Methods, systems and equipment for safety management and monitoring of outdoor signs

Abstract

This application provides a method, system, and device for monitoring and managing the safety of outdoor signs. The method acquires video surveillance data, IoT sensor data, and map point-of-interest (POI) data of the target outdoor sign. Based on the video surveillance data, a defect score characterizing the severity of appearance defects of the target outdoor sign is determined. Then, based on the IoT sensor data, a stability score characterizing the structural stability of the target outdoor sign is determined. Next, based on the spatial relationship between the sign's corresponding POI and surrounding sensitive POIs in the map POI data, a spatial risk weight is determined. Finally, based on the defect score, stability score, and spatial risk weight, a comprehensive risk score for the target outdoor sign is calculated. Finally, based on the comprehensive risk score and the distribution of surrounding sensitive POIs, the scope of accident impact is determined, and early warning information is pushed out, thereby effectively predicting safety problems of surrounding outdoor signs and achieving efficient safety monitoring and early warning.

Description

Technical Field

[0001] This application relates to safety monitoring technology, and more particularly to a method, system and equipment for safety management and monitoring of outdoor signs. Background Technology

[0002] Outdoor signs are an important part of urban public spaces, and their safety is directly related to the safety of public life and property.

[0003] Currently, the safety management of outdoor signs mainly relies on manual inspections. This method is inefficient, has limited coverage, and struggles to accurately identify subtle, potential structural defects. Furthermore, existing technological solutions often focus solely on image recognition to determine sign appearance defects or rely solely on sensor data to assess structural condition, failing to effectively integrate appearance, structure, and surrounding environmental information. This single-dimensional risk assessment method can not only lead to biased risk classification but also fail to accurately predict the actual impact of a safety incident on surrounding pedestrian and traffic flow. Consequently, early warning and response measures lack specificity and precision, making it difficult to meet the demands of modern, sophisticated urban management.

[0004] Therefore, there is an urgent need for a method for monitoring and managing the safety of outdoor signs that can integrate information from multiple sources and achieve accurate risk assessment and impact prediction. Summary of the Invention

[0005] This application provides a method, system, and equipment for monitoring and managing the safety of outdoor signs, in order to solve the technical problems of the existing technology, such as the single dimension of safety management of outdoor signs, inaccurate risk assessment, and inability to effectively predict the surrounding impact, and to achieve efficient safety monitoring and early warning.

[0006] Firstly, this application provides a method for monitoring and managing the safety of outdoor signs, including: Acquire video surveillance data, IoT sensor data, and map point-of-interest data of the target outdoor signboard; Based on the video surveillance data, a defect score is determined to characterize the severity of the appearance defects of the target outdoor sign. Based on the IoT sensor data, a stability score characterizing the structural stability of the target outdoor signboard is determined; Based on the spatial relationship between the sign interest point corresponding to the target outdoor sign and the surrounding sensitive interest points in the map interest point data, the spatial risk weight is determined. The comprehensive risk score of the target outdoor sign is determined based on the defect score, the stability score, and the spatial risk weight. Based on the comprehensive risk score and the distribution of surrounding sensitive points of interest, the scope of the accident's impact is determined, and early warning information is pushed out.

[0007] The outdoor sign safety management and monitoring method, system and equipment provided in this application can obtain quantitative indicators representing the safety status of outdoor signs from three dimensions: appearance defects, structural stability and spatial environment, by collaboratively analyzing video surveillance data, IoT sensor data and map point of interest data.

[0008] Specifically, by calculating defect scores, stability scores, and spatial risk weights, and then integrating them into a comprehensive risk score, a multi-dimensional and quantitative assessment of the safety risks of outdoor signs is achieved, overcoming the one-sidedness of single-dimensional assessment and thus more accurately reflecting the true risk level of outdoor signs.

[0009] Furthermore, this application combines the calculated comprehensive risk score with the distribution of surrounding sensitive points of interest to dynamically determine the scope of accident impact, so that the early warning information can not only reflect the risk level, but also indicate specific potential danger areas, thereby making risk warnings more accurate and targeted.

