A management system for pedestrian-aware smart lighting

Abstract

The present invention discloses a management system for pedestrian-aware smart lighting, which specifically relates to the field of sensor technology, including a regional element extraction module, a signal recognition module, an anomaly detection module, and an upload processing module. The regional element extraction module extracts feature information from the sensor and transmits it to the signal recognition module. The signal recognition module classifies and identifies pedestrians in the detection area, uses a target detection algorithm to detect and identify human targets in the detection area, marks their behavior patterns and motion characteristics and classifies and marks them. The anomaly detection module detects abnormal behavior signals, and when there is an abnormality in the marked pedestrian target behavior and there is uncertainty in the detection reliability, an abnormal signal is issued. The upload processing module is used to save the detection data, and makes a manual access operation judgment based on the anomaly detection result, and uploads the collected information to a cloud server. The present invention can effectively integrate lighting and public safety elements, assist in maintaining public stability, and reduce the burden on staff.

Description

Technical Field

[0001] The present invention relates to the field of sensor technology, and more specifically, to a management system for pedestrian-aware smart lighting. Background Art

[0002] The pedestrian-aware smart lighting management system can use sensing technology and intelligent control algorithms to realize the intelligent lighting of urban streets, public areas and buildings, and provide efficient and safe public services by real-time monitoring and analysis of changes in the surrounding environment.

[0003] Public safety services must take into account both openness and concealment. Public displays of public safety services can effectively deter potential dangerous elements, while concealed public safety services can effectively target dangerous elements with a certain degree of anti-detection awareness. Integrating smart lighting with security monitoring can avoid causing anxiety among citizens and effectively expand the scope and depth of public safety services. Using deep learning algorithms to intelligently upgrade lighting systems to assist in maintaining public safety has become a new challenge.

[0004] In order to solve the above defects, a technical solution is now proposed. Summary of the invention

[0005] The purpose of the present invention is to provide a management system for pedestrian-aware smart lighting to solve the problems in the above-mentioned background technology.

[0006] To achieve the above-mentioned purpose, the present invention provides the following technical solutions: a management system for pedestrian-aware smart lighting, comprising a regional element extraction module, a signal recognition module, an anomaly detection module, and an upload processing module;

[0007] The regional feature extraction module is used to extract feature information from the sensor, provide necessary data for analysis and modeling, and transmit the feature information to the signal recognition module;

[0008] The signal recognition module is used to classify and identify pedestrians in the detection area, mark human targets in the detection area, use the target detection algorithm to detect and identify targets in the detection area, mark their behavior patterns and motion characteristics, and classify and mark them;

[0009] The anomaly detection module is used to detect abnormal behavior signals. When the behavior of the marked pedestrian target is abnormal and there is uncertainty in the detection reliability, an abnormal signal is issued;

[0010] The upload processing module is used to save the regional element signals and anomaly detection signals obtained by the acquisition equipment, make manual access operation decisions based on the anomaly detection results, and upload the collected information to the cloud server.

[0011] Preferably, the method for detecting anomalies of pedestrian targets is:

[0012] S1. Collect video data containing pedestrian targets and annotate them as training sets. During the annotation process, ensure that each pedestrian target has a corresponding bounding box annotation;

[0013] S2. In the YOLOv7 code base, make sure that the configuration file correctly points to the training dataset path and the number of categories, and pedestrian targets are one of the categories;

[0014] S3. Use the training script to train the model, set the configuration file path, weight saving path, etc.;

[0015] S4. Evaluate the trained model through the validation set, and adjust the hyperparameters, model structure, etc. according to the evaluation results to improve the detection performance;

[0016] S5. Use the trained model to perform target detection on the video signal and mark the targets detected as pedestrians;

[0017] S6. Use deep learning models to detect key feature points of pedestrian targets, which usually include key positions such as head, shoulders, hands, and feet, and classify and annotate pedestrian behavior patterns;

[0018] S7. Acquire environmental status information through sensors, and comprehensively evaluate the environmental status information and pedestrian behavior pattern information to obtain a status signal.

