A safety gate management system and method based on artificial intelligence

Abstract

The present invention relates to the field of data processing technology, and specifically to a security gate management system and method based on artificial intelligence, comprising: a face recognition module, used to recognize facial expressions of pedestrians and determine facial features of pedestrians in the current scene; a flow calculation module, used to determine the average time for pedestrians to pass through and the gate reaction time in the current scene according to the number and frequency of pedestrians passing through, and to identify the moving average of pedestrians passing through in different time periods and the length of the corresponding calculated time period; a gate control module, used to obtain a first configuration scheme for the current gate according to the light intensity based on the received facial features and the output data of the flow calculation module, and to obtain a second configuration scheme based on the peak number of people entering and leaving, thereby realizing personalized control of the gate and improving the passage efficiency of pedestrians.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to a safety gate management system and method based on artificial intelligence. Background Art

[0002] With the rapid development of science and technology, artificial intelligence (AI) has demonstrated its great potential and application value in many fields. In the field of secure access management, AI-based gate management systems are gradually becoming a key tool to improve security and efficiency. This system usually combines technologies such as computer vision, deep learning and the Internet of Things to achieve intelligent access control.

[0003] For example, China Patent Publication No. Disclosed are a gate control method, device, equipment and computer-readable storage medium. The gate control method includes obtaining facial information collected by the gate; matching the collected facial information with a first priority database corresponding to a current traffic time period, wherein different traffic time periods correspond to different first priority databases, and the first priority database corresponding to the current traffic time period stores initial facial information of passers-by whose probability of passing in the current traffic time period is greater than a preset value; if the match is successful in the first priority database corresponding to the current traffic time period, the gate is controlled to open.

[0004] However, the existing technology ignores the fact that when pedestrians are passing through, the gate opening time also needs to be adjusted according to the different needs of pedestrians to reduce the time spent by pedestrians when passing through and reduce congestion when passing through the gate. Summary of the invention

[0005] In order to solve the above technical problems, the technical solution adopted by the present invention is:

[0006] As a preferred technical solution of the present invention, a safety gate management system based on artificial intelligence includes:

[0007] The face recognition module is used to recognize the facial expressions of pedestrians and determine the facial features of pedestrians in the current scene;

[0008] The flow calculation module is used to determine the average time for pedestrians to pass and the gate reaction time in the current scene according to the number and frequency of pedestrians passing through, and to identify the moving average of pedestrians passing through in different time periods and the length of the corresponding calculated time period;

[0009] The gate control module is used to obtain the first configuration plan for the current gate according to the light intensity based on the received facial features and the output data of the flow calculation module, obtain the second configuration plan according to the peak number of people entering and exiting, and obtain the third configuration plan according to the pedestrian's traffic habits, and adjust the gate opening and closing time according to the set first configuration plan, second configuration plan, and third configuration plan.

[0010] The control method of the gate control module also includes: a gate operation control module and a gate control adjustment module, wherein the gate operation control module is used to obtain the stability of the gate opening and closing speed and the stability of the gate opening and closing angle when the gate opening and closing time is adjusted, and determine the operation status of the gate; the gate control adjustment module is used to determine the change trend of the gate operation parameters under the corresponding gate operation status according to the number of people passing through the gate, the average time passing through the gate, and the length of the passing time, and to issue an early warning to the gate according to the change trend of the gate operation parameters.

[0011] A safety gate management method based on artificial intelligence, comprising:

[0012] Step S1, processing the collected face image to identify the facial features of the pedestrian;

[0013] Step S2, identifying the pedestrians currently passing by by using the acquired facial features, and determining the moving average of the pedestrians' passing time according to the number of pedestrians passing by, and recording the corresponding time period;

[0014] Step S3, adjusting the configuration scheme of the gate according to the current pedestrian traffic data, so that the gate is set to different configuration schemes according to different scenarios;

[0015] Step S4, determining the stability of the gate operation after setting the configuration scheme, and issuing an early warning when the gate is abnormal. The beneficial effects of the present invention are: 1. The present invention sets the first configuration scheme, the second configuration scheme, and the third configuration scheme according to the identified light intensity, the number of pedestrians, and the pedestrians' travel habits, and adjusts the gate opening and closing time, dynamically adjusts the gate opening and closing time, can reduce the waiting time of pedestrians, and thus improve the travel efficiency; by identifying the operating state of the gate after adjustment, the gate can maintain stable operation after adjustment, thereby improving the working efficiency of the gate itself.

[0016] Second, the present invention uses face recognition technology to identify pedestrians multiple times, thereby improving the accuracy of pedestrian face recognition;

[0017] 3. The present invention can identify the environment when pedestrians pass through, so that when pedestrians pass through, the gate opening and closing can be adjusted according to the average time of pedestrians passing through, and the working state of the gate under the corresponding state can be determined to monitor the stability of the gate operation.

