Traffic violation evidence obtaining method and evidence obtaining system for school bus parking state

Abstract

The invention relates to the field of intelligent traffic monitoring, in particular to a traffic violation evidence obtaining method and evidence obtaining system for the school bus parking state, and the system comprises a vehicle-mounted host which is used for monitoring the state of a school bus STOP plate, sending an instruction to a violation monitoring module according to the state of the STOP plate, receiving monitoring data, generating a monitoring result, and uploading the monitoring result to a police server; the illegal monitoring module is used for collecting video images, calling the parking evidence obtaining sub-module and the driving recognition sub-module according to instructions to generate monitoring data and sending the monitoring data to the vehicle-mounted host, the parking evidence obtaining sub-module is used for analyzing violation conditions when the STOP board is opened and intercepting violation video segments if violation exists, and the driving recognition sub-module is used for identifying the violation video segments when the STOP board is closed and sending the violation video segments to the vehicle-mounted host. Identifying a license plate, and comparing the license plate with a blacklist to generate a comparison result; the police server comprises a traffic police server and a criminal police server and is used for processing and storing the monitoring result uploaded by the vehicle-mounted host. The problems of high cost and evidence obtaining delay are solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of intelligent traffic monitoring, in particular to a traffic violation evidence collection method and system for school bus parking state. BACKGROUND

[0002] In some countries, the safety mechanism of school buses is of great significance to traffic management. When the school bus is parked for students to get on and off, the STOP sign of the school bus will be opened and the warning light will be turned on, and the front and rear vehicles must be completely stopped until the STOP sign is retracted. This regulation aims to ensure the safety of students getting on and off the bus and crossing the road. However, illegal overtaking, not stopping to avoid, and other behaviors still occur frequently, leading to an increase in traffic accident risk. Traditional manual law enforcement or ordinary electronic police cannot fully cover the school bus parking scene, and there is an urgent need for automated and intelligent violation evidence collection means.

[0003] Currently, there are two main ways to set up a school bus traffic violation evidence collection system. One is to set up two main machines independently to run the parking evidence collection system and the driving identification system, and the two machines control the corresponding cameras for evidence collection. This setting method will double the hardware cost, and the independently running system needs independent operation and maintenance, which increases the failure rate (such as increased connection cables and increased power consumption). The other setting method is to set up different algorithms for the parking evidence collection system and the driving identification system, respectively. When detecting the school bus start-stop state, the current parking evidence collection system algorithm is dynamically destroyed, and the driving identification system algorithm is loaded. This setting method involves algorithm restart and destruction, and the model needs to be reloaded from the disk, which takes 1.5-3 seconds (depending on the performance of the storage medium). The delay is further worsened in low temperature environment, and during the algorithm switching period, the camera needs to be reinitialized, resulting in 200-400ms of video frame loss. The switching speed is slow; and frequent destruction / loading of algorithms can easily cause memory leakage or NPU calculation unit crash. The existing technical solutions have the problems of high cost and evidence collection delay. SUMMARY

[0004] The present application aims to provide a traffic violation evidence collection method and system for school bus parking state to solve the problems of high cost and evidence collection delay in the prior art.

[0005] To achieve the above-mentioned purpose, the present application provides a technical solution: a traffic violation evidence collection method for school bus parking state, comprising: S1: Real-time monitoring and judging the state of the STOP sign. If the STOP sign is in the open state, proceed to S2, if the STOP sign is in the closed state, proceed to S3; S2: The illegal monitoring module on the school bus collects video image data in real time, analyzes the violation situation according to the video image data, and if there is a violation, intercepts the violation video segment and sends it to the police server through the vehicle-mounted host; S3: The violation monitoring module on the school bus collects video image data in real time, and performs license plate recognition based on the video image data, and compares the recognized license plate with the blacklist. If the license plate exists in the blacklist, the blacklist license plate image is captured, and the blacklist license plate image, license plate information, and comparison results are sent to the on-board host. The on-board host sends the blacklist license plate image, license plate information, and GPS positioning information to the criminal police server.

