Method and equipment for detecting actions of vehicle to give way to pedestrians for temporary sidewalk

Abstract

The invention provides a method and a device for detecting a pedestrian commission action of a vehicle for a temporary sidewalk, and the method comprises the steps: S1, obtaining the position image data of a plurality of guide cylinders in a target region at regular time through an image collection module, and constructing a temporary sidewalk region through a central control module, s2, the central control module sets a first vehicle stop detection area and a second vehicle stop detection area on vehicle channels on the two sides of the temporary sidewalk area respectively; S3, the image acquisition module acquires target vehicle block diagrams located in the first vehicle stop detection area and the second vehicle stop detection area; s4, the central control module judges whether a target vehicle block diagram passing through the first vehicle stop detection area and the second vehicle stop detection area has a deceleration stop action or not, and when the target vehicle block diagram does not have the deceleration stop action, the central control module obtains and outputs warning information of the target vehicle. According to the invention, a low-cost and high-efficiency function of detecting the actions of the vehicles to give way to the pedestrians on the temporary sidewalk is realized.

Description

Technical Field

[0001] The present invention belongs to the technical field of traffic devices, and in particular relates to a method and device for detecting a vehicle yielding to pedestrians action on a temporary sidewalk. Background Art

[0002] The technology for detecting vehicles yielding to pedestrians is an important component of intelligent transportation systems, which aims to improve road safety and reduce traffic accidents, especially in pedestrian-dense areas such as crosswalks and intersections. It can monitor the behavior of vehicles and pedestrians in real time and determine whether vehicles comply with the regulations on yielding to pedestrians.

[0003] Under the existing technology, the conventional method of detecting vehicles giving way to pedestrians includes setting up video surveillance devices, such as setting up cameras on specific road sections to monitor road traffic conditions in real time, and using image processing technologies such as edge detection, background modeling, and target tracking to identify and track vehicles in the video, thereby detecting whether the vehicle fails to give way to pedestrians. However, with only conventional image recognition technology, the system needs to perform complex processing of image data during the process of detecting vehicles giving way to pedestrians, including factors such as background noise, light changes, and occlusion, which places high requirements on the image recognition algorithm. At the same time, due to the limitations of image recognition technology, the system may make misjudgments, increasing the cost of later manual review. Summary of the invention

[0004] The present invention aims to provide a method and device for detecting the vehicle yielding to pedestrians on temporary sidewalks, so as to solve the technical problems of high image data processing complexity and low automatic recognition accuracy in conventional methods and devices for detecting the vehicle yielding to pedestrians in the prior art.

[0005] To solve the above problems, the technical solution of the present invention is: a method for detecting a vehicle yielding to pedestrians on a temporary sidewalk, comprising the following steps: S1: The image acquisition module periodically acquires the position image data of several guide tubes in the target area, and the central control module constructs a temporary pedestrian area; S2: The central control module sets a first vehicle stop detection area and a second vehicle stop detection area respectively in the vehicle passages on both sides of the temporary pedestrian area; S3: the image acquisition module acquires a frame diagram of a target vehicle located in the first vehicle stop detection area and the second vehicle stop detection area, and the central control module determines whether the frame diagram of the target vehicle passing through the first vehicle stop detection area and the second vehicle stop detection area has a deceleration and stop action; S4: When the target vehicle block diagram does not have a deceleration and stop action, the central control module obtains and outputs the warning information of the target vehicle.

[0006] Preferably, the step of constructing a temporary pedestrian area in S1 further comprises the following steps: S11: the image acquisition module acquires all guide tube frame diagrams in the target area, and the central control module performs clustering processing on the center points of all the guide tube frame diagrams by using the DBSCAN clustering algorithm, and marks the center point connecting lines of the two groups of guide tube frame diagrams with the shortest interval as the first guide tube combination line segment and the second guide tube combination line segment respectively; S12: The central control module performs clustering processing on the first guide tube combination line segment and the second guide tube combination line segment again through the DBSCAN clustering algorithm, obtains a center point connection line between the first guide tube combination line segment and the second guide tube combination line segment, and records it as a temporary sidewalk detection line segment.