[0018] 4. The present invention recognizes the gap time corresponding to the gate machine during operation and determines whether the gap time changes during operation, so as to accurately detect whether the gate machine has an abnormal operation and realize intelligent management of the gate machine. BRIEF DESCRIPTION OF THE DRAWINGS

[0019] The present invention is further described below in conjunction with the accompanying drawings and embodiments.

[0020] Figure 1 It is a system diagram of a safety gate management system based on artificial intelligence;

[0021] Figure 2 It is a flow chart of a safety gate management method based on artificial intelligence. DETAILED DESCRIPTION

[0022] The embodiments of the present invention are described in detail below. The embodiments described below are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention. If no specific techniques or conditions are specified in the embodiments, the techniques or conditions described in the literature in the art or the product specifications are used.

[0023] When pedestrians use the turnstile, the requirements for the gate opening time will vary in different lighting environments and scenarios, so the turnstile will process at a fixed time, which will cause congestion. For example, in a dark scene, it takes a long time to recognize pedestrians, and the time for pedestrians to pass through the turnstile will also increase accordingly. At this time, the turnstile needs to dynamically adjust its opening time according to the corresponding environment and scenario to improve the accuracy and efficiency of the turnstile's recognition of pedestrians.

[0024] To solve this problem, see Figure 1 , a safety gate management system based on artificial intelligence, including:

[0025] The face recognition module is used to recognize the facial expressions of pedestrians and determine the facial features of pedestrians in the current scene;

[0026] The flow calculation module is used to determine the average time for pedestrians to pass and the gate reaction time in the current scene according to the number and frequency of pedestrians passing through, and to identify the moving average of pedestrians passing through in different time periods and the length of the corresponding calculated time period;

[0027] The gate control module is used to obtain the first configuration plan for the current gate according to the light intensity based on the received facial features and the output data of the flow calculation module, obtain the second configuration plan according to the peak number of people entering and exiting, and obtain the third configuration plan according to the pedestrian's traffic habits, and adjust the gate opening and closing time according to the set first configuration plan, second configuration plan, and third configuration plan.

[0028] Preferably, in order to control the gate to be able to adaptively change according to the scene, the control method for the gate control module also includes: a gate operation control module and a gate control adjustment module, wherein the gate operation control module is used to obtain the gate opening and closing speed stability and the gate opening and closing angle stability when the gate opening and closing time is adjusted to determine the gate operation status; the gate control adjustment module is used to determine the change trend of the gate operation parameters under the corresponding gate operation status according to the number of people passing through the gate, the average time passing through the gate, and the length of the passing time, and to warn the gate according to the change trend of the gate operation parameters.

[0029] When the face of the current pedestrian is obtained, the recognized face is compared with a preset face feature library, and the face feature with the highest similarity to the current expression is selected as the recognized face feature.

[0030] The recognition methods of pedestrian facial features include:

[0031] Determine eye movement data, facial image angle, and facial image size ratio of a target object in the corresponding image area from within the acquired image area;

[0032] The recognition of facial image angles includes:

[0033] ;

[0034] in, It represents the pixel of the corrected face image. It represents the pixel of the face image before correction. It is the angle of face correction. In this scenario, the z-axis is kept unchanged to determine the angle of the face relative to the plane.

[0035] In addition, since the scale ratio of the face image is affected by the distance between the pedestrian and the gate, when obtaining the face features of the target image in the image area, it is necessary to determine the distance of the target object from the current gate based on the ratio value of the face image and the preset face image;

[0036] The ratio of a face image is expressed as:

[0037] ;

[0038] in, Indicates the ratio of the target object's face image to the preset face image.

[0039] The above formula is used to inversely solve the correction angle and the ratio of the corresponding face image to the preset face image after face correction, so as to determine the angle and ratio of the corresponding face after rotation transformation, so as to facilitate face recognition.

[0040] The method of acquiring the eye movement data includes determining the angle of the eye movement data in the collected image to identify which gate the pedestrian is observing, and when the pedestrian reaches a designated location where the gate recognizes a face, the corresponding face is recognized.

[0041] The recognition image corresponding to the current target object is determined by the angle of the face image and the size ratio of the face image. At the same time, combined with the target object's eye movement data, it is determined whether the gate the current target object is waiting for is the current gate. According to the ratio value of the target object's face image and the preset face image, the relative distance between the target object and the gate is determined. When the relative distance between the target object and the gate reaches the preset distance value, the gate is opened.

