An intelligent city traffic monitoring system based on AI technology and a method thereof

Abstract

This invention discloses a smart city traffic monitoring system and method based on AI technology, relating to the field of urban traffic monitoring technology. It includes a video data acquisition unit, a video recognition unit, a data analysis unit, a data judgment unit, a model generation unit, a voice playback unit, an adjustment time determination unit, and a traffic light control unit. It solves the technical problem that traffic monitoring cannot identify and judge visually impaired individuals when they cross road sections, and cannot adjust the corresponding traffic light timing based on the estimated time of their crossing. The system identifies and analyzes the canes or guide dogs in traffic monitoring images to mark visually impaired individuals. It combines the location of the visually impaired individual with traffic light information to determine whether they can cross the corresponding road section. It analyzes historical accident data for that road section to generate the corresponding adjustment time value for the traffic light, and adjusts the traffic light timing based on the adjustment time value.

Description

Technical Field

[0001] This invention relates to the field of urban traffic monitoring technology, specifically to a smart city traffic monitoring system and method based on AI technology. Background Technology

[0002] The concept of a smart city originated in the media field. It refers to the use of various information technologies or innovative concepts to connect and integrate urban systems and services in order to improve the efficiency of resource utilization. Currently, smart city traffic monitoring systems are used to monitor traffic conditions in smart cities and provide feedback on traffic situations so that traffic congestion can be dealt with in a timely manner.

[0003] Patent publication number CN113744528A discloses a smart city traffic video surveillance system, including a traffic video surveillance platform, network switch, traffic command center, cloud database, main server, and backup server. This invention effectively integrates traffic flow detection equipment, license plate recognition equipment, video surveillance equipment, traffic signal control equipment, and GPS positioning equipment installed on road sections. After processing and analysis, it forms a traffic information system that integrates data collection, processing, and dispatch. The system determines the location of traffic accidents by uploading traffic accident videos and quickly dispatches the nearest police force to handle the accident through mobile policing, thereby reducing the impact of traffic accidents on surrounding traffic.

[0004] However, the above-mentioned solutions cannot identify and mark visually impaired individuals when they cross road sections, nor can they combine the location of visually impaired individuals with traffic light information to determine whether they can pass through the corresponding road section. Furthermore, they cannot analyze historical accident data for that road section to generate corresponding traffic light adjustment times, thus failing to guarantee the safe passage of visually impaired individuals and increasing the probability of accidents involving them. Based on this, a smart city traffic monitoring system and method based on AI technology is proposed. Summary of the Invention

[0005] The purpose of this invention is to provide a smart city traffic monitoring system and method based on AI technology, which solves the technical problem that when visually impaired people pass through a road section, traffic monitoring cannot identify and judge visually impaired people, and cannot adjust the corresponding traffic light timing according to the estimated time of visually impaired people crossing the intersection.

[0006] The objective of this invention can be achieved through the following technical solutions:

[0007] A smart city traffic monitoring system and method based on AI technology, comprising:

[0008] The video data acquisition unit is used to acquire traffic monitoring images and send them to the video recognition unit;

[0009] The video recognition unit is used to perform frame difference analysis on the images in traffic monitoring to obtain moving targets. Then, it uses a recognition model to identify visually impaired people and marks them as target people. Otherwise, no moving targets are marked. The unit extracts the contour features of the target people and sends them to the data analysis unit.

[0010] The data analysis unit is used to track and acquire the movement distance of the target person based on the contour features of the target person, analyze the movement distance of the target person, obtain the standard movement speed of the target person, and send the standard movement speed to the data judgment unit.

[0011] The data judgment unit is used to acquire the traffic light information corresponding to the target person in the same direction and the target person's standard movement speed. The traffic light information includes the traffic light status and the traffic light countdown. By combining and analyzing the traffic light status, the traffic light countdown and the target person's standard movement speed, the unit determines whether to generate a reminder signal or an adjustment signal and transmits the reminder signal to the voice playback unit and the adjustment time judgment unit.