[0010] Ultimately, the early warning information determined in this way can be pushed to the corresponding handling terminal, realizing closed-loop management from risk identification, assessment, prediction to early warning handling, which significantly improves the intelligence level and efficiency of outdoor sign safety management. Attached Figure Description

[0011] The accompanying drawings, which are incorporated in and form part of this specification, illustrate embodiments consistent with this application and, together with the description, serve to explain the principles of this application.

[0012] Figure 1 This is a flowchart illustrating an outdoor sign safety management and monitoring method according to an example embodiment of this application; Figure 2 This is a schematic diagram of the structure of an outdoor sign safety management and monitoring system according to an example embodiment of this application; Figure 3 This is a schematic diagram of the structure of an electronic device according to an example embodiment of this application.

[0013] The accompanying drawings illustrate specific embodiments of this application, which will be described in more detail below. These drawings and descriptions are not intended to limit the scope of the concept in any way, but rather to illustrate the concept of this application to those skilled in the art through reference to particular embodiments. Detailed Implementation

[0014] Exemplary embodiments will now be described in detail, examples of which are illustrated in the accompanying drawings. When the following description relates to the drawings, unless otherwise indicated, the same numbers in different drawings denote the same or similar elements. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with this application. Rather, they are merely examples of apparatuses and methods consistent with some aspects of this application as detailed in the appended claims.

[0015] The outdoor sign safety management and monitoring method provided in this application can be implemented in the outdoor facility safety supervision system of urban management departments. This system integrates functions such as image acquisition, IoT sensing, geographic information analysis, and early warning push notifications. This application aims to achieve quantitative assessment and accurate early warning of the safety status of outdoor signs through collaborative analysis of multi-source heterogeneous data.

[0016] In this application, the key terms that need to be explained first include: Defect Score: A quantitative assessment value used to characterize the severity of appearance defects (such as deformation, rust, loose supports, and missing parts) of target outdoor signs identified from video surveillance data. The higher the score, the more serious the appearance defect.

[0017] Stability Score: A quantitative assessment value used to characterize the structural stability of a target outdoor sign based on data analyzed from IoT sensors. This score comprehensively reflects the object's tilt, vibration, and other characteristics; a higher score indicates poorer structural stability.

[0018] Spatial risk weight: A weight calculated based on geospatial relationships to characterize the degree of additional risk arising from the distribution of sensitive facilities around the geographical location of a target outdoor sign. This weight is determined by spatial relationships such as distance and density between the sign's point of interest and surrounding sensitive points of interest.

[0019] Comprehensive Risk Score: A multi-dimensional integrated assessment value that comprehensively measures the overall probability of a target outdoor signboard experiencing a safety accident by integrating the aforementioned defect score, stability score, and spatial risk weight using a preset calculation method.

[0020] Figure 1 This is a flowchart illustrating an outdoor sign safety management and monitoring method according to an example embodiment of this application. Figure 1 As shown, the method provided in this embodiment includes: S100: Acquire video surveillance data, IoT sensor data, and map point of interest data of the target outdoor sign.

[0021] Optionally, the aforementioned IoT sensor data includes tilt data, vibration data, and temperature and humidity data. Map point-of-interest (POI) data includes attribute information for signboard POIs, type information for surrounding sensitive POIs, and road network POI information.

[0022] Specifically, tilt data directly reflects the offset posture of the signpost and support, and is a core indicator for judging the risk of overturning. Vibration data reveals the dynamic response caused by loose supports, component fatigue or strong winds. Temperature and humidity data provide environmental correction basis for the above structural data. For example, the thermal expansion and contraction of metal components will affect the tilt reading.

[0023] In map point of interest data, the attribute information of signboard points of interest, such as size, material, and installation height, is the basis for calculating the potential hazard range of an accident.