[0042] At the same time, since the face image recognized in a black light environment or a strong light environment will have a corresponding loss, resulting in face recognition failure, the face recognition module also includes obtaining the intensity and color temperature of the ambient light through an optical sensor, such as using a illuminance meter to measure the intensity of the current ambient light, and using a color temperature sensor to measure the color temperature of the ambient light;

[0043] After acquiring the face image of the target object, histogram enhancement is used to enhance the contrast of the acquired image to improve the face recognition effect under uneven lighting conditions.

[0044] The sensitivity is used to evaluate the currently acquired image. Sensitivity is also called true positive rate or recall rate. It can be determined by calculating the ratio of true positive cases (TP) to all actual positive cases (TP+FN). In face recognition scenarios, this usually involves a labeled test data set containing known positive and negative cases; the sensitivity can be calculated by counting the ratio of the number of correctly identified faces to the total number of faces.

[0045] The processing method of the flow calculation module includes obtaining a first number of people appearing in the image area within a target time period, and determining whether the first number of people exceeds a target number of people;

[0046] If the target number of people is exceeded, the image data collected by the adjacent gate at the corresponding position is obtained to determine the distribution of the number of people at the current position. The corresponding distribution of the number of people is used to determine the entrance of the gate where the people in the current area are located and to determine the number of people corresponding to each gate.

[0047] In order to accurately identify the changes in the number of people, the processing method of the traffic calculation module also includes: obtaining continuous frame images of the target object, determining the calculation results of the continuous frame images of the gate machine, determining the recognition accuracy of the corresponding facial features under the changes in the image frames, and extracting three image frames to determine the difference between the target image frame and the previous image frame and the next image frame.

[0048] First, obtain the calculation formula of the target image frame and the previous and next image frames:

[0049] ; ;

[0050] Then the overall comparison result between the target image frame and the previous and next image frames is:

[0051] ;

[0052] ;

[0053] in, is the target image frame and the The difference value of the image frames, is the target image frame and the previous The difference value of the image frames, is the pixel point in the target image frame, is the target image frame Pixels in an image frame, After the target image frame Pixels in an image frame, is the target image frame and the The difference between the image frame and the target image frame and the previous image frame The intersection of the difference values ​​of image frames, is the target image frame and the The difference between the image frame and the target image frame and the previous image frame The union of the difference values ​​of the image frames.

[0054] The absolute value of the brightness difference between the two frames is obtained by differential operation on the continuous images and compared with the defined threshold. If it is greater than the threshold T, it means that there is a certain difference in the recognized face image after the lighting adjustment, otherwise it is recorded as 0.

[0055] When recognizing images of three image frames, the movement of the target object can be tracked by taking the intersection and union methods. For example, when using the intersection operation, movement is considered to have occurred only when the differential values ​​of the two image frames have changed. At this time, small movements are ignored and only large movements are obtained to identify the current position of the pedestrian. When taking the union at the same time, small movements in the image frames are captured to identify the situation in the current scene.

[0056] The scenario for taking the intersection is that, assuming that the turnstile is used for the entrance and exit of a subway or train station, it is necessary to accurately count the number of people passing through the turnstile; when passengers pass through the turnstile, it is hoped that the system can accurately detect each person passing through without being disturbed by other factors (such as clothing blown by the wind, changes in light and shadow, etc.). By taking the intersection of the difference between two frames, it can be ensured that only when obvious motion is detected in two consecutive frames, it is considered that a person has passed through.

[0057] The scenario of taking the union can be, for example, in areas that require high security, such as security gates at airports, any unauthorized passage must be detected immediately. By taking the union of the difference between two frames, the system can capture any small movement at the first time; for example, in the airport security area, if someone tries to pass quickly or evade gate detection, the method of taking the union can trigger the alarm faster, allowing security personnel to respond quickly.

[0058] After determining the number of people identified, record the timestamp of each pedestrian passing through the gate, the face recognition result, and the time when the gate opens and closes.

[0059] The traffic calculation module will calculate the moving average time of pedestrians based on the identified data. The specific implementation method also includes counting the travel time of each pedestrian, including the time from when the pedestrian enters the identification area to when he / she completely passes through the gate;

[0060] Analyze the gate’s reaction time, that is, the time from successful face recognition to the gate opening;

[0061] According to the number and frequency of pedestrians passing through, the average time of pedestrians passing through is calculated as the moving average of the output.

[0062] ;

[0063] in, is the moving average of pedestrian traffic, is the number of pedestrians passing through, It is Pedestrian travel time.

[0064] After obtaining the average travel time of pedestrians, the number of people passing through the preset time period is extracted, the frequency, number and duration of pedestrian passage are identified, and the number of people passing through the current gate, the speed and scale of crowd gathering, and the predicted peak time are determined.