[0019] Preferably, the logic for comprehensively evaluating the environmental status information and the pedestrian behavior pattern information is:

[0020] The information recognition module obtains the behavior deviation coefficient, temperature fluctuation coefficient, and pedestrian retention coefficient of the detection area, and calibrates the behavior deviation coefficient as Bd, the temperature fluctuation coefficient as Tf, and the pedestrian retention coefficient as Ps, and calculates the sensitivity index, which is expressed as follows: In the formula, Se is the sensitivity index, α and β are the proportional coefficients of temperature fluctuation coefficient and pedestrian retention coefficient, respectively, and α and β are both greater than 0;

[0021] The behavior deviation coefficient is the movement state of pedestrians in the detection area of ​​the pedestrian perception smart lighting management system when the target detected as a pedestrian enters. When the deep learning model detects that the movement state is a high-risk state, Bd takes 1, and when the deep learning model detects that the movement state is a low-risk state, Bd takes 0;

[0022] The high-risk state is when the deep learning model determines that the pedestrian target is in the categories of human disability, fighting, and mass demonstrations, and the low-risk state is other categories;

[0023] The temperature fluctuation coefficient is the degree of decrease in the ambient temperature of the lighting source area, and its calculation expression is: Where, T(k) is the change of ambient temperature over time, [t1, t2 is the time period that the pedestrian target exists in the detection area;

[0024] The pedestrian retention coefficient is the change in the density of pedestrian flow in the detection area, and its calculation expression is: Where P(c) is the change in the number of pedestrian targets in the detection area over time, and t3,t4] is the anomaly detection activation time period.

[0025] Preferably, the method for verifying the detection reliability is:

[0026] A preset reference value is set for the comprehensive evaluation index of the alarm accuracy of the pedestrian perception smart lighting management system within the T time, and the preset reference value of the comprehensive evaluation index is calibrated as Re, and Re is greater than 1;

[0027] The preset reference value of the comprehensive evaluation index is a specific reference value after quantification, which is not specifically limited here. Re is a value greater than 1. The comprehensive evaluation index, that is, the F value, has a value range between 0 and 1, which can comprehensively consider the indicators of matching accuracy P and recall rate R, and is used to measure the judgment ability of the pedestrian perception smart lighting management system for abnormal detection. A high F value means that the pedestrian perception smart lighting management system can more accurately identify abnormal states and avoid misjudging normal states as abnormal states, which helps to ensure that the pedestrian perception smart lighting management system detects and processes abnormal states more accurately and improves the detection effect. A high F value means that the pedestrian perception smart lighting management system has a high recall rate for abnormal states, which can effectively reduce errors and omissions to extract the real abnormal state;

[0028] Obtain the comprehensive evaluation index of the management system for pedestrian perception smart lighting at different time periods within the time T, and mark the comprehensive evaluation index as Tg, where g represents the number of the comprehensive evaluation index of the management system for pedestrian perception smart lighting at different time periods within the time T, and g=1,2,3…r}, where r is a positive integer;

[0029] The expression for calculating the comprehensive evaluation index is: In the formula, P represents the matching accuracy rate, which refers to the proportion of abnormal states that are judged as abnormal states by the management system for pedestrian perception smart lighting, and is actually abnormal. The calculation method is: matching accuracy rate = number of abnormal states correctly matched / number of all abnormal states judged as abnormal states; R represents the recall rate, which refers to the ratio between the number of abnormal states correctly matched by the management system for pedestrian perception smart lighting and the number of all actual abnormal states. The calculation method is: recall rate = number of abnormal states correctly matched / number of all actual abnormal states;

[0030] The number of all abnormal states refers to the number of all abnormal states determined by the pedestrian perception smart lighting management system within a certain time range, including real abnormal states and misjudged normal states. The pedestrian perception smart lighting management system usually classifies and marks the detection states, or records the abnormal judgment of transactions in the log, which can be obtained by counting the number of all states marked as abnormal;

[0031] The number of all actual abnormal states refers to the number of all abnormal states that exist in actual situations. Since the actual abnormal state may be caused by detection and judgment errors, the management system for pedestrian perception smart lighting needs to interact with the terminal server or database to obtain the number of actual abnormal states;

[0032] The expression for calculating the anomaly detection efficiency coefficient is: Re is the preset reference value of the comprehensive evaluation index.

[0033] Preferably, the method for comprehensively evaluating the pedestrian target state and detection reliability is:

[0034] Based on the comprehensive analysis of the sensitivity index and the anomaly detection efficiency coefficient, a monitoring model is established to calculate the reliability index of the management system for pedestrian perception smart lighting. The calculation expression of the reliability index is: In the formula, γ and δ are the proportional coefficients of the sensitivity index and the anomaly detection efficiency coefficient, respectively, and both γ and δ are greater than 0.