[0007] Preferably, obtaining the block diagram of all guide tubes in the target area in S11 further includes the following steps: S111: a guide tube detection model and a guide tube classification model are pre-stored in the image acquisition module through deep learning training. The image acquisition module obtains the guide tube frame diagram of the guide tube position image data in the target area based on the guide tube detection model, and selects the target guide tube frame diagram from all the guide tube frame diagrams based on the guide tube classification model. The screening criteria of the target guide tube frame diagram include the guide tube color.

[0008] Preferably, in S2, the central control module sets a first vehicle stop detection area and a second vehicle stop detection area on the vehicle passages on both sides of the temporary pedestrian area, respectively, further comprising the following steps: S21: After the central control module obtains the temporary pedestrian detection line segment, a first vehicle stop detection line segment and a second vehicle stop detection line segment parallel to the temporary pedestrian detection line segment are respectively provided at preset intervals in the vertical direction on both sides of the extension direction of the length of the temporary pedestrian detection line segment, and a strip area from the first vehicle stop detection line segment to the temporary pedestrian detection line segment is recorded as a first vehicle stop detection area, and a strip area from the second vehicle stop detection line segment to the temporary pedestrian detection line segment is recorded as a second vehicle stop detection area; The preset spacing lengths between the first vehicle stop detection line segment and the temporary sidewalk detection line segment, and between the second vehicle stop detection line segment and the temporary sidewalk detection line segment are at least greater than the body length of the target vehicle.

[0009] Preferably, in S3, the image acquisition module acquires a frame diagram of a target vehicle located in the first vehicle stop detection area and the second vehicle stop detection area, further comprising the following steps: S31: a vehicle detection model is pre-stored in the image acquisition module through deep learning training, the image acquisition module acquires continuous image frame data from video data captured by an external device source, and pre-processes the image frame data, and the image acquisition module acquires the target vehicle block diagram existing in all image frame data based on the vehicle detection model; S32: The central control module locks and tracks the motion trajectory of the target vehicle frame diagram in the continuous image frame data through the ByteTrack multi-target tracking algorithm, and allocates a set of unique identification IDs to the same target vehicle frame diagram.

[0010] Preferably, in S3, the central control module determines whether the target vehicle passing through the first vehicle stop detection area and the second vehicle stop detection area has a deceleration stop action, further comprising the following steps: S33: The central control module records the driving data of the target vehicle frame diagram that continuously passes through the first vehicle stop detection area and the second vehicle stop detection area, and the driving data includes the frame diagram position of the target vehicle frame diagram with the same identification ID in the first vehicle stop detection area and the second vehicle stop detection area in different frames.

[0011] Preferably, in S3, the central control module determines whether the target vehicle passing through the first vehicle stop detection area and the second vehicle stop detection area has a deceleration stop action, further comprising the following steps: S34: The central control module selects the driving data of the target vehicle frame diagram with the same identification ID every 10 frames, and calculates the overlapping ratio between continuous frame diagrams through the IoU intersection-over-union algorithm for the frame diagram positions in the driving data under different frames. When the overlapping ratio between continuous frame diagrams of the target vehicle frame diagram is less than a preset threshold, the central control module determines that there is no deceleration and stop action when the target vehicle frame diagram passes through the first vehicle stop detection area and the second vehicle stop detection area.

[0012] Preferably, in S4, the central control module acquires and outputs the warning information of the target vehicle, further comprising the following steps: S41: A license plate number recognition model is pre-stored in the image acquisition module through deep learning training. When the central control module determines that there is no deceleration and stop action when the target vehicle block diagram passes through the first vehicle stop detection area and the second vehicle stop detection area, the image acquisition module obtains the license plate number information of the target vehicle in the target vehicle block diagram through the license plate number recognition model.

[0013] Preferably, in S4, the central control module acquires and outputs the warning information of the target vehicle, further comprising the following steps: S42: The image acquisition module packages the license plate information of the target vehicle that has not performed the deceleration and stop action, the image of the vehicle passing through the temporary sidewalk detection segment, the image of the vehicle first entering the vehicle stop detection area, the image of the vehicle exiting the vehicle stop detection area, and the continuous frame video playback address of the vehicle passing through the vehicle stop detection area into an alarm information, and transmits it to the designated web page.