[0012] The time adjustment determination unit is used to analyze the historical accident data of the acquired road segment, generate the corresponding risk coefficient of the road segment, and combine the risk coefficient with the time difference of the target personnel for analysis. At the same time, it generates an adjustment time value based on the analysis results and sends the adjustment time value to the traffic light control unit.

[0013] The model generation unit is used to iteratively train the image set to obtain a recognition model, and then send it to the video recognition unit.

[0014] As a further aspect of the present invention, the specific method for obtaining the standard movement speed of the target personnel is as follows:

[0015] S1: Acquire the movement distance of the target personnel every g1 time intervals, and continuously obtain the movement distance of the target personnel n+1 times, and mark them as D1, D2, ..., Dn, D(n+1) respectively. Here, n+1 refers to the number of times the movement distance of the target personnel is acquired, 1≤n;

[0016] S2: Calculate r1 movement differences of the target personnel using the formula Zr1={D(r1+1)-Dr1} / Dp, and label them as Z1, Z2, ..., Zr1 respectively, where Dp={D1+D2+, ...,+Dr1+D(r1+1)} / r1, 1≤r1≤n;

[0017] S3: Calculate the movement speed corresponding to the movement difference of the target person r1 using the formula Zr2 / g1=vr2, and label them as v1, v2, ..., vr1 respectively, and obtain their average value vp. Then obtain all vr2 values ​​in v1, v2, ..., vr1 that satisfy the judgment formula E1. The judgment formula E1 is |vr2-vp|≤Y2, where Y2 is a preset value, 1≤r2≤r1;

[0018] Calculate the average of all movement speeds v1, v2, ..., vr1 that satisfy the judgment formula E1, and use it as the standard movement speed vb1 of the target person.

[0019] As a further aspect of the present invention, the specific method for generating the reminder signal or adjustment signal is as follows:

[0020] A1: Obtain the real-time location of the target person, mark the closest distance between the target person and the waiting line at the intersection as L1, mark the distance between the waiting lines at the intersections on both sides of the road as L2, and mark the traffic light countdown as Ht3;

[0021] A2: The estimated time Dt2 for the target person to cross the intersection is calculated using the formula {(L1+L2)∕vb1}×θ1=Dt2, where θ1 is a correction coefficient.

[0022] When Dt2≥Ht3, an alert signal is generated;

[0023] When Dt2 < Ht3, the time difference C1 of the target personnel is obtained by the formula Ht3 - Dt2 = C1; when C1 < Y3, an adjustment signal is generated; when C1 ≥ Y3, an alert signal is generated, where Y3 is a preset value.

[0024] As a further aspect of the present invention, the specific method for obtaining the adjustment time value is as follows:

[0025] P1: Obtain historical accident data for this road segment within the time range W. The historical accident data includes the number of accidents and the accident interval. The accident interval is the time interval between two accidents.

[0026] P2: Label the number of accidents and the interval between accidents within the time range W as H and J, respectively. H-1 ;

[0027] Through formula Calculate the accident interval J within the time range W. H-1 The discrete values, where The accident interval J for this road section within the time range W. H-1x The mean, =(J1+J2+、…、+J H-1) / H - 1, 1 ≤ x ≤ H - 1;

[0028] If U ≤ Q1, then is used as the accident interval duration jg1 for this section of the road; if U > Q1, then according to |J x - |, the corresponding J x values are deleted in descending order, and after each deletion, the discrete value U of the remaining J x is recalculated, and at the same time, the number K1 of the deleted J x is recorded until U ≤ Q1 is satisfied; if K1 < Q2, then calculate the mean value of the remaining J x and use it as the accident interval duration jg1 for this section of the road; if K1 ≥ Q2, then calculate the mean value of the maximum and minimum values of the remaining J x as the accident interval duration jg1 for this section of the road, that is , where both Q1 and Q2 are preset coefficients;

[0029] P3: Calculate the hazard coefficient WX1 of this section of the road through the formula (jg1 × α1 + H × α2) × θ2 = WX1;

[0030] Quantify the hazard coefficient WX1 and the target person time difference C1 and take their numerical values for calculation. Through the formula , calculate and obtain the adjusted time value , where α1, α2, α3, and α4 are all preset coefficients, and θ2 is a correction coefficient.