[0006] Preferably, in S1, the onboard host obtains the STOP sign status by monitoring the status of the IO port connected to the STOP sign. When the STOP sign status change is detected, the onboard host of the school bus sends a switching instruction and the capture parameters corresponding to the instruction.

[0007] Preferably, in S2, the violation analysis specifically includes: a. Use the Faster R-CNN model to identify and locate the vehicle in the video image and determine its type. If it is not a school bus or special vehicle, proceed to the next step. b. Use the DeepLabV3 model to segment video images, identify road boundaries, sidewalks, and lane markings, and set restricted areas; c. Determine whether there is a vehicle in the restricted area. If there is a vehicle, track the vehicle in the restricted area and determine whether the vehicle is in motion through the continuous trajectory of the vehicle in the video image. If it is in motion, determine that the vehicle is a traffic violation vehicle; d. Extract the key frame of the illegal vehicle at the moment of violation from the video image to form a violation video segment, and extract the license plate information. Send the violation video segment and license plate information to the vehicle host, which then sends it to the traffic police server.

[0008] Preferably, in S3, license plate recognition specifically includes: a. Use the Faster R-CNN model to locate the license plate area in the video, perform perspective transformation correction on tilted license plates, perform affine transformation on deformed license plates, and use the U-Net semantic segmentation network to extract the license plate character area; b. Use the CRNN convolutional recurrent neural network to recognize character sequences in the license plate character area, and perform end-to-end training and optimization of the convolutional recurrent neural network using the CTC loss function; c. Perform format verification and error correction on the recognition results according to the license plate coding rules.

[0009] Further preferably, in S3, the blacklist comparison strategy includes: a. Filter the identified license plates and set a time window to remove duplicates; b. Search for non-duplicate license plates in the blacklist. If the license plate is on the blacklist, the vehicle is considered a blacklisted vehicle. c. Intercept the image corresponding to the blacklisted vehicle from the video image data, and send the image and license plate information to the vehicle host.

[0010] Preferably, when the vehicle host uploads data to the traffic police server or criminal police server, the data reporting exception handling mechanism is as follows: a. When the network is interrupted, unsuccessfully reported events are classified by type and stored in the local disk of the vehicle host; b. After the network is restored, the stored data will be automatically extracted and uploaded to the traffic police server or criminal police server according to priority; c. After receiving the successful upload information from the traffic police server or criminal police server, delete the locally stored reported data.

[0011] The principles and advantages of this solution are: 1. Algorithms for parking evidence collection and vehicle identification are switched based on changes in the STOP sign status. This ensures precise adaptation of different evidence collection functions to the school bus status, ensuring efficient and accurate evidence collection, avoiding delays or missed shots caused by the destruction of old algorithms, and ensuring continuous capture. When the STOP sign is on, the system focuses solely on collecting evidence for "traffic violations," avoiding ineffective monitoring in unnecessary scenarios and reducing resource waste. When the STOP sign is off, the system switches to "blacklist vehicle identification," monitoring vehicles involved in crimes, fleeing the scene, and other illegal vehicles. This enables precise switching of monitoring targets in different scenarios, ensuring that each monitoring task matches the current scenario needs.

[0012] 2. Traffic violation evidence (video) is sent directly to the traffic police server, seamlessly integrating with the traffic police department's traffic enforcement responsibilities and shortening the violation handling process. Blacklisted vehicle information (images, license plates, GPS) is sent to the criminal police server, providing targeted support for the criminal police department's investigation and tracking of vehicles involved in the case, avoiding processing delays caused by cross-departmental information mixing. Through multi-platform targeted transmission, the problem of "data mis-transmission and omission" is resolved, and "different law enforcement departments are connected by scenario", achieving a precise match between data and law enforcement needs, enhancing the collaborative efficiency of different law enforcement links, and forming a closed loop from monitoring, evidence collection, to law enforcement processing.