[0014] Based on the same concept, the present invention also provides a vehicle-to-pedestrian-yielding action detection device for a temporary sidewalk, which executes a vehicle-to-pedestrian-yielding action detection method for a temporary sidewalk as described in any one of the above, including: An image acquisition module, the image acquisition module is used to acquire the image data of the guide tube and the vehicle in the target area, and obtain the frame diagram of the guide tube and the frame diagram of the target vehicle respectively; The central control module is used to establish a temporary sidewalk detection line segment, a first vehicle stop detection area and a second vehicle stop detection area, and indirectly calculate and determine the speed range of the target vehicle frame when passing through the first vehicle stop detection area and the second vehicle stop detection area.

[0015] Due to the adoption of the above technical solution, the present invention has the following advantages and positive effects compared with the prior art: The present invention provides a method and device for detecting a vehicle yielding to pedestrians action on a temporary sidewalk. In the process of detecting a vehicle yielding to pedestrians action, first, a temporary sidewalk area can be automatically established in real time based on the position of a temporary guide tube, and a first vehicle stop detection area and a second vehicle stop detection area can be set according to the temporary sidewalk area. Then, the IoU calculation is performed on the frame positions of the vehicles passing through the first vehicle stop detection area and the second vehicle stop detection area, and the speed range of the target vehicle passing through the temporary sidewalk area is indirectly judged, and then it is determined whether the vehicle has a deceleration and stop action. In the present invention, the image recognition technology can simplify the processing complexity of image data under the condition of limited software and hardware resources, and the accuracy of the vehicle yielding to pedestrians action detection can be improved, so as to realize the automatic recognition and judgment of vehicles that do not perform the yielding to pedestrians action, and reduce the burden of later manual review. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1 A flow chart of a method for detecting a vehicle giving way to pedestrians on a temporary sidewalk provided by the present invention; Figure 2 An on-site schematic diagram of a method for detecting a vehicle yielding to pedestrians action on a temporary sidewalk provided by the present invention.

[0017] Explanation of the reference numerals: 1: guide tube; 2: first vehicle stop detection area; 3: second vehicle stop detection area; 4: target vehicle block diagram; 5: guide tube block diagram; 6: first guide tube combination line segment; 7: second guide tube combination line segment; 8: temporary sidewalk detection line segment; 9: first vehicle stop detection line segment; 10: second vehicle stop detection line segment. DETAILED DESCRIPTION

[0018] The following is a further detailed description of a method and device for detecting a vehicle yielding to pedestrians on a temporary sidewalk in conjunction with the accompanying drawings and specific embodiments. The advantages and features of the present invention will become more apparent from the following description and claims.

[0019] First embodiment See also Figure 1 and Figure 2 This embodiment provides a method for detecting a vehicle yielding to pedestrians on a temporary sidewalk, which is used to automatically determine and warn whether a vehicle passing through a temporary sidewalk area performs a deceleration and stop action, and specifically includes the following steps: S1: The image acquisition module periodically obtains the position image data of several guide cylinders 1 in the target area, and the central control module establishes a temporary pedestrian area in the target area based on the position image data of several guide cylinders 1. Among them, the guide cylinder 1 refers to a warning prop temporarily placed at an intersection where no fixed pedestrian sign is demarcated. For example, if a group of guide cylinders 1 are set on both sides of the road at the starting intersection and the ending intersection of a road, it proves that the current road is a temporary pedestrian walkway, and pedestrians have a higher priority for passage. When a vehicle needs to cross the temporary pedestrian walkway, it needs to perform a deceleration and stop action, thereby preventing the vehicle from colliding and injuring pedestrians due to excessive speed. It is worth noting that temporary pedestrian walkways are often set in roads where there is a significant difference in the change of pedestrian flow over time. Therefore, in this embodiment, when the guide cylinder 1 is not set in the road, the road is currently a normal road, and there is no need to perform a deceleration and stop action when the vehicle crosses. When the guide cylinder 1 is set in the road, the road is currently a temporary pedestrian walkway, and the vehicle needs to perform a deceleration and stop action when crossing. Therefore, in this embodiment, the detection device can determine the automatic start and stop of the vehicle deceleration and stop action detection function according to the actual setting state of the guide cylinder 1 in the temporary pedestrian walkway.