[0031] As a further solution of the present invention: a signal light control unit, used to obtain the adjusted time value and increase the corresponding signal light time through the controller according to its numerical value.

[0032] As a further solution of the present invention: a voice playback unit, used to obtain a reminder signal and play a voice prompt according to the reminder signal.

[0033] A smart city traffic monitoring method based on AI technology, the method specifically includes the following steps:

[0034] Step 1: Analyze the obtained traffic monitoring images, identify moving targets, identify blind sticks or guide dogs in the moving target images in the real-time monitoring through an identification model, and compare the identified blind stick or guide dog images with the blind stick or guide dog images in the image set used for training the identification model. Determine the moving targets with a similarity greater than the preset value Y1 as visually impaired persons and label them as target persons;

[0035] Step 2: Track and acquire the target person's movement distance based on the target person's outline features, analyze the target person's movement distance to obtain the target person's standard movement speed, combine the traffic light status, traffic light countdown, and the target person's standard movement speed for analysis, and determine whether to generate a reminder signal or an adjustment signal.

[0036] Step 3: The traffic light control unit acquires the adjustment signal, analyzes the historical accident data of the road segment, generates the risk factor corresponding to the road segment, analyzes the risk factor, and generates the adjustment time value based on the analysis results.

[0037] Step 4: Acquire the reminder signal through the voice playback unit, and play a voice prompt according to the reminder signal to remind the target person to pay attention to safety and stop and wait when they arrive at the intersection waiting line.

[0038] The beneficial effects of this invention are:

[0039] (1) This invention identifies and analyzes the blind sticks or guide dogs in traffic monitoring images to mark visually impaired persons. By combining the location of the visually impaired person with the information of the traffic lights, it determines whether the visually impaired person can pass through the corresponding road segment. By analyzing the historical accident data of the road segment, it generates the corresponding danger coefficient of the road segment. Then, it combines the corresponding danger coefficient of the road segment with the time difference of the target person to generate the corresponding adjustment time value of the traffic lights. The traffic light time corresponding to the adjustment time value is increased and adjusted so that the allowed passage time of the traffic lights is increased, which further ensures that the visually impaired person can safely pass through the road segment and reduces the probability of the visually impaired person having a safety accident when passing through the road segment.

[0040] (2) In this invention, the voice playback unit plays voice prompts to remind the target person to pay attention to safety when arriving at the intersection waiting line position and to stop and wait, which further ensures the safety of visually impaired people when passing through the section of road. Attached Figure Description

[0041] The invention will now be further described with reference to the accompanying drawings.

[0042] Figure 1 This is a schematic diagram of the framework structure of a smart city traffic monitoring system based on AI technology according to the present invention;

[0043] Figure 2 This is a schematic diagram of the method structure of a smart city traffic monitoring method based on AI technology according to the present invention. Detailed Implementation

[0044] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0045] Example 1

[0046] Please see Figure 1 As shown, the present invention is a smart city traffic monitoring system based on AI technology, including a video data acquisition unit, a video recognition unit, a data analysis unit, a data judgment unit, a model generation unit, a voice playback unit, an adjustment time determination unit, and a traffic light control unit;

[0047] The model generation unit is used to iteratively train the image set to obtain a recognition model, which is then fed into the video recognition unit. The specific method for generating the recognition model is as follows:

[0048] First, images of blind canes and guide dogs under different angles and lighting conditions are obtained from a large database and used as a training image set. The blind canes and guide dogs in the training image set are labeled and then input into a basic recognition model for iterative training to obtain a recognition model. This model can automatically identify blind canes and guide dogs from video images. The trained recognition model is then applied to the video stream of urban traffic monitoring to automatically detect blind canes and guide dogs in each frame of the urban traffic monitoring images.

