments of this application, calculating the Gaussian kernel for each pixel of each frame of the pedestrian video includes:

[0027] For each frame of the pedestrian video, calculate each value in the Gaussian kernel for each pixel based on the following formula 1:

[0028]

[0029] In the formula, g(x,y,I) represents the value of pixel (x,y) in the Gaussian kernel, and σ x,y Represents the pixel mapping parameter of pixel (x,y), where I represents the pixel value of pixel (x,y);

[0030] The pixel mapping parameters of pixel (x, y) are calculated based on Equation 2:

[0031]

[0032] In the formula, σ x,y The pixel mapping parameters represent the pixel point (x,y), and N*M represents the size of the Gaussian kernel.

[0033] In one or more optional embodiments of this application, the distance between the first target tracking path and the second target tracking path is calculated in the following manner:

[0034] A preset number of feature points are taken from the first target tracking path and the second target tracking path respectively to obtain a preset number of feature point pairs;

[0035] The mean distance between each pair of feature points is calculated to obtain the distance between the first target tracking path and the second target tracking path.

[0036] Secondly, embodiments of the present invention provide an HSE management method based on risk items and hazard sources, the method comprising:

[0037] Pedestrian videos were obtained based on multimodal data acquisition;

[0038] Based on the pedestrian video, the dangerous behavior detection results obtained by the dangerous behavior detection method based on target tracking;

[0039] Based on the results of the dangerous behavior detection, risk warning actions will be implemented.

[0040] Thirdly, embodiments of the present invention provide a dangerous behavior detection device based on target tracking, the device comprising:

[0041] The first transformation module is used to perform mapping transformation on the acquired pedestrian video based on an improved Gaussian model to obtain a transformed image sequence;

[0042] The first extraction module is used to extract a frame from the pedestrian video as an initial background image;

[0043] The first segmentation module is used to divide the pedestrian video and the transformed image sequence into multiple video segments and multiple transformed image segments, respectively, based on a preset window length and a preset step size; the preset window length is equal to the preset step size plus one.

[0044] The first analysis module is used to analyze the multiple video segments and multiple transformed image sequences based on the initial background image and the target tracking algorithm, respectively, to obtain multiple video segment paths and multiple transformed image sequences.

[0045] The first stitching module is used to stitch together the multiple video fragment paths and the multiple transformed image fragment paths to obtain a first target tracking path and a second target tracking path, respectively.

[0046] The first judgment module is used to determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold:

[0047] The first update module is configured to, if the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold, obtain a new background image corresponding to the plurality of video segments based on the initial background image; obtain a new video segment path based on each video segment and the corresponding new background image, and splice the plurality of new video segment paths to obtain a new first target tracking path, and re-execute the above judgment steps.

[0048] The second update module is used to take the first target tracking path as the target path if the distance between the first target tracking path and the second target tracking path is not greater than a first preset threshold.

[0049] The first acquisition module is used to obtain the dangerous behavior detection results based on the preset safe action path and the target path.

[0050] Fourthly, embodiments of the present invention provide an HSE management device based on risk items and hazard sources, the device comprising:

[0051] The first acquisition module is used to acquire pedestrian videos based on multimodal data.

[0052] The first transformation module is used to perform mapping transformation on the acquired pedestrian video based on an improved Gaussian model to obtain a transformed image sequence;

[0053] The first extraction module is used to extract a frame from the pedestrian video as an initial background image;

[0054] The first segmentation module is used to divide the pedestrian video and the transformed image sequence into multiple video segments and multiple transformed image segments, respectively, based on a preset window length and a preset step size; the preset window length is equal to the preset step size plus one.

[0055] The first analysis module is used to analyze the multiple video segments and multiple transformed image sequences based on the initial background image and the target tracking algorithm, respectively, to obtain multiple video segment paths and multiple transformed image sequences.

[0056] The first stitching module is used to stitch together the multiple video fragment paths and the multiple transformed image fragment paths to obtain a first target tracking path and a second target tracking path, respectively.

[0057] The first judgment module is used to determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold:

[0058] The first update module is configured to, if the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold, obtain a new background image corresponding to the plurality of video segments based on the initial background image; obtain a new video segment path based on each video segment and the corresponding new background image, and splice the plurality of new video segment paths to obtain a new first target tracking path, and re-execute the above judgment steps.

[0059] The second update module is used to take the first target tracking path as the target path if the distance between the first target tracking path and the second target tracking path is not greater than a first preset threshold.

[0060] The first acquisition module is used to obtain the dangerous behavior detection results based on the preset safe action path and the target path;

[0061] The first implementation module is used to implement risk warning actions based on the results of the dangerous behavior detection.

[0062] Fifthly, embodiments of the present invention provide a computer-readable storage medium having a computer program stored thereon, which, when executed by a processor, implements the above-described method for detecting dangerous behavior based on target tracking, and / or an HSE management method based on risk items and hazard sources.