[0020] S2: The central control module constructs a first vehicle stop detection area 2 and a second vehicle stop detection area 3 in the vehicle passages on both sides of the temporary pedestrian area. When a vehicle passes through the first vehicle stop detection area 2 and the second vehicle stop detection area 3 successively, it is proved that the vehicle passes through the temporary pedestrian area and is recorded as a target vehicle.

[0021] S3: The image acquisition module obtains the target vehicle frame 4 located in the first vehicle stop detection area 2 and the second vehicle stop detection area 3, and the central control module further determines whether the target vehicle frame 4 passing through the first vehicle stop detection area 2 and the second vehicle stop detection area 3 has a deceleration and stop action.

[0022] S4: When the target vehicle (block 4) passes through the temporary pedestrian area, if there is no deceleration and stop action, the central control module obtains and outputs the warning information of the target vehicle.

[0023] The specific steps and functions of the method for detecting a vehicle yielding to pedestrians on a temporary sidewalk provided by this embodiment will be further described in detail below: Preferably, in this embodiment, the step of constructing a temporary pedestrian area in S1 further includes the following steps: S11: The image acquisition module obtains all guide tube frame diagrams 5 in the target area. The target area refers to the overall area covering the starting intersection and the ending intersection of the road. The central control module clusters the center points of all guide tube frame diagrams 5 through the DBSCAN clustering algorithm. In this embodiment, the length of the road is set to be greater than the width of the road, and when the guide tubes 1 are placed, only one group of guide tubes 1 is placed on the sides of the starting intersection and the ending intersection of the road. Therefore, the two groups of guide tubes 1 with a shorter interval are the starting intersection and the ending intersection of the road. Therefore, the central control module marks the center point connection line of the two groups of guide tube frame diagrams 5 with the shortest interval as the first guide tube combination line segment 6 and the second guide tube combination line segment 7. The first guide tube combination line segment 6 and the second guide tube combination line segment 7 are respectively used to represent the starting intersection and the ending intersection of the temporary sidewalk.

[0024] S12: The central control module performs clustering processing on the first guide tube combined line segment 6 and the second guide tube combined line segment 7 again through the DBSCAN clustering algorithm, obtains the center point connection line of the first guide tube combined line segment 6 and the second guide tube combined line segment 7, and records it as the temporary sidewalk detection line segment 8. Therefore, in this embodiment, the temporary sidewalk area is only selected as a line segment, thereby significantly reducing the amount of calculation in subsequent steps and avoiding several interference factors in the two-dimensional plane area structure. At the same time, the vehicle courtesy pedestrian action detection method provided in this embodiment only involves the analysis of the movement path of the vehicle and the temporary sidewalk. Through the simplified processing of the temporary sidewalk area, the detection accuracy of the vehicle courtesy pedestrian action can also be improved.

[0025] It is worth noting that, in this embodiment, the DBSCAN clustering algorithm is used to obtain the center point of each guide tube frame diagram 5 and the center point of the guide tube combination line segment. The specific implementation method includes using the DBSCAN clustering algorithm to calculate the centroid or geometric midpoint of the frame diagram or line segment, which will not be described in detail in this embodiment.

[0026] Preferably, obtaining all guide tubes in the target area in S11 further includes the following steps: S111: A guide tube detection model and a guide tube classification model are pre-stored in the image acquisition module through deep learning training. The image acquisition module obtains the guide tube frame diagram 5 of the guide tube position image data in the target area based on the guide tube detection model, and screens the target guide tube frame diagram from all the guide tube frame diagrams 5 based on the guide tube classification model.

[0027] Among them, the screening criteria of the target guide tube frame diagram include the color and shape of the guide tube. Taking color as an example, the guide tube 1 includes red, red and black, and black and yellow. Guide tubes 1 of different colors have different warning functions. In this embodiment, only the red guide tube is used to indicate a temporary sidewalk. Therefore, the image acquisition module can accurately select the guide tube frame diagram 5 with the red guide tube as the target through the guide tube classification model, thereby improving the accuracy of the construction of the temporary sidewalk detection segment 8.