[0049] The video data acquisition unit is used to acquire traffic monitoring images and send them to the video recognition unit. The traffic monitoring images are captured and recorded by cameras set up on the roadside.

[0050] The video recognition unit performs frame difference analysis on traffic monitoring images to identify moving targets. It then uses a recognition model to acquire images of these moving targets in real-time monitoring and identifies canes or guide dogs within them. The unit compares the similarity scores of the identified canes or guide dogs with those in the recognition model. Moving targets with similarity scores greater than a preset value Y1 are identified as visually impaired individuals and labeled as such; otherwise, no labeling is applied. Morphological processing is then performed on the target individuals, extracting their contour features, and the data is sent to the data analysis unit. Frame difference analysis is an existing and mature technology, so it will not be elaborated upon here.

[0051] The data analysis unit is used to track and acquire the movement distance of the target person based on their contour features, analyze the movement distance to obtain the target person's standard movement rate, and send the standard movement rate to the data judgment unit. The specific method for obtaining the target person's standard movement rate is as follows:

[0052] S1: Acquire the movement distance of the target personnel every g1 time intervals, and continuously obtain the movement distance of the target personnel n+1 times, and mark them as D1, D2, ..., Dn, D(n+1) respectively. Here, n+1 refers to the number of times the movement distance of the target personnel is acquired, 1≤n;

[0053] S2: Calculate r1 movement differences of the target personnel using the formula Zr1={D(r1+1)-Dr1} / Dp, and label them as Z1, Z2, ..., Zr1 respectively, where Dp={D1+D2+, ...,+Dr1+D(r1+1)} / r1, 1≤r1≤n;

[0054] S3: Calculate the movement speed corresponding to the movement difference of the target person r1 using the formula Zr2 / g1=vr2, and label them as v1, v2, ..., vr1 respectively, and obtain their average value vp. Then, obtain all vr2 values ​​in v1, v2, ..., vr1 that satisfy the judgment formula E1. The judgment formula E1 is |vr2-vp|≤Y2, where Y2 is a preset value, 1≤r2≤r1. Relevant personnel can adjust its value according to actual needs.

[0055] Calculate the average of all movement speeds v1, v2, ..., vr1 that satisfy the judgment formula E1, and use it as the standard movement speed vb1 of the target person;

[0056] The data judgment unit is used to acquire traffic light information corresponding to the direction of the target person. This information includes the traffic light status and countdown timer. By analyzing the traffic light status and countdown timer, it determines whether to generate a reminder signal or an adjustment signal. The reminder signal is then transmitted to the voice playback unit and the adjustment time determination unit. The specific method for generating the reminder signal or adjustment signal is as follows:

[0057] A1: Obtain the real-time location of the target person, mark the closest distance between the target person and the waiting line at the intersection as L1, mark the distance between the waiting lines at the intersections on both sides of the road as L2, and mark the traffic light countdown as Ht3;

[0058] Here, the waiting line at the intersection is defined as the start and end points of the zebra crossing, and the intersection is defined as a crossroads or a straight-ahead intersection. The traffic light information corresponding to the direction of the target person refers to the control traffic light information corresponding to the direction of movement of the target person. Turning and other special cases are not considered here.

[0059] A2: The estimated time Dt2 for the target person to cross the intersection is calculated using the formula {(L1+L2)∕vb1}×θ1=Dt2, where θ1 is a correction coefficient.

[0060] When Dt2≥Ht3, an alert signal is generated;

[0061] When Dt2 < Ht3, the time difference C1 of the target personnel is obtained by the formula Ht3 - Dt2 = C1; when C1 < Y3, an adjustment signal is generated; when C1 ≥ Y3, an alert signal is generated, where Y3 is a preset value.