[0024] Information about the types of nearby sensitive points of interest, such as "schools," "hospitals," and "gas stations," determines the risk sensitivity level of the area. Road network point of interest information provides information on road capacity and the number of lanes, which is indispensable when analyzing the risk of traffic paralysis caused by accidents.

[0025] S200. Based on video surveillance data, determine the defect score that characterizes the severity of appearance defects of the target outdoor sign.

[0026] In this step, video surveillance data can be input into a pre-trained video image recognition model to obtain a defect score. The video image recognition model is used to extract appearance defect features from the video surveillance data and generate a defect score based on the appearance defect features.

[0027] Specifically, the process involves extracting frames from the collected video surveillance data to obtain a continuous sequence of video frame images. Then, the video frame image sequence is normalized in size and pixel values ​​to obtain standardized image data that meets the input requirements of the video image recognition model. Next, the standardized image data is input into a pre-trained video image recognition model, which includes convolutional neural network layers for extracting appearance defects such as cracks, damage, corrosion, peeling, and deformation on the signboard surface. The feature maps generated by the convolutional neural network layers are then classified into defect levels through fully connected layers, and quantified defect scores are generated based on the defect level classification results according to a preset mapping relationship. This preset mapping relationship is used to convert the confidence levels corresponding to appearance defect features such as crack length, damaged area, and rust spot density into numerical defect scores.

[0028] It is worth noting that the construction and training of a pre-trained video image recognition model includes: The original training video of the outdoor sign scene is analyzed frame by frame to obtain a labeled training sample set containing different appearance defect types such as cracks, damage, rust, peeling and deformation. The labeled training sample set includes the defect type label and defect severity label corresponding to each video frame image.

[0029] A video image recognition model is constructed with the following structure: an input layer, a multi-layer convolutional neural network layer for extracting local texture information, a transformer self-attention layer for aggregating global defect region relationships, a feature compression layer for generating appearance defect feature vectors, and a fully connected classification layer for outputting defect level prediction results.

[0030] Downsampling convolutional units with a stride of 2 are set in the multi-layer convolutional neural network of the model to enhance the ability to identify large-area peeling and large-scale cracks. A multi-head attention mechanism is set in the self-attention layer of the transformer of the model, with each attention head used to focus on the subtle defect areas at different locations on the surface of the sign to capture the spatial correlation of the defects in the appearance of the sign.

[0031] Then, using the labeled training sample set as input samples, the video image recognition model is trained in a supervised manner, and the cross-entropy loss function is used to backpropagate the error of the defect level classification results.

[0032] During training, the model parameters were updated using the Adam optimizer with a learning rate of 0.0001, and the generalization ability of the model to changes in lighting and shooting angle in real-world outdoor signboard scenes was improved by batch training with 32 frames per batch.

[0033] After training, the confidence vector corresponding to the surface defect features of the signboard is input into the fully connected classification layer. Based on the preset defect level label, the defect level prediction value is output, and the defect score is generated based on the defect level prediction value according to the preset scoring mapping rule.

[0034] Among them, the defect score is used to characterize the severity of the appearance defects of outdoor signs within the corresponding time period of video surveillance data, and is used in conjunction with the structural stability score and spatial risk weight of outdoor signs to calculate the comprehensive risk score.

[0035] S300: Based on IoT sensor data, determine the stability score characterizing the structural stability of the target outdoor sign.

[0036] In this step, IoT sensor data can be input into a pre-trained time-series monitoring model to obtain a stability score. The time-series monitoring model is used to perform time-series encoding on IoT sensor data and generate a stability score based on the encoding results.

[0037] Specifically, the construction and training of the aforementioned pre-trained time-series monitoring model includes: The tilt data, vibration data, and temperature and humidity data collected in the outdoor sign monitoring scenario are time-series labeled to form a labeled training sample set containing samples of normal structural state and samples of potential structural anomalies. The labeled training sample set includes the structural stability level label corresponding to each time segment.