[0028] Preferably, in this embodiment, in S2, the central control module sets a first vehicle stop detection area 2 and a second vehicle stop detection area 3 on the vehicle passages on both sides of the temporary pedestrian area, respectively, and further includes the following steps: S21: After the central control module obtains the temporary sidewalk detection line segment 8, a first vehicle stop detection line segment 9 and a second vehicle stop detection line segment 10 parallel to the temporary sidewalk detection line segment 8 are respectively provided at preset intervals in the perpendicular directions on both sides of the length extension direction of the temporary sidewalk detection line segment 8, that is, in the lanes on both sides of the temporary sidewalk detection line segment 8 for one-way or two-way vehicle passage. The strip area from the first vehicle stop detection line segment 9 to the temporary sidewalk detection line segment 8 is recorded as the first vehicle stop detection area 2, and the strip area from the second vehicle stop detection line segment 10 to the temporary sidewalk detection line segment 8 is recorded as the second vehicle stop detection area 3.

[0029] Among them, the preset spacing lengths from the first vehicle stop detection segment 9 to the temporary sidewalk detection segment 8, and from the second vehicle stop detection segment 10 to the temporary sidewalk detection segment 8, are at least greater than the body length of the target vehicle, and are preferably less than or equal to 20 meters. On the premise of ensuring accurate acquisition of the target vehicle block diagram 4 and the vehicle moving path in the first vehicle stop detection area 2 and the second vehicle stop detection area 3, the data calculation amount of the vehicle information farther away from the temporary sidewalk detection segment 8 is reduced, and the misjudgment of the stop and deceleration action of the vehicle farther away from the temporary sidewalk detection segment 8 is reduced.

[0030] Preferably, in this embodiment, in S3, the image acquisition module obtains the target vehicle frame 4 located in the first vehicle stop detection area 2 and the second vehicle stop detection area 3, further comprising the following steps: S31: A vehicle detection model is pre-stored in the image acquisition module through deep learning training. The image acquisition module obtains continuous image frame data from the video data captured by the external device source and pre-processes the image frame data. The image acquisition module obtains the target vehicle block diagram 4 existing in all the image frame data based on the vehicle detection model.

[0031] Among them, the image frame data is preprocessed, specifically including cropping the image frame data to remove unnecessary background areas, graying the image frame data to reduce computational complexity, filtering the image frame data to reduce image noise and enhance image edge features, and normalizing the image frame data to improve image brightness and contrast.

[0032] S32: The central control module locks and tracks the motion trajectory of the target vehicle frame 4 in the continuous image frame data through the ByteTrack multi-target tracking algorithm, and allocates a set of unique identification IDs to the same target vehicle frame 4, that is, realizes the image data tracking function of the target vehicle's driving trajectory.

[0033] Among them, the ByteTrack multi-target tracking algorithm is an algorithm based on a single-stage detector and a multi-target tracker, which can detect the target vehicle frame 4 in each frame, and associate the detection results of the target vehicle frame 4 in different frames, thereby forming a continuous motion trajectory of the target vehicle frame 4 with the corresponding identification ID when passing through the first vehicle stop detection area 2 and the second vehicle stop detection area 3.

[0034] Preferably, in this embodiment, the central control module determines in S3 whether the target vehicle block diagram 4 passing through the first vehicle stop detection area 2 and the second vehicle stop detection area 3 has a deceleration stop action, further comprising the following steps: S33: The central control module records the driving data of the target vehicle frame 4 that passes through the first vehicle stop detection area 2 and the second vehicle stop detection area 3 continuously. The driving data includes the frame positions of the target vehicle frame 4 with the same identification ID in the first vehicle stop detection area 2 and the second vehicle stop detection area 3 in different frames.

[0035] Furthermore, the method further comprises the following steps: S34: The central control module selects the driving data of the target vehicle frame diagram 4 with the same identification ID every 10 frames, and calculates the overlapping ratio between continuous frame diagrams by the IoU intersection-over-union algorithm for the frame diagram positions in the driving data under different frames. When the overlapping ratio between continuous frame diagrams of the target vehicle frame diagram 4 is less than the preset threshold, the central control module determines that there is no deceleration and stop action when the target vehicle frame diagram 4 passes through the first vehicle stop detection area 2 and the second vehicle stop detection area 3.