[0062] The time adjustment determination unit analyzes the acquired historical accident data for the road segment, generates the corresponding risk coefficient for the segment, and combines the risk coefficient with the time difference of the target personnel. Based on the analysis results, it generates an adjustment time value and sends this value to the traffic light control unit. The specific method for obtaining the adjustment time value is as follows:

[0063] P1: Obtain historical accident data for this road segment within the time range W. The historical accident data includes the number of accidents and the accident interval. The accident interval is the time interval between two accidents. The accident number and accident interval data here are from the traffic management department's data recording center. Data that has not been recorded by the traffic management department's data recording center is not included in the analysis and calculation.

[0064] Here, the time range W is 30 days, specifically referring to the time range of 30 days prior to the moment the data is acquired, excluding the data available on the day the data is acquired.

[0065] P2: Label the number of accidents and the interval between accidents within the time range W as H and J, respectively. H-1 ;

[0066] Through formula Calculate the accident interval J within the time range W. H-1 The discrete values, where The accident interval J for this road section within the time range W. H-1x The mean, =(J1+J2+、…、+J H-1 ) / H-1, 1≤x≤H-1;

[0067] If U≤Q1, then As the accident interval duration jg1 of this section; if U > Q1, then according to the value of |Jx - |, delete the corresponding Jx values in descending order, and recalculate the discrete value U of the remaining Jx after each deletion. At the same time, record the number K1 of deleted Jx until U ≤ Q1 is satisfied; if K1 < Q2, calculate the mean value of the remaining Jx and use it as the accident interval duration jg1 of this section; if K1 ≥ Q2, calculate the mean value of the maximum and minimum values of the remaining Jx as the accident interval duration jg1 of this section, that is ;

[0068] where both Q1 and Q2 are preset coefficients, and relevant staff can adjust their values according to actual needs;

[0069] P3: Calculate the hazard coefficient WX1 of this section through the formula (jg1 × α1 + H × α2) × θ2 = WX1;

[0070] Quantify the hazard coefficient WX1 and the target person time difference C1 and take their values for calculation. Through the formula calculate the corresponding adjustment time value where α1, α2, α3, and α4 are all preset coefficients, and θ2 is a correction coefficient. Relevant staff can adjust their values according to actual needs;

[0071] The signal lamp control unit is used to obtain the adjustment time value and increase the corresponding signal lamp time through the controller according to its value;

[0072] The voice playback unit is used to play a voice prompt according to the reminder signal, reminding the target person to pay attention to safety when arriving at the intersection waiting line position and stop to wait. Here, the voice prompt is pre-recorded and implanted, which is prior art and will not be elaborated here;

[0073] Embodiment 2

[0074] On the basis of the embodiment, please refer to Figure 2 shown. The present invention is a smart city traffic monitoring method based on AI technology. This urban traffic monitoring method is executed by an urban traffic monitoring system. The method specifically includes the following steps:

[0075] Step 1: Analyze the acquired traffic monitoring images to identify moving targets. Use a recognition model to identify blind sticks or guide dogs in the moving target images in real-time monitoring. Compare the similarity between the identified blind stick or guide dog images and the blind stick or guide dog images in the image set used to train the recognition model. Moved targets with a similarity greater than the preset value Y1 are identified as visually impaired persons and marked as target persons.

[0076] Step 2: Track and acquire the target person's movement distance based on the target person's outline features, analyze the target person's movement distance to obtain the target person's standard movement speed, combine the traffic light status, traffic light countdown, and the target person's standard movement speed for analysis, and determine whether to generate a reminder signal or an adjustment signal.

[0077] Step 3: The traffic light control unit acquires the adjustment signal, analyzes the historical accident data of the road segment, generates the risk factor corresponding to the road segment, analyzes the risk factor, and generates the adjustment time value based on the analysis results.

[0078] Step 4: Acquire the reminder signal through the voice playback unit, and play a voice prompt according to the reminder signal to remind the target person to pay attention to safety and stop and wait when they arrive at the intersection waiting line.