[0013] Preferably, when analyzing vehicle violations, first screen the vehicle models to reduce invalid judgments and improve the specificity of violation identification. Through area definition, establish a standard violation space to ensure that the restricted area space is defined objectively and uniformly, avoid the problem of "different judgments in the same scene" caused by the ambiguity of artificially set areas, and improve the fairness of the judgment.

[0014] Preferably, by setting a time window for deduplication, repeated generation of multiple same records is avoided, reducing the bandwidth and storage occupation of the vehicle-mounted host and the police server; avoiding "false positives" caused by repeated data during subsequent comparison, ensuring that the server processes "effective new information", and improving the comparison efficiency.

[0015] The application provides another technical scheme: a system for the traffic violation evidence collection method of the school bus parking state according to claim 1, comprising a vehicle-mounted host configured to monitor the school bus STOP sign state in real time, send a switching instruction to the illegal monitoring module according to the STOP sign state, receive monitoring data from the illegal monitoring module, and generate monitoring results and upload the monitoring results to a police server according to the monitoring data; an illegal monitoring module comprising a camera unit, a video data analysis unit and a communication unit, wherein the camera unit comprises a first camera on the front side of the school bus and a second camera on the rear side of the school bus, the camera unit is configured to collect video images, the video data analysis unit is configured to call a parking evidence collection submodule and a driving identification submodule to analyze the video and generate monitoring data according to the instruction of the vehicle-mounted host, and the communication unit is configured to send the monitoring data to the vehicle-mounted host, the parking evidence collection submodule is configured to analyze the violation according to the video images when the STOP sign is in an open state, and if there is a violation, the violation video segment is intercepted, the driving identification submodule is configured to identify the license plate according to the video images when the STOP sign is in a closed state, compare the license plate with a blacklist, and generate a comparison result; a police server connected to the vehicle-mounted host, comprising a traffic police server and a criminal police server, configured to process and store the monitoring results uploaded by the vehicle-mounted host.

[0016] Preferably, the system further comprises a law enforcement terminal, the law enforcement terminal is connected to the police server, comprising a traffic law enforcement terminal and a criminal case terminal, and is configured to allow staff to process the monitoring results uploaded by the vehicle-mounted host.

[0017] Further preferably, the vehicle-mounted host is further configured to store the blacklist, and the law enforcement terminal is further configured to send the updated blacklist to the vehicle-mounted host through the police server.

[0018] The principle and advantages of the present scheme are as follows: 1. Relying on the same host and a group of cameras, the vehicle-mounted host accurately monitors the school bus STOP sign state, sends corresponding instructions to the illegal monitoring module, and realizes the rapid and stable switching of the violation evidence collection algorithm and the driving identification algorithm. The violation evidence collection is started when the school bus is parked, and the driving identification is switched when the school bus is running, which ensures the accurate adaptation of the two evidence collection methods to the school bus state and ensures the efficiency and accuracy of the evidence collection work.

[0019] 2, the vehicle-mounted host is connected with different police servers respectively, a perfect data classification and reporting mechanism is built, the illegal video data during parking and the license plate, vehicle, GPS data during driving can be accurately reported to the traffic police department system and the criminal case law enforcement department system respectively according to the evidence scene and data use, so that the timeliness and accuracy of data transmission are ensured.

[0020] Preferably, the law enforcement terminal also includes querying and processing the uploaded monitoring results in the police server through the terminal, which can also be used to update the blacklist to ensure that the evidence system compares the latest blacklist and ensures the timeliness of the evidence.

[0021] Noun explanation: the blacklist refers to a list of vehicle information pre-entered into the system, suspected of illegal crime or needing to be monitored, usually including the license plate number, vehicle type, associated case information and other key data of the vehicle. BRIEF DESCRIPTION OF DRAWINGS

[0022] Fig. 1 The method flowchart of the application; Fig. 2 The system logic block diagram of the application; Fig. 3 The data flow direction schematic diagram of the application. DETAILED DESCRIPTION

[0023] The following will be further described in detail through specific embodiments: Example 1 Reference Figs. 1-3 A traffic violation evidence collection method for school bus parking state, comprising: S1: real-time monitoring and judging the state of the STOP sign, if the STOP sign is in the open state, then S2 is performed, if the STOP sign is in the closed state, then S3 is performed; In S1, the vehicle-mounted host obtains the state of the STOP sign by listening to the state of the IO port connected to the STOP sign, and when the state of the STOP sign is monitored to change, the vehicle-mounted host of the school bus sends a switching instruction and a corresponding snapshot parameter of the instruction.