[0038] A time-series monitoring model is constructed, comprising the following structure: an input layer, a long short-term memory network layer for encoding time-series features of multi-dimensional IoT sensor data, a time-series attention layer for enhancing cross-time segment dependencies, a feature aggregation layer for dimensionality reduction and compression of time-series features, and a fully connected prediction layer for outputting stability scores.

[0039] A two-layer structure with 128 hidden units is set up in the Long Short-Term Memory (LSTM) network layer to capture the coupling relationship between the slow trend of tilt change and the instantaneous fluctuation characteristics of vibration. A multi-head attention mechanism is set up in the temporal attention layer to focus on key time segments in the sensor data that may characterize risk features such as structural fatigue, loosening, and wind displacement.

[0040] The time-series monitoring model to be trained is trained in a supervised manner using the labeled training sample set as input samples, and the mean squared error loss function is used to backpropagate the error between the predicted stability score and the actual structural stability level label.

[0041] During training, the Adam optimizer with a learning rate of 0.0005 was used to update the model parameters using gradients, and batch training with 64 time slices per batch was used to improve the model's adaptability to different climatic conditions, installation environment and physical load changes.

[0042] After training, the tilt data, vibration data and temperature and humidity data collected in real time are input into the time-series monitoring model in chronological order. The model generates a structural state time-series feature vector through a long short-term memory network layer and a time-series attention layer, and outputs a quantified stability score based on a fully connected prediction layer.

[0043] Among them, the stability score is used to characterize the structural stability of outdoor signs within a corresponding time window, and is used together with the defect score and spatial risk weight to determine the comprehensive risk score.

[0044] Furthermore, for S200-S300, the training samples for the video image recognition model and / or time-series monitoring model are obtained by data augmentation of the original sample data, wherein data augmentation includes at least one of the following processes: Perform at least one of the following on the signboard video frame image: brightness adjustment, rotation processing, and noise injection; Apply random fluctuations within a preset range to the sensor readings; Perform a neighborhood replacement operation on the spatial data of points of interest.

[0045] Optionally, the rotation processing angle range is ±5 degrees, the random fluctuation range is ±0.05 degrees for tilt data and ±0.005 times the gravitational acceleration for vibration data, and the neighborhood replacement operation replaces some features of the current signboard point of interest with features of other business district points of interest with similar pedestrian density attributes.

[0046] The replacement operation described above can be performed by obtaining the pedestrian density attribute corresponding to each sign point of interest from the map point of interest data. The pedestrian density attribute includes the average and peak pedestrian density calculated based on historical passenger flow statistics and / or mobile terminal trajectory data.

[0047] Based on the average pedestrian density, all signs of interest are divided into multiple preset pedestrian density intervals, and signs of interest within the same interval are marked as a set of candidate business district points of interest with similar pedestrian density.

[0048] For the current signboard point of interest, determine the pedestrian density range to which it belongs, and select at least one target commercial district point of interest from the candidate commercial district point of interest set with similar pedestrian density, based on geographical location similarity and point of interest type similarity. Geographical location similarity is calculated based on distance from the city center or major transportation hub, and point of interest type similarity is calculated based on matching category tags such as catering, retail, and entertainment venues.

[0049] Select at least one feature from the attribute information of the target business district's points of interest to describe the distribution of surrounding businesses, peak hours of pedestrian traffic, and density of parking facilities. Replace the corresponding feature of the current sign point of interest with at least one feature to form a data augmentation sample with neighborhood replacement.

[0050] Among them, the data augmentation samples after neighborhood replacement, while maintaining the basic consistency of the current signboard interest point pedestrian density attributes, introduce spatial feature changes in different business district environments to expand the diversity of the distribution patterns of sensitive interest points around outdoor signs in the training samples, so as to improve the generalization ability of the model trained based on map interest point data in different business district scenarios.

[0051] S400. Based on the spatial relationship between the target outdoor sign's point of interest and surrounding sensitive points of interest in the map's point of interest data, determine the spatial risk weight.