[0065] The number of times each pedestrian passes through the gate within a preset time period reflects the frequency of his / her passage; the passage records of each pedestrian are counted to obtain the total number of passages; for each passage of each pedestrian, the passage time is recorded to analyze the duration of the pedestrian's stay at the gate; by observing the changes in the number of people passing through the gate in a continuous time period, the speed of crowd gathering can be estimated.

[0066] The moving average is used to establish a time series according to the corresponding time period. According to the preset time length, the corresponding moving average of pedestrians at each time point is determined. According to the change of the moving average, the change of the moving average in the adjacent time length is determined to obtain the degree of pedestrian traffic at different times. The moving average is the average movement time of pedestrians.

[0067] When calculating the moving average of pedestrians, the length of the calculated time period needs to meet the following conditions:

[0068] ;

[0069] in, Based on the time period, is the total number of people passing through the gate that exceeds the preset number threshold. is the length of time corresponding to the calculation of the average moving time of pedestrians, is the time period increment.

[0070] At this time, the time period for calculating the moving average is adjusted by observing the number of people passing through the gate to find out the changing trajectory of the corresponding moving average in different time periods.

[0071] At the same time, when the total number of people passing through the gate exceeds the preset number threshold, the length of the calculated time period will be adjusted according to the proportion of the number of people, specifically expressed as:

[0072] ;

[0073] in, It is the proportional coefficient, which represents the ratio of the total number of people passing through the gate that exceeds the preset number threshold to the preset number threshold.

[0074] At this time, the degree of traffic passing through the current gate will be detected based on the actual flow of people passing through.

[0075] The identified data are sent to the gate control module so that the gate control module can adjust the gate opening time according to the current scene.

[0076] In one embodiment of the present invention, after determining the movement of personnel, the processing method for the gate control module includes setting multiple control schemes according to the current scene to determine the opening and closing time of the gate under the corresponding scene.

[0077] Set environmental tags for the collected images and determine the gate configuration based on light intensity, peak time of in and out traffic, and pedestrian personality habits.

[0078] When the light intensity is greater than the preset intensity value, the current gate is set to the first configuration scheme; when the corresponding time period is the peak time for the number of people entering and exiting, it is set to the second configuration scheme; when the identified pedestrian is a pedestrian who frequently passes through the current gate, it is set to the third configuration scheme according to the pedestrian's traffic habits.

[0079] The first configuration scheme determines the face recognition time under the current light intensity, determines the changing trend of the face recognition time with the time period, and determines the moving average value under the corresponding time.

[0080] The relationship between face recognition time and light intensity can be expressed by a function:

[0081] ;

[0082] in, is the face recognition time, is the light intensity, It is the functional relationship between light intensity and face recognition time.

[0083] Determine the change value of face recognition time based on light intensity, calculate the total passage time required after adding the moving average to the face recognition time, determine the maximum difference in the total passage time under different lighting conditions, and use the time calculated after weighting the average of the maximum difference as the time length set for the first configuration scheme. The selected weight is determined based on the maximum weight value of the functional relationship between light intensity and face recognition time to obtain the weight under the light intensity environment.

[0084] In the scenario of the first configuration scheme, the speed of face recognition slows down due to strong light. At the same time, strong light will have a certain impact on the speed of crowd movement, making the crowd movement speed appear too fast or too slow. At this time, the average passing time of pedestrians in the current scene is first obtained, and the movement trajectory of pedestrians is tracked. The current gate opening and closing time is adjusted according to the movement trajectory of pedestrians, so that the next waiting pedestrian can reduce the waiting time and improve the passage efficiency.

[0085] The second configuration scheme is manifested as obtaining the number of people corresponding to the peak time of the number of people entering and exiting, determining the expected number of people passing through and the expected passage time within a fixed time period based on the queue time and queue length of the crowd, and adjusting the opening and closing time of the gate according to the obtained expected passage time.

[0086] The estimated travel time is expressed as:

[0087] ;

[0088] in, is the estimated time of travel; , , , are the parameters of the model, also called regression coefficients, which are estimated from historical data using statistical methods such as the method of least squares. is the intercept term, which represents the baseline value of the estimated travel time when all independent variables are 0. is a coefficient related to the number of people, which indicates how the expected travel time will change when the number of people increases by one unit. is a coefficient related to queue length, reflecting the impact of queue length on the expected travel time. It is a coefficient related to the time period, indicating the impact of different time periods on the estimated travel time; is the number of people; is the queue length; It is a quantitative indicator of time period. Since the traffic conditions in different time periods (such as morning peak, noon, evening, etc.) may be different, it is necessary to consider the time period as a variable. This variable can be a categorical variable, which is converted into a numerical form through one-hot encoding or other methods for use in the regression model; is the error term, which represents the partial variation that cannot be explained by the model. In practical applications, even the most accurate model cannot accurately predict every observation, so the error term is used to capture the random variation that is not explained by the model.