[0063] In a sixth aspect, embodiments of the present invention provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the above-described method for detecting dangerous behaviors based on target tracking, and / or, an HSE management method based on risk items and hazard sources.

[0064] In a seventh aspect, embodiments of the present invention provide a computer program product containing instructions that, when the computer program product is run on a computer device, cause the computer device to perform the above-described method for detecting dangerous behavior based on target tracking, and / or, an HSE management method based on risk items and hazard sources.

[0065] Eighthly, embodiments of the present invention provide a chip, the chip including a processor and a communication interface, the communication interface and the processor being coupled, the processor being used to run computer programs or instructions to implement the above-described method for detecting dangerous behavior based on target tracking, and / or, an HSE management method based on risk items and hazard sources.

[0066] The beneficial effects of the above-described technical solutions provided in the embodiments of the present invention include at least the following:

[0067] The hazardous behavior detection method based on target tracking provided in this invention uses an improved Gaussian model to map and transform pedestrian videos, enhancing the features of the image sequence and making the target easier to track. This improves the fidelity of the transformed image. Simultaneously, by processing the pedestrian video and the transformed image sequence through a window sliding method with a preset step size, a first target tracking path and a second target tracking path are obtained. The comparison between the first and second target tracking paths determines the background image corresponding to each segment of the pedestrian video, thus determining the final target path. This allows the target path to better adapt to background changes, improving the accuracy of target tracking. Therefore, based on a more accurate target path, hazardous behavior detection results are obtained, enabling early warning and control of hazardous behaviors, improving the accuracy of risk analysis and the effectiveness of safety management.

[0068] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in the written description, claims, and drawings.

[0069] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0070] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used together with the embodiments of the invention to explain the invention, but do not constitute a limitation thereof; in the drawings:

[0071] Figure 1 This is a schematic diagram illustrating the steps of a dangerous behavior detection method based on target tracking provided in an embodiment of the present invention;

[0072] Figure 2 A schematic diagram illustrating the steps of the HSE management method based on risk items and hazard sources provided in an embodiment of the present invention;

[0073] Figure 3 This is a schematic diagram of the safety evaluation and early warning system provided in an embodiment of the present invention;

[0074] Figure 4 This is a schematic diagram of a safety evaluation and early warning system provided in an embodiment of the present invention;

[0075] Figure 5 This is a schematic diagram of the structure of a dangerous behavior detection device based on target tracking provided in an embodiment of this application;

[0076] Figure 6 A schematic diagram of the structure of an HSE management device based on risk items and hazard sources provided in an embodiment of this application. Detailed Implementation

[0077] In the following description, specific details such as particular system architectures and techniques are set forth for illustrative purposes and not for limitation, in order to provide a thorough understanding of the embodiments of this application. However, those skilled in the art will understand that this application may also be implemented in other embodiments without these specific details. In other instances, detailed descriptions of well-known systems, apparatuses, circuits, and methods have been omitted so as not to obscure the description of this application with unnecessary detail.

[0078] It should be understood that, when used in this application specification and the appended claims, the term "comprising" indicates the presence of the described features, integrals, steps, operations, elements and / or components, but does not exclude the presence or addition of one or more other features, integrals, steps, operations, elements, components and / or a collection thereof.

[0079] It should also be understood that the term “and / or” as used in this application specification and the appended claims means any combination of one or more of the associated listed items and all possible combinations, and includes such combinations.

[0080] As used in this application specification and the appended claims, the term "if" may be interpreted, depending on the context, as "when," "once," "in response to determination," or "in response to detection." Similarly, the phrase "if determined" or "if detected [the described condition or event]" may be interpreted, depending on the context, as meaning "once determined," "in response to determination," "once detected [the described condition or event]," or "in response to detection [the described condition or event]."

[0081] Furthermore, in the description of this application and the appended claims, the terms "first," "second," "third," etc., are used only to distinguish descriptions and should not be construed as indicating or implying relative importance.

[0082] References to "one embodiment" or "some embodiments" as described in this specification mean that one or more embodiments of this application include a specific feature, structure, or characteristic described in connection with that embodiment. Therefore, the phrases "in one embodiment," "in some embodiments," "in other embodiments," "in still other embodiments," etc., appearing in different parts of this specification do not necessarily refer to the same embodiment, but rather mean "one or more, but not all, embodiments," unless otherwise specifically emphasized. The terms "comprising," "including," "having," and variations thereof mean "including but not limited to," unless otherwise specifically emphasized.

[0083] It should be understood that the sequence number of each step in the following embodiments does not imply the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of this application.

[0084] To illustrate the technical solution of this application, specific embodiments are described below.