[0052] Specifically, this can involve obtaining the geographic coordinates of the signboard's point of interest, as well as the type and geographic coordinates of surrounding sensitive points of interest from map point of interest data.

[0053] Based on the geographic coordinates between the signboard's point of interest and each surrounding sensitive point of interest, the corresponding spatial distance is calculated. The spatial distance is calculated using at least one of the following: path distance based on road network point of interest information or Euclidean distance based on geographic coordinates.

[0054] Based on the type information of surrounding sensitive points of interest, the corresponding sensitivity level weight is determined from the preset sensitive point of interest risk level mapping table. The sensitivity level weight is used to characterize the risk level of different sensitive point of interest types such as densely populated areas, schools, bus stops, and open road sections.

[0055] Combining spatial distance and sensitivity level weights, a distance decay function is used to calculate the spatial risk contribution value of each surrounding sensitive interest point to the signboard interest point. The distance decay function is used to reduce the corresponding spatial risk contribution value as the distance increases.

[0056] The spatial risk contribution values ​​are weighted and summed to obtain the spatial risk weight used to characterize the sign's point of interest in its spatial environment.

[0057] Among them, the spatial risk weight is used to reflect the degree of potential accident risk of the target outdoor sign under different types of surrounding sensitive points of interest and different spatial distance conditions, and is used together with the defect score and stability score to determine the comprehensive risk score of the target outdoor sign.

[0058] S500 determines the overall risk score of the target outdoor sign based on defect scores, stability scores, and spatial risk weights.

[0059] Specifically, the defect score, stability score, and spatial risk weight are normalized to obtain normalized defect score, normalized stability score, and normalized spatial risk weight. The normalization process is used to map data with different dimensions to a unified numerical range.

[0060] Based on the preset risk weight coefficients, the normalized defect score, normalized stability score, and normalized spatial risk weight are weighted with their respective risk weight coefficients to obtain the defect risk item, stability risk item, and spatial risk item. The risk weight coefficients represent the degree of contribution of each score to the comprehensive risk assessment.

[0061] The defect risk item, stability risk item, and spatial risk item are weighted and summed to obtain a comprehensive risk score that characterizes the overall safety status of the target outdoor sign.

[0062] The comprehensive risk score is used to characterize the overall risk level of the target outdoor sign in terms of appearance damage, structural stability, and surrounding spatial environment, and is used to determine the scope of the accident impact and to send early warning information.

[0063] S600 determines the scope of an accident's impact based on a comprehensive risk score and the distribution of surrounding sensitive points of interest, and pushes out early warning information.

[0064] Specifically, when the surrounding sensitive points of interest include points of interest in densely populated areas, the preset influence range will be expanded to the first range; when the surrounding sensitive points of interest include points of interest in open road sections, the preset influence range will be reduced to the second range.

[0065] Specifically, expanding the preset impact range to the first range can be achieved by acquiring the category attributes and geographic coordinates of points of interest in densely populated areas. Based on the spatial distance between these points of interest and the signboard points of interest, a corresponding impact factor for densely populated areas is determined. This impact factor characterizes the amplification effect of densely populated areas on the spread of accidents. Then, the preset impact range is scaled according to the impact factor, expanding radially around the signboard points of interest to obtain the first range. This first range reflects situations where the accident propagation path is more complex and the impact on the population is higher near densely populated areas.

[0066] To narrow the preset impact range to the second range, one can obtain the category attributes and geographic coordinate information of the points of interest in the open road section. Then, based on the spatial distance between the points of interest in the open road section and the points of interest on the sign, a mitigation factor for the open area is determined. The mitigation factor is used to characterize the attenuation effect of the open area on the degree of impact of the accident.

[0067] Next, the preset impact range is scaled down according to the mitigation factor, causing the preset impact range to shrink radially around the point of interest of the sign, resulting in a second range. The second range is used to reflect the situation where the open area is highly accessible, there are few obstacles, and the resistance to accident propagation is low, making the impact prediction more consistent with the actual environmental conditions.