[0089] In the second configuration scheme, by analyzing the degree to which the moving average value of pedestrians is affected by time period and crowd size, the required travel time under different crowd sizes and queue numbers is determined to improve the pedestrian travel speed.

[0090] The third configuration scheme is manifested as determining the corresponding moving average values ​​and gate opening and closing times of pedestrians in different time periods; identifying the movement characteristics corresponding to the current movement of pedestrians, and the movement characteristics are extracted from the movement speed, position, and movement amplitude of the pedestrians; comparing the obtained movement characteristics with historical data, and selecting the preset movement characteristics with the highest similarity to obtain the movement characteristics, and taking the gate opening and closing time corresponding to the preset movement characteristics with the highest similarity as the configuration result of the third configuration scheme.

[0091] The third configuration scheme is to set up for pedestrians who frequently pass through this specific gate. By obtaining the time and speed when pedestrians pass each other and comparing the current movement characteristics of the passage with the currently collected image frame, the estimated movement time required by the current pedestrian can be determined. When a pedestrian is identified, the gate can be opened and closed in time to achieve unimpeded travel for pedestrians.

[0092] When controlling the gate machine, in order to reduce the energy required for the operation of the gate machine and speed up the passage of people, the opening angle of the gate machine can also be adjusted to improve the opening efficiency and adjust the retraction and extension speed of the gate machine.

[0093] Specifically, the gate operation control module also includes obtaining the current gate configuration plan, determining the number of gate operations when the gate is running according to the gate configuration plan, and determining the gate load based on the number of gate operations and the number of people passing through.

[0094] The load of the gate machine is determined by the gate opening and closing time, the number of people passing through, and the number of gate machine operations set in the configuration plan. The working intensity and frequency of the gate machine in different time periods are determined to represent the working load of the gate machine.

[0095] The load of the gate is expressed as:

[0096] ;

[0097] in, is the coefficient corresponding to the load, is the total time the gate is opened and closed. It is the length of the time period corresponding to the total time of gate opening and closing. is the number of people passing through the gate, is the maximum number of people that can pass through the gate. is the number of gate operations, is the maximum number of gate operation times. is the weight coefficient corresponding to the total time of gate opening and closing, is the weight coefficient corresponding to the number of people passing through the gate, It is the weight coefficient corresponding to the number of gate operation times.

[0098] After determining the load of the gate machine, the processing of the gate machine itself can be set according to the flow of people in its working scene, and the control angle of the corresponding gate door can be adjusted to improve the control effect of the gate machine.

[0099] According to the acquired gate load and gate configuration time, the pedestrian flow corresponding to each angle of the gate within the preset adjustment angle range is determined, and the setting angle of the gate is adjusted based on the determined pedestrian flow.

[0100] Assume that in a gate system at a subway station, the opening angle is adjusted to improve traffic efficiency and save energy.

[0101] Initial state: The gate was originally set to open at full angle (for example, 90 degrees). Although the passage is smooth, the energy consumption is high.

[0102] Through observation and analysis, it was found that pedestrians do not need a fully opened door width when passing through the turnstile.

[0103] Therefore, it was decided to adjust the opening angle of the gate to 60 degrees, which not only ensures the smooth passage of pedestrians but also reduces the opening range.

[0104] After the adjustment, the energy required for the gate to open each time was significantly reduced.

[0105] At the same time, since the gate opening speed was not affected, the pedestrian traffic speed also remained at the original level.

[0106] Continue to collect data and observe pedestrian traffic, and fine-tune the gate opening angle and speed settings according to actual conditions to achieve optimal traffic efficiency and energy saving.

[0107] Preferably, in order to check the stability of the gate opening and closing speed stability and the gate opening and closing angle stability of the current gate under the corresponding load, the gate operation control module processes the gate opening and closing speed stability and the gate opening and closing angle stability, including determining the relative acceleration of adjacent gates, determining the gap time between the opening and closing of the gate under the corresponding configuration scheme, and performing a stability test on the obtained relative acceleration and gap time, keeping the configuration scheme unchanged when pedestrians pass, determining the value of the relative acceleration, and when the relative acceleration fluctuates, determining the corresponding time, and sending the data of the corresponding time to the database;

[0108] For the gap time, determine the difference between the predicted gap time calculated by the configuration scheme and the actually measured gap time under the corresponding configuration scheme, and determine the feedforward coefficient corresponding to the gap time;

[0109] The feed-forward coefficient is expressed as:

[0110] ;

[0111] in, is the feed-forward coefficient, is the adjusted gap time, is the unadjusted gap time, is the running unevenness value, It is the running roughness value measured after adjustment.