[0024] S2: the illegal monitoring module on the school bus collects video image data in real time, analyzes the illegal situation according to the video image data, if there is illegal, intercepts the illegal video segment, and sends the illegal video segment to the traffic police server through the vehicle-mounted host; In S2, the illegal situation analysis specifically includes: a. using Faster R-CNN model to identify and locate the vehicle in the video image, judging the vehicle type of the vehicle, if it is not a school bus or a special vehicle, then the next step is performed; b. The video images are divided by a DeepLabV3 model to identify the road boundary, sidewalk and lane line area, and to set the forbidden zone; c. It is determined whether there is a vehicle in the forbidden zone. If there is a vehicle, the vehicle in the forbidden zone is tracked, and whether the vehicle is in a motion state is determined through the continuous trajectory of the vehicle in the video image. If it is in a motion state, the vehicle is determined to be a violation vehicle; d. The violation moment key frame of the violation vehicle is extracted from the video image to form a violation video segment, and the license plate information is extracted. The violation video segment and the license plate information are sent to the vehicle host, and the vehicle host sends them to the traffic police server.

[0025] S3: The illegal monitoring module on the school bus collects video image data in real time, and performs license plate recognition according to the video image data. The recognized license plate is compared with the blacklist. If it exists in the blacklist, the blacklist license plate picture is intercepted, and the blacklist license plate picture, the license plate information and the comparison result are sent to the vehicle host. The vehicle host sends the blacklist license plate picture, the license plate information and the GPS positioning information to the criminal police server.

[0026] In S3, the license plate recognition specifically includes: a. The license plate region in the video is located by a Faster R-CNN model, and the perspective transformation correction is performed on the inclined license plate. Affine transformation is implemented on the deformed license plate. The U-Net semantic segmentation network is used to extract the license plate character region; b. The character sequence recognition is performed on the license plate character region using the CRNN convolutional recurrent neural network, and the convolutional recurrent neural network is optimized through end-to-end training using the CTC loss function; c. The format verification and error correction are performed on the recognition result according to the license plate coding rule.

[0027] The blacklist comparison strategy includes: a. The recognized license plate is filtered, and a time window is set for deduplication; b. The non-duplicate license plate is queried in the list containing the blacklist. If the license plate exists in the blacklist, the vehicle is determined to be a blacklist vehicle; c. The picture corresponding to the blacklist vehicle is intercepted in the video image data, and the picture and the license plate information are sent to the vehicle host.

[0028] Preferably, when the vehicle host uploads data to the traffic police server or the criminal police server, the data reporting exception handling mechanism is as follows: a. When the network is interrupted, the events that are not successfully reported are stored in the local disk of the vehicle host according to type; b. After the network is restored, the stored data is automatically extracted and transmitted to the traffic police server or the criminal police server according to priority; c. After receiving the upload success information from the traffic police server or the criminal police server, delete the locally stored reported data.

[0029] Embodiment 2 Referring to Figs. 2-3 A traffic violation evidence collection system for school bus parking state, for the above method, comprising The vehicle-mounted host is used for monitoring the STOP sign state of the school bus in real time, issuing switching instructions to the illegal monitoring module according to the STOP sign state, receiving monitoring data from the illegal monitoring module, and generating monitoring results according to the monitoring data and uploading the monitoring results to the police server. The vehicle-mounted host uses an embedded SoC chip, and in this embodiment, hisiv500 is used.