[0085] The inventors discovered that in existing technologies, the generalization ability of target tracking algorithms is limited by changes in background scenarios. Using the same background repeatedly leads to inaccurate tracking, making it difficult to adapt to all situations, resulting in performance degradation, insufficient accuracy, and ultimately, inaccurate detection results for dangerous behaviors. Based on this, the inventors conducted further research and development, resulting in this invention, which provides a method, related methods, and apparatus for detecting dangerous behaviors based on target tracking.

[0086] Example 1

[0087] This invention provides a method for detecting dangerous behaviors based on target tracking, referring to... Figure 1 As shown, the method includes:

[0088] S101: Based on the improved Gaussian model, the acquired pedestrian video is mapped and transformed to obtain a transformed image sequence.

[0089] In this embodiment of the application, step S101 above, which involves mapping and transforming the acquired pedestrian video based on an improved Gaussian model to obtain a transformed image sequence, includes:

[0090] For each frame of the pedestrian video, the Gaussian kernel of each pixel is calculated;

[0091] Normalize the Gaussian kernel for each pixel;

[0092] The transformed value of a pixel is obtained by multiplying the Gaussian kernel of each pixel element-wise with the pixel values ​​of the corresponding region and summing the results.

[0093] A sequence of transformed images is obtained by transforming the values ​​of each pixel in each frame of the pedestrian video.

[0094] Specifically, for each frame of the pedestrian video, the Gaussian kernel for each pixel is calculated, including:

[0095] For each frame of the pedestrian video, calculate each value in the Gaussian kernel for each pixel based on the following formula 1:

[0096]

[0097] In the formula, g(x,y,I) represents the value of pixel (x,y) in the Gaussian kernel, and σ x,y Represents the pixel mapping parameter of pixel (x,y), where I represents the pixel value of pixel (x,y);

[0098] The pixel mapping parameters of pixel (x, y) are calculated based on Equation 2:

[0099]

[0100] In the formula, σ x,y The pixel mapping parameters represent the pixel point (x,y), and N*M represents the size of the Gaussian kernel.

[0101] In this embodiment, each frame of the pedestrian video is traversed, and for each pixel in each frame, the Gaussian kernel corresponding to that pixel is calculated based on the above formulas 1 and 2. The Gaussian kernel calculated for each pixel is normalized to ensure that the sum of the weights of the Gaussian kernels is 1, thus preserving the brightness information of the transformed image and preventing information loss. The Gaussian kernel corresponding to each pixel is multiplied element-wise with the pixel values ​​of the local region centered on that pixel, and the results of the multiplication are summed to obtain the transformed value of that pixel. Based on the transformed values ​​of each pixel in each frame, a transformed image sequence is obtained.

[0102] In this embodiment, an improved Gaussian model is used to perform a mapping transformation on pedestrian videos. The resulting transformed image sequence incorporates both positional and pixel features to calculate the pixel mapping parameters of the Gaussian model, which serve as a quality evaluation standard. Compared to the traditional Gaussian function, which only considers positional relationships, the improved Gaussian model proposed in this method considers both positional and pixel information, improving the fidelity of the transformed image and achieving noise reduction. This lays the foundation for accurate subsequent target tracking.

[0103] S102: Extract a frame from the pedestrian video as the initial background image.

[0104] S103: Based on a preset window length and a preset step size, the pedestrian video and the transformed image sequence are divided into multiple video segments and multiple transformed image segments, respectively; the preset window length is equal to the preset step size plus one.

[0105] In this embodiment of the application, a frame is extracted from the pedestrian video as the initial background image. The initial background image can be a frame randomly selected from the pedestrian video, or the first frame of the pedestrian video can be used directly as the initial background image.

[0106] Based on a preset window length and a preset step size, starting from the first frame, the pedestrian video is divided into multiple video segments. The preset window length is equal to the preset step size plus one; that is, if the preset window length is K, then the preset step size is K-1. The preset window length should be set to a positive integer greater than 1. The pedestrian video will be divided into video segments consisting of frames 1 to K, video segments consisting of frames K to 2K-1, and so on, dividing the pedestrian video into multiple video segments. The transformed image sequence is also divided in the same way to obtain multiple segment transformed image sequences.

[0107] S104: Based on the initial background image and the target tracking algorithm, analyze the multiple video segments and multiple transformed image sequences to obtain multiple video segment paths and multiple transformed image sequences.

[0108] S105: Based on the multiple video fragment paths and the multiple transformed image fragment paths, the first target tracking path and the second target tracking path are respectively spliced ​​together.

[0109] In this embodiment, each frame of the video segment and the initial background image are input into a pre-trained convolutional neural network to extract image features, which include semantic information and local structure in the image. Based on the extracted image features, a target detection algorithm is used to obtain the initial position of the target, and a target tracking algorithm is used to track the target to obtain a new position. For each frame of a video segment, the above-mentioned input convolutional neural network is used to extract features and the target position is updated using the target tracking algorithm to obtain the video segment path corresponding to that video segment. In this way, all video segments and segment-transformed image sequences are processed in the same way to obtain multiple video segment paths and multiple segment-transformed image paths.