[0112] The running unevenness value obtained here is obtained by measuring the wear of the internal mechanical parts of the gate machine, determining the wear of the gate machine under the corresponding running time, and determining the impact of the wear on the closing and opening time of the gate machine operation. At the same time, when calculating the feedforward coefficient, all data are normalized to obtain a coefficient that can express the current operation.

[0113] Relative acceleration is expressed as the corresponding acceleration value when adjacent gates are opened and closed at the same time. It is used to determine whether the current gate opens too quickly or too slowly. Under normal circumstances, the gate opening for a large number of pedestrians needs to be completed within 0.5S. If the opening time exceeds this time, or is faster than the normal setting standard, it means that the gate is unstable under the current configuration adjustment. Obtaining the recommended time between the opening and closing of the gate is suitable for obtaining the total time of the gate's operation. At this time, it will be judged whether the gap time meets the current situation settings to prevent erroneous settings and failure to change them in time after setting. At the same time, obtaining the gap time can more conveniently determine the average time for pedestrians to pass. The gap time is included in the pedestrian's passage time. The pedestrian passage time includes face recognition time, the gap time between the gate door opening and closing, and the gap time between the gate opening and closing.

[0114] Based on the obtained feedforward coefficient, the change value of the feedforward coefficient is determined when the gate is running. When the change value of the feedforward coefficient is a smooth curve, it means that the current gate operation is stable. When the change value of the feedforward coefficient exceeds the preset feedforward coefficient value, it means that the current gate operation is unstable.

[0115] In order to deal with the unstable condition of the gate machine, the gate machine control adjustment module includes the following processing methods:

[0116] When the comparison is unstable, the timestamp of the pedestrian's passing time at the gate is used to determine whether the machine-recognized data and the calculated data are consistent;

[0117] Determine whether the current pedestrian passing time exceeds the preset maximum passing time. When the pedestrian passing time is less than or equal to the preset maximum passing time, if there are pedestrians passing through the current gate, determine whether the gate gap time is less than the first threshold; when there are no pedestrians in the gate, determine whether the gate gap time is less than the second threshold; when there are pedestrians passing through the gate and there are waiting pedestrians, determine whether the gate gap time is less than the third threshold; the first threshold is less than the second threshold and less than the third threshold; when the above situation is judged to be yes, the current corresponding situation is a stable state, when the judgment result is no, it is abnormal data, and abnormal analysis is performed.

[0118] When performing abnormal analysis, the acquired data is used to perform curve fitting on the corresponding operating parameters according to the different gap times, and the degree of instability of the current operation is determined based on the area of ​​the graph composed of the curve fitting.

[0119] When the pedestrian's passage time meets the requirements, the system will further check the status inside the gate.

[0120] Case A: If there are still pedestrians passing through the gate, the time interval from the last pedestrian leaving to the current pedestrian starting to pass (i.e. the gate interval time) will be checked to determine whether it is less than the set first threshold. This first threshold is set relatively small to ensure that pedestrians can pass continuously and quickly during peak hours.

[0121] Case B: If there are no pedestrians in the gate, the system will check whether the time interval from the last pedestrian completely leaving the gate to the current pedestrian starting to pass through, that is, whether the gate interval time is less than the second threshold. Since there are no other pedestrians in the gate at this time, the second threshold can be set slightly larger to give the system more preparation time.

[0122] When there are pedestrians passing through the gate, if pedestrians waiting to pass through are detected outside the gate, the system will further determine whether the gate clearance time at this time is less than the third threshold. This third threshold is usually set to the maximum to balance the efficiency and safety of passage, ensuring that the waiting pedestrians can start to pass within a reasonable time, while avoiding the gate from opening and closing too frequently.

[0123] At the same time, when an abnormality occurs, the current gap time will be greater than the set first threshold, second threshold, and third threshold. The equipment operation information corresponding to the gap time when the abnormality occurs is identified and sent to the terminal to assist the background personnel in analysis and processing.

[0124] like Figure 2 As shown, the present invention also provides a safety gate management method based on artificial intelligence, comprising:

[0125] Step S1, processing the collected face image to identify the facial features of the pedestrian;

[0126] Step S2, identifying the pedestrians currently passing by by using the acquired facial features, and determining the moving average of the pedestrians' passing time according to the number of pedestrians passing by, and recording the corresponding time period;

[0127] Step S3, adjusting the configuration scheme of the gate according to the current pedestrian traffic data, so that the gate is set to different configuration schemes according to different scenarios;

[0128] Step S4, determining the stability of the gate operation after setting the configuration plan, and issuing an early warning when the gate has an abnormality.