[0030] The illegal monitoring module is in communication connection with the vehicle-mounted host, and the illegal monitoring module comprises a camera unit, a video data analyzer, and a communication unit. The camera unit is used for collecting video images. The video data analyzer comprises a parking evidence collection submodule and a driving identification submodule, and is used for calling the parking evidence collection submodule and the driving identification submodule to analyze the video to generate monitoring data according to the instructions of the vehicle-mounted host. The parking evidence collection submodule is used for analyzing the illegal parking according to the video images when the STOP sign is in the open state, and if there is illegal parking, the illegal parking video segment is intercepted. The driving identification submodule is used for identifying the license plate according to the video images when the STOP sign is in the closed state, comparing the license plate with the blacklist, and generating a comparison result. The communication unit is used for data exchange with the vehicle-mounted host, and specifically, the illegal parking video segment generated by the parking evidence collection submodule is sent to the vehicle-mounted host, and the comparison result generated by the driving identification submodule and the blacklist license plate picture and license plate information of the vehicle confirmed as the blacklist vehicle are sent to the vehicle-mounted host. In this embodiment, the license plate information is the license plate number.

[0031] The camera unit comprises a first camera located at the front side of the school bus and a second camera located at the rear side of the school bus, and a multimedia processing chip is arranged in the camera. In this embodiment, the camera model is D43C27, and the multimedia processing chip is hisiv300.

[0032] Preferably, the communication mode between the vehicle-mounted host and the illegal monitoring module can be wired network communication.

[0033] The police server is in network connection with the vehicle-mounted host, and comprises a traffic police server and a criminal police server, and is used for processing and storing the monitoring results uploaded by the vehicle-mounted host. The network connection mode between the police server and the vehicle-mounted host can be 4G or 5G.

[0034] The system also includes law enforcement terminals connected to the police server network, including traffic enforcement terminals and criminal case terminals, for personnel to process monitoring results uploaded by the vehicle host. The law enforcement terminals can be computers or mobile phones. The connection between the law enforcement terminals and the police server network can be wireless or wired. This embodiment utilizes wireless communication, specifically 4G, 5G, or WiFi.

[0035] Preferably, the vehicle-mounted host is further used to store a blacklist, and the law enforcement terminal is further used to send the updated blacklist to the vehicle-mounted host via the police server.

[0036] The above is only an embodiment of the present invention, and the common knowledge such as the specific technical solutions and / or characteristics in the solution are not described in detail here. It should be pointed out that for those skilled in the art, without departing from the technical solution of the present invention, several variations and improvements can be made, which should also be regarded as the scope of protection of the present invention, and these will not affect the effect of the implementation of the present invention and the practicality of the patent. The scope of protection required by this application shall be based on the content of its claims, and the specific implementation methods and other records in the description can be used to interpret the content of the claims.

Claims

1. A method for collecting traffic violation evidence for a school bus in a parked state, characterized in that: include: S1: Real-time monitoring and judgment of the status of the STOP sign. If the STOP sign is on, proceed to S2; if the STOP sign is off, proceed to S3. S2: The violation monitoring module on the school bus collects video image data in real time and analyzes the violation situation based on the video image data. If a violation occurs, the violation video segment is intercepted and sent to the traffic police server through the on-board host; S3: The violation monitoring module on the school bus collects video image data in real time, and performs license plate recognition based on the video image data, and compares the recognized license plate with the blacklist. If the license plate exists in the blacklist, the blacklist license plate image is captured, and the blacklist license plate image, license plate information, and comparison results are sent to the on-board host. The on-board host sends the blacklist license plate image, license plate information, and GPS positioning information to the criminal police server.

2. The traffic violation evidence collection method for a school bus in a parked state according to claim 1 is characterized by: In S1, the onboard host obtains the STOP sign status by monitoring the IO port status connected to the STOP sign. When the STOP sign status changes, the onboard host of the school bus sends a switching command and the corresponding capture parameters.