[0110] The video paths corresponding to every two adjacent video segments from multiple video clips are concatenated. Since there are overlapping frames between adjacent video clips, the transition between the two video clip paths is relatively smooth, avoiding unnatural jumps at the concatenation points. After concatenating the multiple video clip paths, the first target tracking path is obtained. Similarly, the video paths corresponding to every two adjacent transformed image sequences from multiple transformed image sequences are concatenated to obtain the second target tracking path.

[0111] S106: Determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold: if yes, proceed to step S107; if no, proceed to step S108.

[0112] In this embodiment of the application, in step S106 above, it is determined whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold, wherein the distance between the first target tracking path and the second target tracking path is calculated in the following manner:

[0113] A preset number of feature points are taken from the first target tracking path and the second target tracking path respectively to obtain a preset number of feature point pairs;

[0114] Calculate the mean distance between each pair of feature points to obtain the distance between the first target tracking path and the second target tracking path.

[0115] In this embodiment of the application, the specific steps for calculating the distance between the first target tracking path and the second target tracking path include: randomly selecting a preset number of feature points from the first target tracking path, finding corresponding feature points in the second target tracking path, and forming a preset number of feature point pairs. The two feature points in each feature point pair need to have the same features in the first target tracking path and the second target tracking path. For example, the two feature points are the first inflection points of the first target tracking path and the second target tracking path, or the slopes of the two feature points are the same, etc.; calculating the distance between each feature point pair, which can be calculated using Euclidean distance, and averaging the distances of each feature point pair to obtain the distance between the first target tracking path and the second target tracking path.

[0116] S107: Based on the initial background image, obtain new background images corresponding to the multiple video segments; based on each video segment and the corresponding new background image, obtain new video segment paths, and splice the multiple new video segment paths to obtain a new first target tracking path, and re-execute the judgment step of S106.

[0117] S108: Use the first target tracking path as the target path;

[0118] In this embodiment of the application, step S107 above, which involves obtaining new background images corresponding to multiple video segments based on the initial background image, includes:

[0119] For each pair of adjacent video segments, the average is calculated to obtain the corresponding average image;

[0120] Perform background subtraction on the two average images and the initial background image respectively to obtain two background difference images;

[0121] Determine whether the similarity between two background difference images is less than a second preset threshold:

[0122] If so, the average image corresponding to the later video segment in the time sequence of two adjacent video segments is fused with the initial background image to obtain a new background image corresponding to that video segment.

[0123] If not, the initial background image will be used as the new background image for that video segment.

[0124] In this embodiment, if the distance between the first target tracking path and the second target tracking path is not greater than a first preset threshold, the background image of the segment image needs to be replaced. The specific steps include: for every two adjacent video segments, performing video frame averaging on the two video segments respectively, that is, accumulating and averaging the pixel values ​​at the same position in all frames within a video segment to obtain the average image corresponding to the two video segments; performing background subtraction operation on the two average images with the initial background image respectively, that is, subtracting the average image from the initial background image to obtain the background difference image corresponding to the two average images; calculating the similarity between the two background difference images, which can be done using similarity graph evaluation algorithms such as Structural Similarity Index (SSIM) and histogram comparison to obtain the similarity result of the two background difference images; determining whether the similarity result is less than a second preset threshold: if yes, then fusing the average image corresponding to the later video segment in the time sequence of the two adjacent video segments with the initial background image to obtain a new background image corresponding to the video segment; if no, then using the initial background image as the new background image of the later video segment in the time sequence of the two adjacent video segments.

[0125] After updating the background images of all adjacent video segments in the pedestrian video, a new video segment path is obtained based on each video segment and its corresponding new background image. Multiple new video segment paths are then concatenated to obtain a new first target tracking path, as detailed in steps S104 and S105. Step S106 is then re-executed based on the new first target tracking path and the second target tracking path until the distance between the first target tracking path and the second target tracking path is no greater than a first preset threshold. The first target tracking path is then used as the target path.

[0126] In this embodiment, by updating the background image corresponding to each video segment, better adaptation to background changes is achieved, thereby improving the accuracy of target tracking. By continuously updating the background model, this method can capture dynamic changes in the environment, such as lighting, occlusion, or scene changes, making the target tracking algorithm more robust. This real-time adaptive mechanism helps ensure that target tracking can work stably in complex and dynamic scenes.

[0127] S109: Based on the acquired preset safe action path and the target path, obtain the dangerous behavior detection result.

[0128] In this embodiment of the application, the obtained preset safe action path and target path are compared. If the comparison results are too different and the target path reaches a preset inaccessible location, it is determined that the pedestrian target in the pedestrian video has engaged in dangerous behavior, and the dangerous behavior detection result is "existing".