[0035] Preferably, the logic for determining the manual access operation based on the abnormal detection result is:

[0036] The calculated reliability index of the management system for pedestrian perception smart lighting is compared with the preset reliability index threshold. If the calculated reliability index is greater than or equal to the preset reliability index, a suspicious signal is generated. If the calculated reliability index is less than the preset reliability index, a steady-state signal is generated.

[0037] Perform processing strategy analysis based on the generated suspicious signal. When the generated suspicious signal is received, a data set is generated based on the continuous reliability index data of the pedestrian perception smart lighting management system within T time after the suspicious signal is generated, and the reliability index in the data set is calibrated as Rv, where v is the reliability index number, that is, v = {1, 2, 3...x}, where x is a positive integer;

[0038] Calculate the standard deviation of several reliability pointing indexes in the data set, and calibrate the standard deviation of the reliability pointing index as So, and compare the reliability pointing index standard deviation So with the preset reliability pointing index standard deviation threshold Do, and perform early warning processing based on the comparison results. The processing logic is as follows:

[0039] If So is greater than or equal to Do, the pedestrian perception smart lighting management system is marked as a high-risk level, and the staff is reminded that the pedestrian perception smart lighting management system has unreliable risk hazards and needs to be inspected and maintained;

[0040] If So is less than Do, the pedestrian perception smart lighting management system is marked as a low-risk level, reminding the staff that the pedestrian perception smart lighting management system has low-risk hidden dangers and does not require inspection and maintenance.

[0041] In the above technical solution, the technical effects and advantages provided by the present invention are:

[0042] The present invention detects the reliability pointing index of the management system for pedestrian perception smart lighting. When it is found that the detection reliability is abnormal under abnormal circumstances, a comprehensive analysis is conducted on the subsequent operation status of the management system for pedestrian perception smart lighting, and the abnormal hidden dangers are determined and early warning prompts are issued. This can effectively assist the staff to obtain the potential fault status, check the potential faults in advance, prevent the operation of the management system from entering a failure state, and can reduce the workload of the staff and improve work efficiency. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments recorded in the present invention. For ordinary technicians in this field, other drawings can also be obtained based on these drawings.

[0044] Figure 1 It is a system module diagram of the present invention. DETAILED DESCRIPTION

[0045] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0046] Example 1

[0047] See also Figure 1As shown, the present invention is a management system for pedestrian-aware smart lighting, including a regional element extraction module, a signal recognition module, an anomaly detection module, and an upload processing module;

[0048] The regional feature extraction module is used to extract feature information from the sensor, provide necessary data for analysis and modeling, and transmit the feature information to the signal recognition module;

[0049] The sensor used in the management system for pedestrian-aware smart lighting can be a video stream acquisition device, and the video stream acquisition device can be a composite camera with infrared sensor, optical camera, and microwave sensor functions;

[0050] First, the system needs to establish a connection with the video stream acquisition device and obtain the parameters of the device, including resolution, frame rate, exposure, etc., to ensure that the acquired video data meets the system requirements. When the acquisition device establishes a connection, the video acquisition module starts to acquire real-time video streams through video capture cards, network cameras, camera APIs, etc. Before passing the video stream to other modules, some preprocessing steps may be required to improve the effect of subsequent analysis, which may include image denoising, color correction, image stabilization, etc. In order to reduce the cost of data transmission and storage, the video stream may need to be compressed and encoded. Common compression algorithms include H.264, H.265, etc. The encoded video data can be transmitted to the signal recognition module more efficiently;

[0051] The compressed and encoded video data is transmitted to the signal recognition module, which is responsible for detecting the target in the video and identifying its motion pattern. The data transmission can be completed through the network communication protocol, such as RTSP or other real-time transmission protocols. The compressed and encoded video data can also be transmitted to the upload processing module for long-term storage and subsequent analysis. The storage can be local storage or cloud storage.

[0052] The signal recognition module is used to classify and identify pedestrians in the detection area, mark human targets in the detection area, use the target detection algorithm to detect and identify targets in the detection area, mark their behavior patterns and motion characteristics, and classify and mark them;

[0053] Target detection algorithms include but are not limited to YOLO algorithm, SSD algorithm, Faster R-CNN algorithm, etc., which analyze and detect the motion information and state information of pedestrian targets;

[0054] The anomaly detection module is used to detect abnormal behavior signals. When the behavior of the marked pedestrian target is abnormal, an abnormal signal is issued;

[0055] Abnormal behavior includes abnormal conditions such as human disability, fighting, and mass demonstrations;

[0056] The upload processing module is used to save the regional element signals and anomaly detection signals obtained by the acquisition equipment, perform manual access operations according to the anomaly detection results, and upload the collected information to the cloud server.