[0068] In addition, for pushing early warning information, the early warning information can be pushed to the corresponding processing terminal based on the preset risk level of the comprehensive risk score. The processing terminal includes at least one of the following: urban management law enforcement terminal, street office terminal, and merchant terminal.

[0069] For example, when the comprehensive risk score reaches the highest risk level (e.g., red), it indicates an extremely high and imminent threat to public safety. In this case, the warning information will be simultaneously pushed to the urban management enforcement terminal, the street office terminal, and the merchant's terminal, requiring the activation of the emergency response procedure. For lower risk levels (e.g., blue), the warning may only be pushed to the merchant's terminal, prompting them to conduct their scheduled inspections and maintenance. This mechanism of routing warning information to different responsible entities based on risk level ensures highly targeted warning information, clearly delineates the boundaries of responsibility between government regulation and merchants, and while ensuring strong control over high-risk situations, also reduces unnecessary consumption of administrative resources and excessive disruption to merchants.

[0070] In this embodiment, video surveillance data, IoT sensor data, and map point-of-interest (POI) data of the target outdoor sign are acquired. Based on the video surveillance data, a defect score representing the severity of the target outdoor sign's appearance defects is determined. Then, based on the IoT sensor data, a stability score representing the structural stability of the target outdoor sign is determined. Next, based on the spatial relationship between the sign's POI and surrounding sensitive POIs in the map POI data, a spatial risk weight is determined. Based on the defect score, stability score, and spatial risk weight, a comprehensive risk score for the target outdoor sign is calculated. Finally, based on the comprehensive risk score and the distribution of surrounding sensitive POIs, the scope of the accident's impact is determined, and early warning information is pushed out. This effectively predicts safety issues of surrounding outdoor signs and achieves efficient safety monitoring and early warning.

[0071] To facilitate understanding, an example can be taken of a large LED outdoor sign installed outside a shop in a bustling commercial district to illustrate the specific implementation of the above embodiments. To continuously monitor the operational status of the sign, the outdoor sign safety management and monitoring method provided in this application, after actual deployment, begins to collect data and conduct risk assessments of the sign around the clock.

[0072] During system operation, a camera installed directly in front of the sign first acquires real-time video monitoring data. Simultaneously, IoT sensors fixed to the main structure of the sign periodically upload data on tilt, vibration, and temperature and humidity. In addition, the system retrieves information on points of interest corresponding to the sign, surrounding sensitive points of interest (such as densely populated areas, bus stops, schools, and open road sections), and the spatial distribution information of road network points of interest from the city map database.

[0073] Subsequently, based on the collected video surveillance data, the system identifies whether there are appearance defects on the signboard surface, such as cracks, peeling, and damaged light panels. The video data is input into a pre-trained video image recognition model, which automatically extracts appearance defect features from the video and outputs a defect score to characterize the current degree of damage to the signboard.

[0074] Meanwhile, multiple IoT sensors on the sign continuously report changes in its tilt, vibration intensity, and temperature and humidity. This sensor data is input into a pre-trained time-series monitoring model. By encoding and analyzing the temporal trends of the sensor data, the model generates a stability score to reflect whether the sign structure exhibits abnormal fluctuations, uneven stress, or potential risk of detachment during that period.

[0075] While determining the structural status, the system further analyzes the risk level of the spatial environment in which the sign is located. To this end, the system reads the spatial relationship between the sign's point of interest and surrounding sensitive points of interest, such as the distance between the sign and densely populated areas, and whether there are schools or open road sections nearby. Based on the type, density, and distance of these sensitive points of interest, the system calculates spatial risk weights, so that the potential risks of the sign in high-density areas are given higher weights, while those in open areas are correspondingly reduced.