[0129] In this embodiment of the application, a schematic diagram of the safety evaluation and early warning system implemented based on this method is shown below. Figure 3 As shown, the input consists of HSE-related parameters, including safety conditions and status records, corresponding to the preset safe action path and pedestrian video mentioned above; the evaluation and early warning model in the calculation part includes HSE data statistical analysis and JSA analysis, which correspond to the dangerous behavior detection method based on target tracking; the output is an evaluation and early warning report, including early warning prompts and evaluation reports.

[0130] A schematic diagram of the safety assessment and early warning system is shown below. Figure 4 As shown, it includes three parts: data collection, HSE status assessment and early warning, and data visualization, which correspond to... Figure 3 The system comprises three parts. The data acquisition function includes personnel management, vehicle / machinery management, HSE inspection management, HSE examination management, green and civilized construction, work permit management, JSA management, safe man-hours, vehicle / machinery mileage, accidents and incidents, violation inspections, unsafe behavior identification, personnel examination pass rate, pre-work safety analysis form, and work permit processing. The HSE status evaluation and early warning function and related data include accident and incident evaluation reports per million man-hours / vehicle mileage, JSA risk database, pre-work safety analysis form, project area risk level early warning, subcontractor HSE priority evaluation, subcontractor scoring form, subcontractor management regulations, and subcontractor scoring deduction item early warning. The data visualization function includes evaluation reports, a project area risk map, and a 3D model.

[0131] In this embodiment, the aforementioned safety evaluation and early warning system can be simplified to seven parts: data collection, specific function management, HSE status and risk early warning model, HSE behavior intelligent recognition model, HSE problem closed-loop tracking, HSE reward and punishment management, and HSE evaluation report.

[0132] The data collection section mainly includes: automatically counting the total number of on-site personnel; using a facial recognition access control system to achieve one-stop digital security management for attendance, personnel qualifications, identity verification, physical examinations, practical training, mobile training, examinations, automatic access control authorization, and hat sticker printing; counting the total number of vehicles; processing vehicle entry applications and approvals; compiling vehicle information and creating QR codes for printing; affixing the QR codes to a prominent position on the vehicle when issuing entry permits; and allowing users to view the vehicle's real-time status and information by scanning the QR code label with a mobile app; and using indicators such as the number of tools and equipment, safe man-hours, number of violations, number of accidents, violation rate, and accident rate as basic statistical data for safety status analysis.

[0133] The specific functional management includes four parts: examination management, work ticket management, inspection management, and emergency management. Examination management involves compiling HSE professional knowledge and operational skills related to construction work into a question bank, importing it into the system, and allowing HSE managers to select the appropriate questions for online testing. The system automatically calculates scores to determine pass / fail. Work ticket management digitizes the process of managing work tickets for direct work operations. Through integration, managers issue work tickets based on system-provided work information and site conditions, binding electronic work tickets to their relevant work areas and corresponding managers. Data is integrated into the platform, providing a layered, transparent display of all types of on-site work information, supporting online querying and supervision of manager performance. High-risk operations are also implemented in the model. The system features a unified map for the industry; inspection and management implements an electronic point system for contractors to assess violations, with points assigned to individuals. Once a person reaches the maximum score, their access authorization is automatically suspended, requiring retraining and passing an assessment before they can enter the site. For serious violations, contractor personnel are directly blacklisted and removed from the site. A grid-based intelligent monitoring function for professional safety personnel performance has been developed to verify the performance of dedicated safety management personnel from the owner, general contractor, supervisor, and construction party. This includes personnel location, trajectory tracking, and online time statistics, and regularly generates performance evaluation reports. Emergency management involves inputting the names, quantities, usage methods, dosages, and storage locations of first-aid equipment and medicines into the system and pushing them to mobile devices. All personnel entering the site can quickly view the corresponding first-aid supplies and contact the administrator in case of an emergency.

[0134] The HSE status and risk warning model is established by collecting HSE data to create an evaluation model that corresponds to high and low risks in a region. Based on the risk value of the work ticket, the risk of the project area is determined to analyze and evaluate the HSE status and provide real-time risk warnings.

[0135] The HSE (Health, Safety, and Environment) behavior intelligent recognition model applies a target tracking-based hazardous behavior detection method. The model is deployed on an industrial control computer, hardware is connected, and debugging is performed. After debugging, the entire system is deployed on-site for intelligent personnel behavior recognition. The recognition results are recorded, analyzed, fed back, and the behavior intelligent recognition model is optimized.

[0136] HSE issue closed-loop tracking involves issuing violation notices via video, images, text, and voice when violations are encountered. After review by the branch HSE manager, the contractor manager receives the notice and provides feedback on the rectification results, thus achieving closed-loop management.

[0137] HSE reward and punishment management, as an important means of safety management, aims to create a management atmosphere that promotes good behavior and discourages bad behavior. On-site, violations are differentiated by level and severity, with good safety behaviors rewarded and violations punished to standardize safe construction practices.