[0036] Among them, the intersection over union (IoU) is an indicator used to measure the degree of overlap of multiple groups of bounding boxes. Its calculation formula is as follows: represents the intersection area of ​​two bounding boxes, Represents the area of ​​the union of two bounding boxes.

[0037] For example, in this embodiment, if a target vehicle frame 4 with a certain identification ID exists in the image frame data from the 1st frame to the 100th frame continuously within a unit time, the central control module selects its driving data at the 1st frame, that is, its frame position in the first vehicle stop detection area 2 and the second vehicle stop detection area 3 at the 1st frame, and its driving data at the 11th frame, the 21st frame, the 31st frame ..., the 91st frame, and performs an IoU on the frame positions of the target vehicle frame 4 at the 1st frame and the 11th frame. Similarly, the intersection over union (IoU) of the frame positions of the target vehicle frame 4 at the 11th frame and the 21st frame, at the 21st frame and the 31st frame, and at the 81st frame and the 91st frame is respectively calculated, and the overlap ratio between a group of consecutive frames with the smallest value is obtained by comparison. If the overlap ratio between the consecutive frames with the smallest value is less than the preset threshold, it proves that the vehicle speed is too fast when passing through the first vehicle stop detection area 2 and the second vehicle stop detection area 3, thereby indirectly proving that the vehicle did not perform the deceleration and stop action.

[0038] Preferably, in this embodiment, in S4, the central control module obtains and outputs the warning information of the target vehicle, further comprising the following steps: S41: A license plate number recognition model is pre-stored in the image acquisition module through deep learning training. When the central control module determines that there is no deceleration and stop action when the target vehicle block diagram 4 passes through the first vehicle stop detection area 2 and the second vehicle stop detection area 3, the image acquisition module obtains the license plate number information of the target vehicle in the target vehicle block diagram 4 through the license plate number recognition model.

[0039] Furthermore, the method further comprises the following steps: S42: If the vehicle does not decelerate and stop, the image acquisition module generates a continuous frame video of the vehicle passing through the vehicle stop detection area and uploads it to the video storage and playback platform. At the same time, the image acquisition module packages the license plate number information of the target vehicle that has not performed the deceleration and stop action, the image of the vehicle passing through the temporary sidewalk detection segment 8, the image of the vehicle first entering the vehicle stop detection area, the image of the vehicle exiting the vehicle stop detection area and the continuous frame video playback address of the vehicle passing through the vehicle stop detection area into an alarm message, and transmits it to the designated web page.

[0040] In summary, this embodiment provides a method for detecting the vehicle yielding to pedestrians action on a temporary sidewalk. In the process of detecting the vehicle yielding to pedestrians action, first, a temporary sidewalk area can be automatically established in real time based on the position of the temporary guide tube 1, and the first vehicle stop detection area 2 and the second vehicle stop detection area 3 are set according to the temporary sidewalk area. Then, the IoU calculation is performed on the frame positions of the vehicles passing through the first vehicle stop detection area 2 and the second vehicle stop detection area 3, and the speed range of the target vehicle passing through the temporary sidewalk area is indirectly judged, and then it is determined whether the vehicle has a deceleration and stop action. In the present invention, the image recognition technology can simplify the processing complexity of image data under the condition of limited software and hardware resources, and the accuracy of the vehicle yielding to pedestrians action detection can be improved, so as to realize the automatic recognition and judgment of vehicles that do not perform the yielding to pedestrians action, and reduce the burden of later manual review.

[0041] Second embodiment Based on the same concept, the present invention also provides a vehicle yielding to pedestrians action detection device for temporary sidewalks, which is used to execute the vehicle yielding to pedestrians action detection method for temporary sidewalks as described in the first embodiment. The detection device includes an image acquisition module and a central control module.

[0042] The image acquisition module includes a camera and a first processor. The camera is used to collect guide tube image data and vehicle image data in the target area. The first processor is used to obtain the guide tube frame diagram 5 and the target vehicle frame diagram 4 respectively according to the guide tube image data and the vehicle image data.