[0079] The working principle of this invention is as follows: By identifying blind canes or guide dogs in real-time traffic monitoring images, and comparing the similarity of the identified blind cane or guide dog images with those in the image set used for training the recognition model, visually impaired persons are identified and marked as target persons. Then, the movement distance of the target person is tracked based on their contour features. By analyzing the movement distance, the standard movement rate of the target person is obtained. Furthermore, by combining the analysis of traffic light status, traffic light countdown, and the standard movement rate of the target person, a warning signal or adjustment signal is generated. Historical accident data of the road segment is analyzed to generate the corresponding danger coefficient of the road segment. The danger coefficient is combined with the time difference of the target person and an adjustment time value is generated based on the analysis results. The traffic light control unit acquires the adjustment time value and increases the traffic light time according to the corresponding adjustment time value. The voice prompt acquires the warning signal and plays a voice prompt to remind the target person to pay attention to safety when reaching the waiting line position at the intersection.

[0080] The above formulas are all dimensionless calculations. The formulas are derived from software simulations based on a large amount of collected data to obtain the most recent real-world results. The preset parameters and thresholds in the formulas are set by those skilled in the art according to the actual situation.

[0081] The above description is merely a specific embodiment of this application, but the scope of protection of this application is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the scope of the technology disclosed in this application should be included within the scope of protection of this application. Therefore, the scope of protection of this application should be determined by the scope of the claims.

Claims

1. A smart city traffic monitoring system based on AI technology, characterized in that, include: The video data acquisition unit is used to acquire traffic monitoring images and send them to the video recognition unit; The video recognition unit is used to perform frame difference analysis on the images in traffic monitoring to obtain moving targets. Then, it uses a recognition model to identify visually impaired people and marks them as target people. Otherwise, no moving targets are marked. The unit extracts the contour features of the target people and sends them to the data analysis unit. The data analysis unit is used to track and acquire the movement distance of the target person based on the contour features of the target person, analyze the movement distance of the target person, obtain the standard movement speed of the target person, and send the standard movement speed to the data judgment unit. The data judgment unit is used to acquire the traffic light information corresponding to the target person in the same direction and the target person's standard movement speed. The traffic light information includes the traffic light status and the traffic light countdown. By combining and analyzing the traffic light status, the traffic light countdown and the target person's standard movement speed, the unit determines whether to generate a reminder signal or an adjustment signal and transmits the reminder signal to the voice playback unit and the adjustment time judgment unit. The time adjustment determination unit is used to analyze the historical accident data of the acquired road segment, generate the corresponding risk coefficient of the road segment, and combine the risk coefficient with the time difference of the target personnel for analysis. At the same time, it generates an adjustment time value based on the analysis results and sends the adjustment time value to the traffic light control unit. The model generation unit iteratively trains the image set to obtain a recognition model, and then sends it to the video recognition unit. The specific method for obtaining the target personnel's standard movement speed is as follows: S1: Acquire the movement distance of the target personnel every g1 time intervals, and continuously obtain the movement distance of the target personnel n+1 times, and mark them as D1, D2, ..., Dn, D(n+1) respectively. Here, n+1 refers to the number of times the movement distance of the target personnel is acquired, 1≤n; S2: Calculate r1 movement differences of the target personnel using the formula Zr1={D(r1+1)-Dr1} / Dp, and label them as Z1, Z2, ..., Zr1 respectively, where Dp={D1+D2+, ...,+Dr1+D(r1+1)} / r1, 1≤r1≤n; S3: Calculate the movement speed corresponding to the movement difference of the target person r1 using the formula Zr2 / g1=vr1, and label them as v1, v2, ..., vr1 respectively, and obtain their average value vp. Then obtain the values ​​of all vr2 in v1, v2, ..., vr1 that satisfy the judgment formula E1. The judgment formula E1 is |vr2-vp|≤Y2, where Y2 is a preset value, 1≤r2≤r1; Calculate the average of all movement speeds v1, v2, ..., vr1 that satisfy the judgment formula E1, and use it as the standard movement speed vb1 of the target person; The specific methods for generating reminder signals or adjustment signals are as follows: A1: Obtain the real-time location of the target person, mark the closest distance between the target person and the waiting line at the intersection as L1, mark the distance between the waiting lines at the intersections on both sides of the road as L2, and mark the traffic light countdown as Ht3; A2: The estimated time Dt2 for the target person to cross the intersection is calculated using the formula {(L1+L2)∕vb1}×θ1=Dt2, where θ1 is a correction coefficient. When Dt2≥Ht3, an alert signal is generated; When Dt2 < Ht3, the time difference C1 of the target personnel is obtained by the formula Ht3 - Dt2 = C1; when C1 < Y3, an adjustment signal is generated; when C1 ≥ Y3, an alert signal is generated, where Y3 is a preset value. The specific method for obtaining the adjustment time value is as follows: P1: Obtain historical accident data for this road segment within the time range W. The historical accident data includes the number of accidents and the accident interval. The accident interval is the time interval between two accidents. P2: Label the number of accidents and the interval between accidents within the time range W as H and J, respectively. H-1 ; Through formula Calculate the accident interval J within the time range W. H-1 The discrete values, where The accident interval J for this road section within the time range W. H-1x The mean, =(J1+J2+、…、+J H-1 ) / H-1, 1≤x≤H-1; If U ≤ Q1, then is used as the accident interval duration jg1 for this section of the road; if U > Q1, then according to the value of |J x - |, the corresponding J x values are deleted in descending order, and after each deletion, the discrete value U of the remaining J x is recalculated, and at the same time, the number K1 of the deleted J x is recorded until U ≤ Q1 is satisfied; if K1 < Q2, then calculate the mean value of the remaining J x and use it as the accident interval duration jg1 for this section of the road; if K1 ≥ Q2, then calculate the mean value of the maximum and minimum values of the remaining J x as the accident interval duration jg1 for this section of the road, that is , where both Q1 and Q2 are preset coefficients; P3: The risk factor WX1 of this road section is calculated using the formula (jg1×α1+H×α2)×θ2=WX1; The risk factor WX1 and the time difference C1 between the target personnel are dequantified and their values ​​are used for calculation, using the formula... Calculate and obtain the adjustment time value α1, α2, α3 and α4 are preset coefficients, and θ2 is a correction coefficient.