[0138] The HSE evaluation report is generated by automatically analyzing the on-site safety status and evaluation indicators based on business processes such as entry safety training, vehicle entry registration, work permit approval, on-site inspection and problem rectification, and compiling the evaluation report according to the corresponding requirements.

[0139] In this embodiment, all-weather, full-cycle analysis of risks and hazards during construction is achieved, accurately reflecting the safety status of the construction site, predicting trends, and providing early warnings and control for impending risks. This improves the accuracy of risk analysis and the effectiveness of safety management. By combining theories of hazard source research, inherently safe systems, and accident causation, the mechanism of construction hazards is analyzed from the source of accidents, establishing risk indicators to safeguard safe construction.

[0140] Example 2

[0141] Based on the same inventive concept, embodiments of the present invention also provide an HSE management method based on risk items and hazard sources, referring to... Figure 2 As shown, the method includes:

[0142] S201: Based on data collection, pedestrian videos are obtained;

[0143] S202: Based on the improved Gaussian model, the acquired pedestrian video is mapped and transformed to obtain a transformed image sequence;

[0144] S203: Extract a frame from the pedestrian video as an initial background image;

[0145] S204: Based on a preset window length and a preset step size, the pedestrian video and the transformed image sequence are respectively divided into multiple video segments and multiple transformed image segments; the preset window length is equal to the preset step size plus one;

[0146] S205: Based on the initial background image and the target tracking algorithm, analyze the multiple video segments and multiple transformed image sequences to obtain multiple video segment paths and multiple transformed image sequences;

[0147] S206: Based on the multiple video fragment paths and the multiple transformed image fragment paths, a first target tracking path and a second target tracking path are respectively spliced ​​together;

[0148] S207: Determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold.

[0149] S208: If so, then based on the initial background image, obtain a new background image corresponding to the multiple video segments;

[0150] Based on each video segment and the corresponding new background image, a new video segment path is obtained, and the multiple new video segment paths are spliced ​​together to obtain a new first target tracking path. The above judgment steps are then re-executed.

[0151] S209: If not, take the first target tracking path as the target path;

[0152] S210: Based on the acquired preset safe action path and the target path, obtain the dangerous behavior detection result.

[0153] S211: Based on the results of the dangerous behavior detection, implement risk warning actions.

[0154] Example 3

[0155] Based on the same inventive concept, embodiments of the present invention also provide a dangerous behavior detection device based on target tracking, referring to... Figure 5 As shown, the device includes:

[0156] The first transformation module 101 is used to perform mapping transformation on the acquired pedestrian video based on an improved Gaussian model to obtain a transformed image sequence;

[0157] The first extraction module 102 is used to extract a frame from the pedestrian video as an initial background image;

[0158] The first segmentation module 103 is used to divide the pedestrian video and the transformed image sequence into multiple video segments and multiple transformed image segments, respectively, based on a preset window length and a preset step size; the preset window length is equal to the preset step size plus one.

[0159] The first analysis module 104 is used to analyze the multiple video segments and multiple transformed image sequences based on the initial background image and the target tracking algorithm, respectively, to obtain multiple video segment paths and multiple transformed image sequences.

[0160] The first stitching module 105 is used to stitch together the multiple video fragment paths and the multiple transformed image fragment paths to obtain a first target tracking path and a second target tracking path, respectively.

[0161] The first judgment module 106 is used to determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold.

[0162] The first update module 107 is used to obtain a new background image corresponding to the plurality of video segments based on the initial background image if the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold; obtain a new video segment path based on each video segment and the corresponding new background image, and splice the plurality of new video segment paths to obtain a new first target tracking path, and re-execute the above judgment steps.

[0163] The second update module 108 is used to take the first target tracking path as the target path if the distance between the first target tracking path and the second target tracking path is not greater than a first preset threshold.

[0164] The first acquisition module 109 is used to obtain dangerous behavior detection results based on the preset safe action path and the target path.

[0165] Example 4

[0166] Based on the same inventive concept, embodiments of the present invention also provide an HSE management device based on risk items and hazard sources, referring to... Figure 6 As shown, the device includes:

[0167] The first acquisition module 201 is used to acquire pedestrian videos based on multimodal data.

[0168] The first transformation module 202 is used to perform mapping transformation on the acquired pedestrian video based on an improved Gaussian model to obtain a transformed image sequence;

[0169] The first extraction module 203 is used to extract a frame from the pedestrian video as an initial background image;

[0170] The first segmentation module 204 is used to divide the pedestrian video and the transformed image sequence into multiple video segments and multiple transformed image segments, respectively, based on a preset window length and a preset step size; the preset window length is equal to the preset step size plus one.

[0171] The first analysis module 205 is used to analyze the multiple video segments and multiple transformed image sequences based on the initial background image and the target tracking algorithm, respectively, to obtain multiple video segment paths and multiple transformed image sequences.