[0057] Example 2

[0058] Pedestrian target anomaly detection is obtained by analyzing the pedestrian feature extraction information. The implementation process is as follows:

[0059] S1. Collect video data containing pedestrian targets and annotate them as training sets. During the annotation process, ensure that each pedestrian target has a corresponding bounding box annotation;

[0060] S2. In the YOLOv7 code base, make sure that the configuration file correctly points to the training dataset path and the number of categories, and pedestrian targets are one of the categories;

[0061] S3. Use the training script to train the model, set the configuration file path, weight saving path, etc.;

[0062] S4. Evaluate the trained model through the validation set, and adjust the hyperparameters, model structure, etc. according to the evaluation results to improve the detection performance;

[0063] S5. Use the trained model to perform target detection on the video signal and mark the targets detected as pedestrians;

[0064] S6. Use deep learning models to detect key feature points of pedestrian targets, which usually include key positions such as head, shoulders, hands, and feet, and classify and annotate pedestrian behavior patterns;

[0065] Select an appropriate posture estimation model. Common models include OpenPose and PoseNet. These models can convert the detected human key points into specific posture information. Use the selected posture estimation model to process the human key points and estimate the target's behavior. The behavior information includes movement direction, speed, limb state, etc., and classify the action through a classifier.

[0066] S7. Acquire environmental status information through sensors, and comprehensively evaluate the environmental status information and pedestrian behavior pattern information to obtain a status signal.

[0067] The anomaly detection module obtains the behavior deviation coefficient, temperature fluctuation coefficient, and pedestrian retention coefficient of the detection area, and calibrates the behavior deviation coefficient as Bd, the temperature fluctuation coefficient as Tf, and the pedestrian retention coefficient as Ps, and calculates the sensitivity index, which is expressed as follows: In the formula, α and β are the proportional coefficients of temperature fluctuation coefficient and pedestrian retention coefficient, respectively, and both α and β are greater than 0;

[0068] The behavior deviation coefficient is the movement state of pedestrians in the detection area of ​​the pedestrian perception smart lighting management system when the target detected as a pedestrian enters. When the deep learning model detects that the movement state is a high-risk state, Bd takes 1, and when the deep learning model detects that the movement state is a low-risk state, Bd takes 0;

[0069] The high-risk state is when the deep learning model determines that the pedestrian target is in the categories of human disability, fighting, and mass demonstrations, and the low-risk state is other categories;

[0070] The temperature fluctuation coefficient is the degree of decrease in the ambient temperature of the lighting source area, and its calculation expression is: Where, T(k) is the change of ambient temperature over time, [t1, t2 is the time period that the pedestrian target exists in the detection area;

[0071] The pedestrian retention coefficient is the change in the density of pedestrian flow in the detection area, and its calculation expression is: Where P(c) is the change in the number of pedestrian targets in the detection area over time, and t3,t4] is the anomaly detection activation time period.

[0072] Example 3

[0073] The pedestrian perception intelligent lighting management system obtains abnormal signals and performs judgment operations to evaluate the judgment effect. The judgment effect is expressed by the F value. The specific calculation process is as follows:

[0074] A preset reference value is set for the comprehensive evaluation index of the alarm accuracy of the pedestrian perception smart lighting management system within the T time, and the preset reference value of the comprehensive evaluation index is calibrated as Re, and Re is greater than 1;

[0075] It should be noted that the preset reference value of the comprehensive evaluation index is a specific reference value after quantification, which is not specifically limited here. Re is a value greater than 1. The comprehensive evaluation index, that is, the F value, has a value range between 0 and 1, which can comprehensively consider the indicators of matching accuracy P and recall rate R, and is used to measure the judgment ability of the pedestrian perception smart lighting management system for abnormal detection. A high F value means that the pedestrian perception smart lighting management system can more accurately identify abnormal states and avoid misjudging normal states as abnormal states, which helps to ensure that the pedestrian perception smart lighting management system detects and processes abnormal states more accurately and improves the detection effect. A high F value means that the pedestrian perception smart lighting management system has a high recall rate for abnormal states, which can effectively reduce errors and omissions to extract the real abnormal state;