[0129] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations on the present invention. A person skilled in the art may make changes, modifications, substitutions and variations to the above embodiments within the scope of the present invention and they are still covered by the protection scope of the present invention.

Claims

1. A safety gate management system based on artificial intelligence, characterized in that: include: The face recognition module is used to recognize the facial expressions of pedestrians and determine the facial features of pedestrians in the current scene; The flow calculation module is used to determine the average time for pedestrians to pass and the gate reaction time in the current scene according to the number and frequency of pedestrians passing through, and to identify the moving average of pedestrians passing through in different time periods and the length of the corresponding calculated time period; The gate control module is used to obtain a first configuration scheme for the current gate according to the light intensity based on the received facial features and the output data of the flow calculation module, obtain a second configuration scheme according to the peak number of people entering and exiting, and obtain a third configuration scheme according to the traffic habits of pedestrians, and adjust the gate opening and closing time according to the set first configuration scheme, second configuration scheme, and third configuration scheme; The control method of the gate control module also includes: a gate operation control module and a gate control adjustment module, wherein the gate operation control module is used to obtain the gate opening and closing speed stability and the gate opening and closing angle stability when the gate opening and closing time is adjusted, and determine the gate operation status; the gate control adjustment module is used to determine the change trend of the gate operation parameters under the corresponding gate operation status according to the number of people passing through the gate, the average time of passing through the gate, and the length of the passing time, and to warn the gate according to the change trend of the gate operation parameters; The processing method of the gate control module includes setting environmental tags for the collected images, determining the configuration scheme of the gate based on light intensity, peak time of the number of people entering and exiting, and the individual habits of pedestrians; when the light intensity is greater than the preset intensity value, the current gate is set to the first configuration scheme, and when the corresponding time period is the peak time of the number of people entering and exiting, it is set to the second configuration scheme; when the identified pedestrian is a pedestrian who often passes through the current gate, it is set to the third configuration scheme according to the pedestrian's travel habits; The first configuration scheme determines the face recognition time under the current light intensity, determines the change trend of the face recognition time over the time period, and determines the moving average value under the corresponding time; The function between face recognition time and light intensity is expressed as: ； in, It is the face recognition time. is the light intensity, It is the functional relationship between light intensity and face recognition time; Determine the change value of the face recognition time based on the light intensity, calculate the total passing time required after adding the moving average value to the face recognition time, determine the maximum difference of the total passing time under different lighting conditions, and calculate the time after weighting the average value of the maximum difference as the time length set by the first configuration scheme; The second configuration scheme is characterized by obtaining the number of people corresponding to the peak time of the number of people entering and exiting, determining the expected number of people passing through and the expected passage time within a fixed time period based on the queue time and queue length of the crowd, and adjusting the opening and closing time of the gate according to the obtained expected passage time; The estimated travel time is expressed as: ； in, is the estimated time of travel; , , , are the parameters of the model, is the intercept term, which represents the baseline value of the estimated travel time when all independent variables are 0. is a coefficient related to the number of people, which indicates how the expected travel time will change when the number of people increases by one unit. is a coefficient related to queue length, reflecting the impact of queue length on the expected travel time. It is a coefficient related to the time period, indicating the impact of different time periods on the estimated travel time; is the number of people; is the queue length; It is a quantitative indicator of the time period; is the error term, which represents the portion of variation that is not explained in the model; The third configuration scheme is manifested as determining the corresponding moving average values ​​and gate opening and closing times of pedestrians in different time periods; identifying the movement characteristics corresponding to the current movement of pedestrians, comparing the obtained movement characteristics with historical data, selecting the preset movement characteristics with the highest similarity to the movement characteristics, and taking the gate opening and closing time corresponding to the preset movement characteristics with the highest similarity as the configuration result of the third configuration scheme.

2. According to the artificial intelligence-based safety gate management system of claim 1, it is characterized in that: The facial feature recognition method of the pedestrian includes: Determine eye movement data, facial image angle, and facial image size ratio of a target object in the corresponding image area from within the acquired image area; The recognition of facial image angles includes: ； in, It represents the pixel of the corrected face image. It represents the pixel of the face image before correction. is the angle of face correction; The ratio of a face image is expressed as: ； in, Indicates the ratio of the target object's face image to the preset face image.

3. The artificial intelligence-based safety gate management system according to claim 1 is characterized in that: The processing method of the flow calculation module also includes: obtaining continuous frame images of the target object, determining the calculation results of the continuous frame images of the gate, determining the recognition accuracy of the corresponding facial features under the change of the image frame, and extracting three image frames to determine the difference between the target image frame and the previous image frame and the subsequent image frame; The calculation formulas for the target image frame and the previous and next image frames are: ； ； The overall comparison results of the target image frame and the previous and next image frames are as follows: ； ； in, is the target image frame and the The difference value of image frames, is the target image frame and the previous The difference value of image frames, is the pixel point in the target image frame, is the target image frame Pixels in an image frame, After the target image frame Pixels in an image frame, is the target image frame and the The difference between the image frame and the target image frame and the previous image frame The intersection of the difference values ​​of image frames, is the target image frame and the The difference between the image frame and the target image frame and the previous image frame The union of the difference values ​​of the image frames.