3. The traffic violation evidence collection method for a school bus in a parked state according to claim 1 is characterized by: In S2, the violation analysis specifically includes: a. Use the Faster R-CNN model to identify and locate the vehicle in the video image and determine its type. If it is not a school bus or special vehicle, proceed to the next step. b. Use the DeepLabV3 model to segment video images, identify road boundaries, sidewalks, and lane markings, and set restricted areas; c. Determine whether there is a vehicle in the restricted area. If there is a vehicle, track the vehicle in the restricted area and determine whether the vehicle is in motion through the continuous trajectory of the vehicle in the video image. If it is in motion, determine that the vehicle is a traffic violation vehicle; d. Extract the key frame of the illegal vehicle at the moment of violation from the video image to form a violation video segment, and extract the license plate information. Send the violation video segment and license plate information to the vehicle host, which then sends it to the traffic police server.

4. The method for collecting traffic violation evidence for a school bus in a parked state according to claim 1, characterized in that: In S3, license plate recognition specifically includes: a. Use the Faster R-CNN model to locate the license plate area in the video, perform perspective transformation correction on tilted license plates, perform affine transformation on deformed license plates, and use the U-Net semantic segmentation network to extract the license plate character area; b. Use the CRNN convolutional recurrent neural network to recognize character sequences in the license plate character area, and perform end-to-end training and optimization of the convolutional recurrent neural network using the CTC loss function; c. Perform format verification and error correction on the recognition results according to the license plate coding rules.

5. The method for collecting traffic violation evidence for a school bus in a parked state according to claim 4, characterized in that: In S3, the blacklist matching strategies include: a. Filter the identified license plates and set a time window to remove duplicates; b. Search for non-duplicate license plates in the blacklist. If the license plate is on the blacklist, the vehicle is considered a blacklisted vehicle. c. Intercept the image corresponding to the blacklisted vehicle from the video image data, and send the image and license plate information to the vehicle host.

6. The traffic violation evidence collection method for a school bus in a parked state according to claim 1, characterized in that: When the vehicle host uploads data to the traffic police server or criminal police server, the data reporting exception handling mechanism is as follows: a. When the network is interrupted, unsuccessfully reported events are classified by type and stored in the local disk of the vehicle host; b. After the network is restored, the stored data will be automatically extracted and uploaded to the traffic police server or criminal police server according to priority; c. After receiving the successful upload information from the traffic police server or criminal police server, delete the locally stored reported data.

7. A system for collecting traffic violation evidence of a school bus in a parked state according to claim 1, characterized in that: include The vehicle host is used to monitor the status of the school bus STOP sign in real time and issue switching instructions to the violation monitoring module according to the STOP sign status. It is also used to receive monitoring data from the violation monitoring module and generate monitoring results based on the monitoring data and upload them to the police server; The illegal monitoring module includes a camera unit, a video data analysis unit, and a communication unit. The camera unit includes a first camera at the front of the school bus and a second camera at the rear of the school bus. The camera unit is used to collect video images. The video data analysis unit is used to call the parking evidence collection submodule and the driving identification submodule according to the instructions of the vehicle host to analyze the video and generate monitoring data. The communication unit sends the monitoring data to the vehicle host. The parking evidence collection submodule is used to analyze the violation situation based on the video image when the STOP sign is on. If there is a violation, the violation video segment is intercepted. The vehicle identification submodule is used to identify the license plate based on the video image when the STOP sign is off, compare the license plate with the blacklist, and generate a comparison result; The police server is connected to the vehicle host network, including the traffic police server and the criminal police server, and is used to process and store the monitoring results uploaded by the vehicle host for processing.

8. The traffic violation evidence collection system for school bus parking according to claim 7, characterized in that: Also includes law enforcement terminals, The law enforcement terminal is connected to the police server network, and includes a traffic law enforcement terminal and a criminal case terminal, which are used for staff to process the monitoring results uploaded by the vehicle-mounted host.

9. The traffic violation evidence collection system for school bus parking according to claim 8, characterized in that: The vehicle-mounted host is further used to store a blacklist, and the law enforcement terminal is further used to send an updated blacklist to the vehicle-mounted host via a police server.

Images