[0076] Obtain the comprehensive evaluation index of the management system for pedestrian perception smart lighting at different time periods within the time T, and mark the comprehensive evaluation index as Tg, where g represents the number of the comprehensive evaluation index of the management system for pedestrian perception smart lighting at different time periods within the time T, and g=1,2,3…r}, where r is a positive integer;

[0077] The expression for calculating the comprehensive evaluation index is: In the formula, P represents the matching accuracy rate, which refers to the proportion of abnormal states that are judged as abnormal states by the management system for pedestrian perception smart lighting, and is actually abnormal. The calculation method is: matching accuracy rate = number of abnormal states correctly matched / number of all abnormal states judged as abnormal states; R represents the recall rate, which refers to the ratio between the number of abnormal states correctly matched by the management system for pedestrian perception smart lighting and the number of all actual abnormal states. The calculation method is: recall rate = number of abnormal states correctly matched / number of all actual abnormal states;

[0078] It should be noted that the number of correctly matched abnormal states refers to the number of abnormal states correctly identified and judged by the pedestrian perception smart lighting management system. The pedestrian perception smart lighting management system will mark the detected abnormal states or put them into a database of abnormal state records. The number of correctly judged abnormal states can be obtained by querying the number of abnormal state records;

[0079] The number of all abnormal states refers to the number of all abnormal states determined by the pedestrian perception smart lighting management system within a certain time range, including real abnormal states and misjudged normal states. The pedestrian perception smart lighting management system usually classifies and marks the detection states, or records the abnormal judgment of transactions in the log, which can be obtained by counting the number of all states marked as abnormal;

[0080] The number of all actual abnormal states refers to the number of all abnormal states that exist in actual situations. Since the actual abnormal state may be caused by detection and judgment errors, the management system for pedestrian perception smart lighting needs to interact with the terminal server or database to obtain the number of actual abnormal states;

[0081] The expression for calculating the anomaly detection efficiency coefficient is

[0082] According to the calculation expression of the anomaly detection efficiency coefficient, it can be seen that the larger the anomaly detection efficiency coefficient of the pedestrian perception smart lighting management system is, the worse the credibility of the pedestrian perception smart lighting management system is. Conversely, the smaller the anomaly detection efficiency coefficient of the pedestrian perception smart lighting management system is, the better the credibility of the pedestrian perception smart lighting management system is.

[0083] Based on the comprehensive analysis of the sensitivity index and the anomaly detection efficiency coefficient, a monitoring model is established to calculate the reliability index of the management system for pedestrian perception smart lighting. The calculation expression of the reliability index is: In the formula, γ and δ are the proportional coefficients of the sensitivity index and the anomaly detection efficiency coefficient, respectively, and both γ and δ are greater than 0;

[0084] The calculated reliability index of the management system for pedestrian perception smart lighting is compared with the preset reliability index threshold. If the calculated reliability index is greater than or equal to the preset reliability index, a suspicious signal is generated. If the calculated reliability index is less than the preset reliability index, a steady-state signal is generated.

[0085] Perform processing strategy analysis based on the generated suspicious signal. When the generated suspicious signal is received, a data set is generated based on the continuous reliability index data of the pedestrian perception smart lighting management system within T time after the suspicious signal is generated, and the reliability index in the data set is calibrated as Rv, where v is the reliability index number, that is, v = {1, 2, 3...x}, where x is a positive integer;

[0086] Calculate the standard deviation of several reliability pointing indexes in the data set, and calibrate the standard deviation of the reliability pointing index as So, and compare the reliability pointing index standard deviation So with the preset reliability pointing index standard deviation threshold Do, and perform early warning processing based on the comparison results. The processing logic is as follows:

[0087] If So is greater than or equal to Do, the pedestrian perception smart lighting management system is marked as a high-risk level, and the staff is reminded that the pedestrian perception smart lighting management system has unreliable risk hazards and needs to be inspected and maintained;

[0088] If So is less than Do, the pedestrian perception smart lighting management system is marked as a low-risk level, reminding the staff that the pedestrian perception smart lighting management system has low-risk hidden dangers and does not require inspection and maintenance.