[0043] The central control module includes a second processor, which is used to establish a temporary sidewalk detection segment 8, a first vehicle stop detection area 2 and a second vehicle stop detection area 3, and realize the block diagram position intersection over union (IoU) calculation when the target vehicle block diagram 4 passes through the first vehicle stop detection area 2 and the second vehicle stop detection area 3, thereby indirectly judging the speed range of the target vehicle block diagram 4 when passing through the first vehicle stop detection area 2 and the second vehicle stop detection area 3.

[0044] It is worth noting that, in this embodiment, both the image acquisition module and the central control module can adopt edge AI devices, and then directly process and analyze the collected guide tube image data and vehicle image data, so as to achieve rapid response of the vehicle courtesy pedestrian action detection function for temporary sidewalks.

[0045] The embodiments of the present invention are described in detail above in conjunction with the accompanying drawings, but the present invention is not limited to the above embodiments. Even if various changes are made to the present invention, if these changes fall within the scope of the claims of the present invention and their equivalents, they still fall within the protection scope of the present invention.

Claims

1. A method for detecting a vehicle yielding to pedestrians on a temporary sidewalk, characterized in that: The steps include: S1: The image acquisition module periodically acquires the position image data of several guide tubes in the target area, and the central control module constructs a temporary pedestrian area; S2: The central control module sets a first vehicle stop detection area and a second vehicle stop detection area respectively in the vehicle passages on both sides of the temporary pedestrian area; S3: the image acquisition module acquires a frame diagram of a target vehicle located in the first vehicle stop detection area and the second vehicle stop detection area, and the central control module determines whether the frame diagram of the target vehicle passing through the first vehicle stop detection area and the second vehicle stop detection area has a deceleration and stop action; S4: When the target vehicle block diagram does not have a deceleration and stop action, the central control module obtains and outputs the warning information of the target vehicle.

2. The method for detecting a vehicle yielding to pedestrians on a temporary sidewalk according to claim 1, characterized in that: The temporary pedestrian area is constructed in S1, further comprising the following steps: S11: the image acquisition module acquires all guide tube frame diagrams in the target area, and the central control module performs clustering processing on the center points of all the guide tube frame diagrams by using the DBSCAN clustering algorithm, and marks the center point connecting lines of the two groups of guide tube frame diagrams with the shortest interval as the first guide tube combination line segment and the second guide tube combination line segment respectively; S12: The central control module performs clustering processing on the first guide tube combination line segment and the second guide tube combination line segment again through the DBSCAN clustering algorithm, obtains a center point connection line between the first guide tube combination line segment and the second guide tube combination line segment, and records it as a temporary sidewalk detection line segment.

3. The method for detecting a vehicle yielding to pedestrians on a temporary sidewalk as claimed in claim 2, characterized in that: In S11, a block diagram of all guide tubes in the target area is obtained, which further includes the following steps: S111: a guide tube detection model and a guide tube classification model are pre-stored in the image acquisition module through deep learning training. The image acquisition module obtains the guide tube frame diagram of the guide tube position image data in the target area based on the guide tube detection model, and selects the target guide tube frame diagram from all the guide tube frame diagrams based on the guide tube classification model. The screening criteria of the target guide tube frame diagram include the guide tube color.

4. The method for detecting a vehicle yielding to pedestrians on a temporary sidewalk as claimed in claim 2, characterized in that: In S2, the central control module sets a first vehicle stop detection area and a second vehicle stop detection area in the vehicle passages on both sides of the temporary pedestrian area, respectively, further comprising the following steps: S21: After the central control module obtains the temporary pedestrian detection line segment, a first vehicle stop detection line segment and a second vehicle stop detection line segment parallel to the temporary pedestrian detection line segment are respectively provided at preset intervals in the vertical direction on both sides of the extension direction of the length of the temporary pedestrian detection line segment, and a strip area from the first vehicle stop detection line segment to the temporary pedestrian detection line segment is recorded as a first vehicle stop detection area, and a strip area from the second vehicle stop detection line segment to the temporary pedestrian detection line segment is recorded as a second vehicle stop detection area; The preset spacing lengths between the first vehicle stop detection line segment and the temporary sidewalk detection line segment, and between the second vehicle stop detection line segment and the temporary sidewalk detection line segment are at least greater than the body length of the target vehicle.