2. The smart city traffic monitoring system based on AI technology according to claim 1, characterized in that, The traffic light control unit is used to adjust the time value. The data is acquired, and the corresponding signal light time is increased by the controller based on the value.

3. The smart city traffic monitoring system based on AI technology according to claim 1, characterized in that, The voice playback unit is used to acquire reminder signals and play voice prompts based on the reminder signals.

4. A smart city traffic monitoring method based on AI technology, wherein the method is executed by the monitoring system according to any one of claims 1 to 3, characterized in that, The method specifically includes the following steps: Step 1: Analyze the acquired traffic monitoring images to identify moving targets. Use a recognition model to identify blind sticks or guide dogs in the moving target images in real-time monitoring. Compare the similarity between the identified blind stick or guide dog images and the blind stick or guide dog images in the image set used to train the recognition model. Moved targets with a similarity greater than the preset value Y1 are identified as visually impaired persons and marked as target persons. Step 2: Track and acquire the target person's movement distance based on the target person's outline features, analyze the target person's movement distance to obtain the target person's standard movement speed, combine the traffic light status, traffic light countdown, and the target person's standard movement speed for analysis, and determine whether to generate a reminder signal or an adjustment signal. Step 3: The traffic light control unit acquires the adjustment signal, analyzes the historical accident data of the acquired road segment, generates the risk coefficient corresponding to the road segment, analyzes the risk coefficient, and generates the adjustment time value based on the analysis results. Step 4: Acquire the reminder signal through the voice playback unit, and play a voice prompt according to the reminder signal to remind the target person to pay attention to safety and stop and wait when they arrive at the intersection waiting line.

Images