[0089] The present invention detects the reliability pointing index of the management system for pedestrian perception smart lighting. When it is found that the detection reliability is abnormal under abnormal circumstances, a comprehensive analysis is conducted on the subsequent operation status of the management system for pedestrian perception smart lighting, and the abnormal hidden dangers are determined and early warning prompts are issued. This can effectively assist the staff to obtain the potential fault status, check the potential faults in advance, prevent the operation of the management system from entering a failure state, and can reduce the workload of the staff and improve work efficiency.

[0090] The above formulas are all dimensionless and numerical calculations. The formula is a formula for the most recent real situation obtained by collecting a large amount of data and performing software simulation. The preset parameters in the formula are set by technicians in this field according to actual conditions.

[0091] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of computer program goods. The computer program goods include one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired or wireless (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that can be accessed by a computer or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.

[0092] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean 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 the present application.

[0093] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.

[0094] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the system described above can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0095] If the functions are implemented in the form of software functional units and sold or used as independent goods, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of software goods, which are stored in a storage medium and include several instructions for a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the methods described in each embodiment of the present application. The aforementioned storage medium includes: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.

[0096] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.

Claims

1. A management system for pedestrian-aware smart lighting, characterized in that: It includes regional element extraction module, signal recognition module, anomaly detection module and upload processing module; The regional feature extraction module is used to extract feature information from the sensor, provide necessary data for analysis and modeling, and transmit the feature information to the signal recognition module; The signal recognition module is used to classify and identify pedestrians in the detection area, mark human targets in the detection area, use the target detection algorithm to detect and identify targets in the detection area, mark their behavior patterns and motion characteristics, and classify and mark them; The anomaly detection module is used to detect abnormal behavior signals. When the behavior of the marked pedestrian target is abnormal and there is uncertainty in the detection reliability, an abnormal signal is issued; The upload processing module is used to save the regional element signals and anomaly detection signals acquired by the acquisition equipment, make manual access operation decisions based on the anomaly detection results, and upload the collected information to the cloud server; The method for detecting anomalies of pedestrian targets is: S1. Collect video data containing pedestrian targets and annotate them as training sets. During the annotation process, ensure that each pedestrian target has a corresponding bounding box annotation; S2. In the YOLOv7 code base, make sure that the configuration file correctly points to the training dataset path and the number of categories, and pedestrian targets are one of the categories; S3. Use the training script to train the model and set the configuration file path and weight saving path; S4. Evaluate the trained model through the validation set, and adjust the hyperparameters and model structure according to the evaluation results to improve the detection performance; S5. Use the trained model to perform target detection on the video signal and mark the targets detected as pedestrians; S6. Use deep learning models to detect key feature points of pedestrian targets, including key positions of the head, shoulders, hands, and feet, and classify and annotate pedestrian behavior patterns; S7, obtaining environmental status information through sensors, and comprehensively evaluating the environmental status information and pedestrian behavior pattern information to obtain a status signal; The logic of comprehensive evaluation of environmental status information and pedestrian behavior pattern information is: The information recognition module obtains the behavior deviation coefficient, temperature fluctuation coefficient, and pedestrian retention coefficient of the detection area, and calibrates the behavior deviation coefficient as Bd, the temperature fluctuation coefficient as Tf, and the pedestrian retention coefficient as Ps, and calculates the sensitivity index, which is expressed as follows: In the formula, Se is the sensitivity index, α and β are the proportional coefficients of temperature fluctuation coefficient and pedestrian retention coefficient, respectively, and α and β are both greater than 0; The behavior deviation coefficient is the movement state of pedestrians in the detection area of ​​the pedestrian perception smart lighting management system when the target detected as a pedestrian enters. When the deep learning model detects that the movement state is a high-risk state, Bd takes 1, and when the deep learning model detects that the movement state is a low-risk state, Bd takes 0; The high-risk state is when the deep learning model determines that the pedestrian target is in the categories of human disability, fighting, and mass demonstrations, and the low-risk state is other categories; The temperature fluctuation coefficient is the degree of decrease in the ambient temperature of the lighting source area, and its calculation expression is: Where T(k) is the change of ambient temperature over time, and [t1, t2] is the time period that the pedestrian target exists in the detection area; The pedestrian