5. The method for detecting a vehicle yielding to pedestrians on a temporary sidewalk according to claim 1, characterized in that: In S3, the image acquisition module acquires a frame diagram of a target vehicle located in the first vehicle stop detection area and the second vehicle stop detection area, further comprising the following steps: S31: a vehicle detection model is pre-stored in the image acquisition module through deep learning training, the image acquisition module acquires continuous image frame data from video data captured by an external device source, and pre-processes the image frame data, and the image acquisition module acquires the target vehicle block diagram existing in all image frame data based on the vehicle detection model; S32: The central control module locks and tracks the motion trajectory of the target vehicle frame diagram in the continuous image frame data through the ByteTrack multi-target tracking algorithm, and allocates a set of unique identification IDs to the same target vehicle frame diagram.

6. The method for detecting a vehicle yielding to pedestrians on a temporary sidewalk as claimed in claim 5, characterized in that: In S3, the central control module determines whether the target vehicle passing through the first vehicle stop detection area and the second vehicle stop detection area has a deceleration stop action, further comprising the following steps: S33: The central control module records the driving data of the target vehicle frame diagram that continuously passes through the first vehicle stop detection area and the second vehicle stop detection area, and the driving data includes the frame diagram position of the target vehicle frame diagram with the same identification ID in the first vehicle stop detection area and the second vehicle stop detection area in different frames.

7. The method for detecting a vehicle yielding to pedestrians on a temporary sidewalk as claimed in claim 6, characterized in that: In S3, the central control module determines whether the target vehicle passing through the first vehicle stop detection area and the second vehicle stop detection area has a deceleration stop action, and further includes the following steps: S34: The central control module selects the driving data of the target vehicle frame diagram with the same identification ID every 10 frames, and calculates the overlapping ratio between continuous frame diagrams through the IoU intersection-over-union algorithm for the frame diagram positions in the driving data under different frames. When the overlapping ratio between continuous frame diagrams of the target vehicle frame diagram is less than a preset threshold, the central control module determines that there is no deceleration and stop action when the target vehicle frame diagram passes through the first vehicle stop detection area and the second vehicle stop detection area.

8. The method for detecting a vehicle yielding to pedestrians on a temporary sidewalk as claimed in claim 1, characterized in that: In S4, the central control module obtains and outputs the warning information of the target vehicle, further comprising the following steps: S41: A license plate number recognition model is pre-stored in the image acquisition module through deep learning training. When the central control module determines that there is no deceleration and stop action when the target vehicle block diagram passes through the first vehicle stop detection area and the second vehicle stop detection area, the image acquisition module obtains the license plate number information of the target vehicle in the target vehicle block diagram through the license plate number recognition model.

9. The method for detecting a vehicle yielding to pedestrians on a temporary sidewalk as claimed in claim 8, characterized in that: In S4, the central control module obtains and outputs the warning information of the target vehicle, and further includes the following steps: S42: The image acquisition module packages the license plate information of the target vehicle that has not performed the deceleration and stop action, the image of the vehicle passing through the temporary sidewalk detection segment, the image of the vehicle first entering the vehicle stop detection area, the image of the vehicle exiting the vehicle stop detection area, and the continuous frame video playback address of the vehicle passing through the vehicle stop detection area into an alarm information, and transmits it to the designated web page.

10. A vehicle yielding to pedestrians action detection device for temporary sidewalks, characterized in that: The method for detecting a vehicle yielding to pedestrians action on a temporary sidewalk according to any one of claims 1 to 9 comprises: An image acquisition module, the image acquisition module is used to acquire the image data of the guide tube and the vehicle in the target area, and obtain the frame diagram of the guide tube and the frame diagram of the target vehicle respectively; The central control module is used to establish a temporary sidewalk detection line segment, a first vehicle stop detection area and a second vehicle stop detection area, and indirectly calculate and determine the speed range of the target vehicle frame when passing through the first vehicle stop detection area and the second vehicle stop detection area.

Images