4. The artificial intelligence-based safety gate management system according to claim 1 is characterized in that: The specific implementation of the flow calculation module also includes counting the travel time of each pedestrian, including the time from when the pedestrian enters the identification area to when he / she completely passes through the gate; Analyze the gate’s reaction time, that is, the time from successful face recognition to the gate opening; According to the number and frequency of pedestrians passing through, the average time of pedestrians passing through is calculated as the moving average of the output; ； in, is the moving average of pedestrian traffic, is the number of pedestrians passing through, It is The travel time of a pedestrian; When calculating the moving average of pedestrians, the length of the calculated time period needs to meet the following conditions: ； in, Based on the time period, is the total number of people passing through the gate that exceeds the preset number threshold. is the length of time corresponding to the calculation of the average moving time of pedestrians, is the time period increment.

5. The artificial intelligence-based safety gate management system according to claim 1 is characterized in that: The gate control module also includes obtaining a configuration scheme of the current gate, determining the number of gate operation times when the gate is running according to the configuration scheme of the gate, and determining the load of the gate based on the number of gate operation times and the number of people passing through the gate; The load of the gate is expressed as: ； in, is the coefficient corresponding to the load, is the total time the gate is opened and closed. It is the length of the time period corresponding to the total time of gate opening and closing. is the number of people passing through the gate, is the maximum number of people that can pass through the gate. is the number of gate operations, is the maximum number of gate operation times. is the weight coefficient corresponding to the total time of gate opening and closing, is the weight coefficient corresponding to the number of people passing through the gate, It is the weight coefficient corresponding to the number of gate operation times.

6. The artificial intelligence-based security gate management system according to claim 1 is characterized in that: The gate operation control module processes the gate opening and closing speed stability and gate opening and closing angle stability by determining the relative acceleration of adjacent gates, determining the interval time between the gate opening and closing under the corresponding configuration scheme, and performing a stability test on the obtained relative acceleration and interval time. When pedestrians pass through, the configuration scheme remains unchanged, the relative acceleration value is determined, and when the relative acceleration fluctuates, the corresponding time is determined, and the data of the corresponding time is sent to the database; For the gap time, determine the difference between the predicted gap time calculated by the configuration scheme and the actually measured gap time under the corresponding configuration scheme, and determine the feedforward coefficient corresponding to the gap time; The feed-forward coefficient is expressed as: ； in, is the feed-forward coefficient, is the adjusted gap time, is the unadjusted gap time, is the running unevenness value, It is the running roughness value measured after adjustment.

7. The artificial intelligence-based safety gate management system according to claim 1 is characterized in that: The processing methods of the gate control adjustment module include: When the comparison is unstable, the timestamp of the pedestrian's passing time at the gate is used to determine whether the machine-recognized data and the calculated data are consistent; Determine whether the current pedestrian passing time exceeds the preset maximum passing time. When the pedestrian passing time is less than or equal to the preset maximum passing time, if there are pedestrians passing through the current gate, determine whether the gate gap time is less than the first threshold; when there are no pedestrians in the gate, determine whether the gate gap time is less than the second threshold; when there are pedestrians passing through the gate and there are waiting pedestrians, determine whether the gate gap time is less than the third threshold; the first threshold is less than the second threshold and less than the third threshold; when the above situation is judged as yes, the current corresponding situation is a stable state, and when the judgment result is no, it is abnormal data, and abnormal analysis is performed; When performing abnormal analysis, the acquired data is used to perform curve fitting on the corresponding operating parameters according to the different gap times, and the degree of instability of the current operation is determined based on the area of ​​the graph composed of the curve fitting.

8. A safety gate management method based on artificial intelligence, using the safety gate management system based on artificial intelligence as claimed in claim 1, characterized in that: include: Step S1, processing the collected face image to identify the facial features of the pedestrian; Step S2, identifying the pedestrians currently passing by by using the acquired facial features, and determining the moving average of the pedestrians' passing time according to the number of pedestrians passing by, and recording the corresponding time period; Step S3, adjusting the configuration scheme of the gate according to the current pedestrian traffic data, so that the gate is set to different configuration schemes according to different scenarios; Step S4, determining the stability of the gate operation after setting the configuration plan, and issuing an early warning when the gate has an abnormality.

Images