retention coefficient is the change in the density of pedestrian flow in the detection area, and its calculation expression is: Where P(c) is the change in the number of pedestrian targets in the detection area over time, and [t3, t4] is the activation time period of anomaly detection; The method to verify the reliability of the test is: A preset reference value is set for the comprehensive evaluation index of the alarm accuracy of the pedestrian perception smart lighting management system within the T time, and the preset reference value of the comprehensive evaluation index is calibrated as Re, and Re is greater than 1; The preset reference value of the comprehensive evaluation index is a specific reference value after quantification, Re is a value greater than 1, and the comprehensive evaluation index, namely the F value, has a value range between 0 and 1, which can comprehensively consider the indicators of matching accuracy P and recall rate R, and is used to measure the judgment ability of the pedestrian perception smart lighting management system for abnormal detection. A high F value means that the pedestrian perception smart lighting management system can more accurately identify abnormal states and avoid misjudging normal states as abnormal states, which helps to ensure that the pedestrian perception smart lighting management system detects and processes abnormal states more accurately and improves the detection effect. A high F value means that the pedestrian perception smart lighting management system has a high recall rate for abnormal states, which can effectively reduce errors and omissions to extract the true abnormal state; Obtain the comprehensive evaluation index of the management system for pedestrian perception smart lighting at different time periods within the time T, and mark the comprehensive evaluation index as Tg, where g represents the number of the comprehensive evaluation index of the management system for pedestrian perception smart lighting at different time periods within the time T, and g={1,2,3…r}, where r is a positive integer; The expression for calculating the comprehensive evaluation index is: In the formula, P represents the matching accuracy rate, which refers to the proportion of abnormal states that are judged as abnormal states by the management system for pedestrian perception smart lighting, and is actually abnormal. The calculation method is: matching accuracy rate = number of abnormal states correctly matched / number of all abnormal states judged as abnormal states; R represents the recall rate, which refers to the ratio between the number of abnormal states correctly matched by the management system for pedestrian perception smart lighting and the number of all actual abnormal states. The calculation method is: recall rate = number of abnormal states correctly matched / number of all actual abnormal states; The number of all abnormal states refers to the number of all abnormal states determined by the pedestrian perception smart lighting management system within a certain time range, including real abnormal states and misjudged normal states. The pedestrian perception smart lighting management system will classify and mark the detection states, or record the abnormal determination of transactions in the log, which can be obtained by counting the number of all states marked as abnormal; The number of all actual abnormal states refers to the number of all abnormal states that exist in actual situations. Since the actual abnormal state may be caused by detection and judgment errors, the management system for pedestrian perception smart lighting needs to interact with the terminal server or database to obtain the number of actual abnormal states; The expression for calculating the anomaly detection efficiency coefficient is: Re is the preset reference value of the comprehensive evaluation index; The method for comprehensive evaluation of pedestrian target status and detection reliability is: Based on the comprehensive analysis of the sensitivity index and the anomaly detection efficiency coefficient, a monitoring model is established to calculate the reliability index of the management system for pedestrian perception smart lighting. The calculation expression of the reliability index is: In the formula, γ and δ are the proportional coefficients of the sensitivity index and the anomaly detection efficiency coefficient, respectively, and both γ and δ are greater than 0; The logic for determining manual access operations based on anomaly detection results is as follows: The calculated reliability index of the management system for pedestrian perception smart lighting is compared with the preset reliability index threshold. If the calculated reliability index is greater than or equal to the preset reliability index, a suspicious signal is generated. If the calculated reliability index is less than the preset reliability index, a steady-state signal is generated. Perform processing strategy analysis based on the generated suspicious signal. When the generated suspicious signal is received, a data set is generated based on the continuous reliability index data of the pedestrian perception smart lighting management system within T time after the suspicious signal is generated, and the reliability index in the data set is calibrated as Rv, where v is the reliability index number, that is, v = {1, 2, 3...x}, where x is a positive integer; Calculate the standard deviation of several reliability pointing indexes in the data set, and calibrate the standard deviation of the reliability pointing index as So, and compare the reliability pointing index standard deviation So with the preset reliability pointing index standard deviation threshold Do, and perform early warning processing based on the comparison results. The processing logic is as follows: If So is greater than or equal to Do, the pedestrian perception smart lighting management system is marked as a high-risk level, and the staff is reminded that the pedestrian perception smart lighting management system has unreliable risk hazards and needs to be inspected and maintained; If So is less than Do, the pedestrian perception smart lighting management system is marked as a low-risk level, reminding the staff that the pedestrian perception smart lighting management system has low-risk hidden dangers and does not require inspection and maintenance.

Images