[0076] After obtaining the defect score, stability score, and spatial risk weight, the system uses a preset weighted calculation strategy to fuse these scores, obtaining the sign's comprehensive risk score at the current moment. The comprehensive risk score can intuitively reflect the sign's overall risk level in terms of appearance damage, structural stability, and the safety of its surrounding environment.

[0077] After calculating the overall risk score, the system will automatically generate the scope of the accident's impact based on the distribution of surrounding sensitive points of interest. For example, if there are densely populated areas of interest around the sign, the system will expand the preset scope of impact to a larger area; if the surrounding area is mainly open road sections, the system will appropriately reduce the scope of impact to reflect the actual risk performance.

[0078] Finally, based on the preset risk level corresponding to the comprehensive risk score, the system will automatically push the generated early warning information to the corresponding handling terminals, including urban management law enforcement terminals, street office terminals, and merchant terminals, so that management departments can take timely inspection, repair, or closure measures, thereby effectively reducing the probability of sign falling accidents and ensuring public safety.

[0079] Figure 2 This is a schematic diagram illustrating the structure of an outdoor sign safety management and monitoring system according to an example embodiment of this application. Figure 2 As shown, the outdoor sign safety management and monitoring system 700 provided in this embodiment includes: The acquisition module 710 is used to acquire video surveillance data, IoT sensor data, and map point of interest data of the target outdoor sign. The determination module 720 is used to determine a defect score that characterizes the severity of the appearance defects of the target outdoor sign based on the video surveillance data; The determining module 720 is further configured to determine a stability score characterizing the stability of the target outdoor sign structure based on the IoT sensor data; The determining module 720 is further configured to determine the spatial risk weight based on the spatial relationship between the sign interest point corresponding to the target outdoor sign and the surrounding sensitive interest points in the map interest point data. The determining module 720 is further configured to determine the comprehensive risk score of the target outdoor sign based on the defect score, the stability score, and the spatial risk weight; The push module 730 is also used to determine the scope of the accident's impact based on the comprehensive risk score and the distribution of the surrounding sensitive points of interest, and to push early warning information.

[0080] Figure 3 This is a schematic diagram of the structure of an electronic device according to an example embodiment of this application. For example... Figure 3 As shown, the electronic device 800 provided in this embodiment includes: a processor 801 and a memory 802; wherein: The memory 802 is used to store computer programs, and the memory can also be flash memory.

[0081] Processor 801 is used to execute the execution instructions stored in memory to implement the various steps in the above method. For details, please refer to the relevant descriptions in the preceding method embodiments.

[0082] Alternatively, the memory 802 can be either standalone or integrated with the processor 801.

[0083] When the memory 802 is a device independent of the processor 801, the electronic device 800 may further include: Bus 803 is used to connect the memory 802 and the processor 801.

[0084] This embodiment also provides a readable storage medium storing a computer program, which, when executed by at least one processor of an electronic device, enables the electronic device to perform the methods provided in the various embodiments described above.

[0085] This embodiment also provides a program product including a computer program stored in a readable storage medium. At least one processor of an electronic device can read the computer program from the readable storage medium, and the at least one processor executes the computer program to cause the electronic device to perform the methods provided in the various embodiments described above.

[0086] Other embodiments of this application will readily occur to those skilled in the art upon consideration of the specification and practice of the invention disclosed herein. This application is intended to cover any variations, uses, or adaptations of this application that follow the general principles of this application and include common knowledge or customary techniques in the art not disclosed herein. The specification and examples are to be considered exemplary only, and the true scope and spirit of this application are indicated by the claims.

[0087] It should be understood that this application is not limited to the precise structure described above and shown in the accompanying drawings, and various modifications and changes can be made without departing from its scope. The scope of this application is limited only by the appended claims.