[0172] The first stitching module 206 is used to stitch together the multiple video fragment paths and the multiple transformed image fragment paths to obtain a first target tracking path and a second target tracking path, respectively.

[0173] The first judgment module 207 is used to determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold.

[0174] The first update module 208 is configured to, if the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold, obtain a new background image corresponding to the plurality of video segments based on the initial background image; obtain a new video segment path based on each video segment and the corresponding new background image, and splice the plurality of new video segment paths to obtain a new first target tracking path, and re-execute the above judgment steps.

[0175] The second update module 209 is used to take the first target tracking path as the target path if the distance between the first target tracking path and the second target tracking path is not greater than a first preset threshold.

[0176] The first acquisition module 210 is used to obtain dangerous behavior detection results based on the preset safe action path and the target path;

[0177] The first implementation module 211 is used to implement risk warning actions based on the results of the dangerous behavior detection.

[0178] Example 5

[0179] Based on the same inventive concept, embodiments of the present invention also provide a computer-readable storage medium storing a computer program thereon, which, when executed by a processor, implements the dangerous behavior detection method based on target tracking as described in Embodiment 1 above, and / or the HSE management method based on risk items and hazard sources as described in Embodiment 2 above.

[0180] Example 6

[0181] Based on the same inventive concept, embodiments of the present invention also provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the dangerous behavior detection method based on target tracking as described in Embodiment 1 above, and / or the HSE management method based on risk items and hazard sources as described in Embodiment 2 above.

[0182] Example 7

[0183] Based on the same inventive concept, this embodiment of the invention also provides a computer program product containing instructions that, when the computer program product is run on a computer device, cause the computer device to execute the dangerous behavior detection method based on target tracking as described in Embodiment 1 above, and / or the HSE management method based on risk items and hazard sources as described in Embodiment 2 above.

[0184] Example 8

[0185] Based on the same inventive concept, this embodiment of the invention also provides a chip, which includes a processor and a communication interface, the communication interface and the processor being coupled together. The processor is used to run computer programs or instructions to implement the dangerous behavior detection method based on target tracking as described in Embodiment 1 above, and / or the HSE management method based on risk items and hazard sources as described in Embodiment 2 above.

[0186] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage and optical storage) containing computer-usable program code.

[0187] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0188] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1 One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0189] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0190] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A method for detecting dangerous behaviors based on target tracking, characterized in that, include: The acquired pedestrian video is mapped and transformed based on the improved Gaussian model to obtain a transformed image sequence; Extract one frame from the pedestrian video as the initial background image; Based on a preset window length and a preset step size, the pedestrian video and the transformed image sequence are respectively divided into multiple video segments and multiple transformed image segments; The preset window length is equal to the preset step size plus one; Based on the initial background image and the target tracking algorithm, the multiple video segments and multiple transformed image sequences are analyzed to obtain multiple video segment paths and multiple transformed image sequences. Based on the multiple video fragment paths and the multiple transformed image fragment paths, the first target tracking path and the second target tracking path are respectively spliced ​​together; Determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold: If so, then based on the initial background image, new background images corresponding to the multiple video segments are obtained; Based on each video segment and the corresponding new background image, a new video segment path is obtained, and the multiple new video segment paths are spliced ​​together to obtain a new first target tracking path. The above judgment steps are then re-executed. If not, the first target tracking path shall be taken as the target path; Based on the acquired preset safe action path and the target path, the dangerous behavior detection results are obtained.

2. The method as described in claim 1, characterized in that, The process of obtaining new background images corresponding to the multiple video segments based on the initial background image includes: For each pair of adjacent video segments, the average is calculated to obtain the corresponding average image; Perform background subtraction on the two average images and the initial background image respectively to obtain two background difference images; Determine whether the similarity between the two background difference images is less than a second preset threshold: If so, the average image corresponding to the later video segment in the two adjacent video segments is fused with the initial background image to obtain a new background image corresponding to that video segment. If not, the initial background image will be used as the new background image corresponding to that video segment.

3. The method as described in claim 1, characterized in that, The modified Gaussian model is used to perform a mapping transformation on the acquired pedestrian video to obtain a transformed image sequence, including: For each frame of the pedestrian video, the Gaussian kernel of each pixel is calculated; Normalize the Gaussian kernel for each pixel; The transformed value of the pixel is obtained by multiplying the Gaussian kernel of each pixel element by the pixel value of the corresponding region and summing the results. The transformed image sequence is obtained based on the transformed values ​​of each pixel in each frame of the pedestrian video.

4. The method as described in claim 3, characterized in that, The calculation of the Gaussian kernel for each pixel in each frame of the pedestrian video includes: For each frame of the pedestrian video, calculate each value in the Gaussian kernel for each pixel based on the following formula 1: In the formula, g(x,y,I) represents the value of pixel (x,y) in the Gaussian kernel, and σ x,y Represents the pixel mapping parameter of pixel (x,y), where I represents the pixel value of pixel (x,y); The pixel mapping parameters of pixel (x, y) are calculated based on Equation 2: In the formula, σ x,y The pixel mapping parameters represent the pixel point (x,y), and N*M represents the size of the Gaussian kernel.