Claims

1. A method for safety management and monitoring of outdoor signs, characterized in that, include: Acquire video surveillance data, IoT sensor data, and map point-of-interest data of the target outdoor signboard; Based on the video surveillance data, a defect score is determined to characterize the severity of the appearance defects of the target outdoor sign. Based on the IoT sensor data, a stability score characterizing the structural stability of the target outdoor signboard is determined; Based on the spatial relationship between the sign interest point corresponding to the target outdoor sign and the surrounding sensitive interest points in the map interest point data, the spatial risk weight is determined. The comprehensive risk score of the target outdoor sign is determined based on the defect score, the stability score, and the spatial risk weight. Based on the comprehensive risk score and the distribution of surrounding sensitive points of interest, the scope of the accident's impact is determined, and early warning information is pushed out.

2. The method according to claim 1, characterized in that, The determination of a defect score, characterizing the severity of the appearance defects of the target outdoor sign based on the video surveillance data, includes: The video surveillance data is input into a pre-trained video image recognition model to obtain the defect score. The video image recognition model is used to extract appearance defect features from the video surveillance data and generate the defect score based on the appearance defect features.

3. The method according to claim 2, characterized in that, The determination of a stability score characterizing the structural stability of the target outdoor signboard based on the IoT sensor data includes: The IoT sensor data is input into a pre-trained time-series monitoring model to obtain the stability score. The time-series monitoring model is used to perform time-series encoding on the IoT sensor data and generate the stability score based on the encoding results.

4. The method according to claim 3, characterized in that, The training samples for the video image recognition model and / or the time-series monitoring model are obtained by data augmentation of the original sample data, wherein the data augmentation includes at least one of the following processes: Perform at least one of the following on the signboard video frame image: brightness adjustment, rotation processing, and noise injection; Apply random fluctuations within a preset range to the sensor readings; Perform a neighborhood replacement operation on the spatial data of points of interest.

5. The method according to claim 4, characterized in that, The rotation angle range is ±5 degrees; The range of the random fluctuations is ±0.05 degrees for tilt data and ±0.005 times the gravitational acceleration for vibration data; The neighborhood replacement operation involves replacing some features of the current signboard point of interest with features of other business district points of interest that have similar pedestrian density attributes.

6. The method according to claim 1, characterized in that, The determination of the accident impact range based on the comprehensive risk score and the distribution of surrounding sensitive points of interest includes: When the surrounding sensitive points of interest include points of interest in densely populated areas, the preset influence range will be expanded to the first range; When the surrounding sensitive points of interest include points of interest in open road sections, the preset influence range is reduced to a second range.

7. The method according to claim 1, characterized in that, The IoT sensor data includes tilt data, vibration data, and temperature and humidity data; The map point of interest data includes attribute information of signboard points of interest, type information of surrounding sensitive points of interest, and road network point of interest information.

8. The method according to claim 1, characterized in that, The push notification information includes: Based on the preset risk level to which the comprehensive risk score belongs, the early warning information is pushed to the processing terminal corresponding to the preset risk level. The processing terminal includes at least one of urban management law enforcement terminal, street office terminal and merchant terminal.

9. An outdoor signboard safety management and monitoring system, characterized in that, include: The acquisition module is used to acquire video surveillance data, IoT sensor data, and map point of interest data of the target outdoor sign. The determination module is used to determine a defect score that characterizes the severity of the appearance defects of the target outdoor sign based on the video surveillance data. The determining module is also used to determine a stability score characterizing the stability of the target outdoor sign structure based on the IoT sensor data; The determining module is further configured to determine the spatial risk weight based on the spatial relationship between the sign interest point corresponding to the target outdoor sign and the surrounding sensitive interest points in the map interest point data. The determining module is further configured to determine the comprehensive risk score of the target outdoor sign based on the defect score, the stability score, and the spatial risk weight. The push module is also used to determine the scope of the accident's impact based on the comprehensive risk score and the distribution of the surrounding sensitive points of interest, and to push early warning information.

10. An electronic device, characterized in that, include: processor; as well as, Memory for storing the executable instructions of the processor; The processor is configured to execute the method of any one of claims 1 to 8 by executing the executable instructions.

Images