5. The method as described in claim 1, characterized in that, The distance between the first target tracking path and the second target tracking path is calculated in the following manner: A preset number of feature points are taken from the first target tracking path and the second target tracking path respectively to obtain a preset number of feature point pairs; The mean distance between each pair of feature points is calculated to obtain the distance between the first target tracking path and the second target tracking path.

6. An HSE management method based on risk items and hazard sources, characterized in that, The method includes: Pedestrian videos were obtained based on multimodal data acquisition; Based on the pedestrian video, the dangerous behavior detection result is obtained based on the dangerous behavior detection method based on target tracking according to any one of claims 1-5; Based on the results of the dangerous behavior detection, risk warning actions will be implemented.

7. A dangerous behavior detection device based on target tracking, characterized in that, include: The first transformation module is used to perform mapping transformation on the acquired pedestrian video based on an improved Gaussian model to obtain a transformed image sequence; The first extraction module is used to extract a frame from the pedestrian video as an initial background image; The first segmentation module is used to divide the pedestrian video and the transformed image sequence into multiple video segments and multiple transformed image segments, respectively, based on a preset window length and a preset step size; the preset window length is equal to the preset step size plus one. The first analysis module is used to analyze the multiple video segments and multiple transformed image sequences based on the initial background image and the target tracking algorithm, respectively, to obtain multiple video segment paths and multiple transformed image sequences. The first stitching module is used to stitch together the multiple video fragment paths and the multiple transformed image fragment paths to obtain a first target tracking path and a second target tracking path, respectively. The first judgment module is used to determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold: The first update module is used to obtain a new background image corresponding to the multiple video segments based on the initial background image if the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold. Based on each video segment and the corresponding new background image, a new video segment path is obtained, and the multiple new video segment paths are spliced ​​together to obtain a new first target tracking path. The above judgment steps are then re-executed. The second update module is used to take the first target tracking path as the target path if the distance between the first target tracking path and the second target tracking path is not greater than a first preset threshold. The first acquisition module is used to obtain the dangerous behavior detection results based on the preset safe action path and the target path.

8. An HSE management device based on risk items and hazard sources, characterized in that, include: The first acquisition module is used to acquire pedestrian videos based on multimodal data. The first transformation module is used to perform mapping transformation on the acquired pedestrian video based on an improved Gaussian model to obtain a transformed image sequence; The first extraction module is used to extract a frame from the pedestrian video as an initial background image; The first segmentation module is used to divide the pedestrian video and the transformed image sequence into multiple video segments and multiple transformed image segments, respectively, based on a preset window length and a preset step size; the preset window length is equal to the preset step size plus one. The first analysis module is used to analyze the multiple video segments and multiple transformed image sequences based on the initial background image and the target tracking algorithm, respectively, to obtain multiple video segment paths and multiple transformed image sequences. The first stitching module is used to stitch together the multiple video fragment paths and the multiple transformed image fragment paths to obtain a first target tracking path and a second target tracking path, respectively. The first judgment module is used to determine whether the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold: The first update module is used to obtain a new background image corresponding to the multiple video segments based on the initial background image if the distance between the first target tracking path and the second target tracking path is greater than a first preset threshold. Based on each video segment and the corresponding new background image, a new video segment path is obtained, and the multiple new video segment paths are spliced ​​together to obtain a new first target tracking path. The above judgment steps are then re-executed. The second update module is used to take the first target tracking path as the target path if the distance between the first target tracking path and the second target tracking path is not greater than a first preset threshold. The first acquisition module is used to obtain the dangerous behavior detection results based on the preset safe action path and the target path; The first implementation module is used to implement risk warning actions based on the results of the dangerous behavior detection.

9. A computer-readable storage medium storing instructions that, when executed on a terminal, cause the terminal to perform the hazardous behavior detection method based on target tracking as described in any one of claims 1-5, and / or the HSE management method based on risk items and hazard sources as described in claim 6.

10. A computer device, characterized in that, It includes a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor executes the computer program to implement the hazardous behavior detection method based on target tracking as described in any one of claims 1-5, and / or the HSE management method based on risk items and hazard sources as described in claim 6.

11. A computer program product containing instructions that, when run on a computer device, causes the computer device to perform the target-tracking-based hazardous behavior detection method as described in any one of claims 1-5, and / or the HSE management method based on risk items and hazard sources as described in claim 6.

12. A chip comprising a processor and a communication interface coupled to the processor, the processor being configured to run computer programs or instructions to implement the target tracking-based hazardous behavior detection method as described in any one of claims 1-5, and / or the HSE management method based on risk items and hazard sources as described in claim 6.

Images