Video vehicle type identification and dynamic weighing fusion system based on AI
By introducing AI technology and high-precision laser sensors into the video model identification system and dynamic weighing system, the dual high-precision fusion of vehicle models and weight is achieved, solving the problem of data correspondence accuracy in complex vehicles and improving the overall accuracy of the system.
Patent Information
- Application Number
- CN202420990720.6
- Authority / Receiving Office
- CN · China
- Patent Type
- Utility models(China)
- Current Assignee / Owner
- Filing Date
- 2024-05-09
- Publication Date
- 2025-04-25
- Estimated Expiration
- 2034-05-09
AI Technical Summary
The existing video model identification system and dynamic weighing system are difficult to achieve one-to-one correspondence between model data and dynamic weighing data under complex vehicle conditions, resulting in a decrease in overall accuracy.
The AI-based video model identification and dynamic weighing fusion system is adopted, and the dual triggering of high-precision laser and ground dynamic weighing sensors ensures the accurate correspondence between model data and dynamic weighing data. The AI control device processes multi-angle video images and weighing data to achieve dual high-precision fusion of model and weight.
In complex vehicles, the system can ensure one-to-one correspondence between vehicle model data and dynamic weighing data, improve the overall accuracy of the system, and ensure high recognition accuracy of passenger and freight vehicles and high weighing accuracy.
Smart Images
Figure CN222801264U_ABST
Abstract
Description
Technical Field
[0001] The utility model relates to an AI-based video vehicle model recognition and dynamic weighing fusion system, belonging to the technical field of vehicle management equipment. Background Art
[0002] In recent years, video vehicle type recognition systems have been widely used in my country's intelligent transportation, and dynamic weighing systems have always been an indispensable part of highway overload and overlimit management. In practical applications, although both have high accuracy, there is often a problem that the data cannot be matched one by one, especially in complex situations such as heavy traffic or reversing or semi-reversing. The video recognition data cannot match the weighing data, resulting in a serious decrease in overall accuracy. In order to solve this problem in practical applications and improve the overall accuracy of the system, we propose an AI-based video vehicle type recognition and dynamic weighing fusion system to solve this application problem. Utility Model Content
[0003] The purpose of the utility model is to provide an AI-based video vehicle model recognition and dynamic weighing fusion system, which can still ensure the one-to-one correspondence between vehicle model data and dynamic weighing data under complex vehicle conditions, and the system has high overall accuracy and good application effect.
[0004] In order to solve the above technical problems, the utility model adopts the following technical solutions: a video vehicle type recognition and dynamic weighing fusion system based on AI, including a recognition detection device and a ground dynamic weighing sensor, the ground dynamic weighing sensor is laid on the lane, and is used to obtain dynamic vehicle weighing data information, including axle weight, axle number, axle type, vehicle speed and other data; the recognition detection device is arranged on one side of the lane, an AI control device is arranged inside the recognition detection device, and the AI control device is electrically connected to the ground dynamic weighing sensor, and a laser finishing device and a video image acquisition device are arranged on the recognition detection device. The laser finishing device and the video image acquisition camera are electrically connected to the AI control device. The video image acquisition camera can provide license plate recognition and vehicle photos, and can provide vehicle video recording. The laser finishing device is used to obtain vehicle start and end information. The AI control device processes the acquired data, performs AI recognition on the acquired vehicle photos, determines the vehicle model, and calculates the vehicle weight, axle weight and other weight information based on the acquired ground dynamic weighing sensor data, so as to achieve a perfect fusion of high-precision and accurate one-to-one correspondence between the vehicle model and the weight. At the same time, the processing results can be reported to the lane computer in real time through the standard serial port.
[0005] The aforementioned AI-based video vehicle type recognition and dynamic weighing fusion system has three video image acquisition cameras, and the three video image acquisition cameras are dispersedly arranged facing the lane direction, and collect information from multiple angles to ensure that multi-angle information of the front, rear and body of the vehicle can be collected, thereby ensuring the comprehensiveness and accuracy of the information.
[0006] The aforementioned AI-based video vehicle model recognition and dynamic weighing fusion system, the recognition and detection device is provided with an LED information indicator screen, and the LED information indicator screen is electrically connected to the AI control device to assist in vehicle guidance, instructing vehicles with designated license plates not to follow too closely and to pass in an orderly manner, thereby ensuring the system recognition rate.
[0007] The aforementioned AI-based video vehicle type recognition and dynamic weighing fusion system has a base at the bottom of the recognition and detection device, and the base is fixed on the ground to make the fixation more stable.
[0008] The aforementioned AI-based video vehicle type recognition and dynamic weighing fusion system, the ground dynamic weighing sensor includes a first sensor component and a second sensor component, and the first sensor component and the second sensor component are electrically connected to the AI control device, the first sensor component and the second sensor component are both composed of two weighing sensors arranged along the same straight line, and the first sensor component and the second sensor component span the lane and are arranged parallel to each other. The weight data and information collection time obtained by the four weighing sensors can provide data information such as axle weight, number of axles, axle type, vehicle speed, etc. of dynamic passing vehicles.
[0009] In the aforementioned AI-based video vehicle type recognition and dynamic weighing fusion system, a fill light is arranged at the lower part of the recognition and detection device near the lane direction.
[0010] Compared with the prior art, the utility model ensures that it is a valid vehicle through dual triggering of high-precision laser and ground dynamic weighing sensor and stipulates that only when the ground dynamic weighing sensor uploads a complete vehicle weighing data, the system considers the video camera recognition to be valid data, otherwise it is considered to be interference data, so that the one-to-one correspondence between the vehicle model data and the dynamic weighing data can be guaranteed, the uniqueness of the weighing, vehicle model and license plate is completed, and only one frame of data is generated for upload; the utility model has the advantage of high recognition accuracy of passenger and freight vehicles, and still has high weighing accuracy under traffic jams and complex vehicle driving conditions. BRIEF DESCRIPTION OF THE DRAWINGS
[0011] Figure 1 It is a structural schematic diagram of the utility model;
[0012] Figure 2 It is a principle block diagram of the utility model.
[0013] Figure numerals: 1-identification and detection device, 2-ground dynamic weighing sensor, 3-AI control device, 4-laser finishing device, 5-video image acquisition camera, 6-LED information indicator screen, 7-base, 8-first sensor assembly, 9-second sensor assembly, 10-fill light.
[0014] The utility model is further described below in conjunction with the accompanying drawings and specific implementation methods. DETAILED DESCRIPTION
[0015] Embodiment 1 of the utility model: A video vehicle type recognition and dynamic weighing fusion system based on AI, comprising a recognition detection device 1 and a ground dynamic weighing sensor 2, wherein the ground dynamic weighing sensor 2 is laid on the lane, and the ground dynamic weighing sensor 2 provides dynamic vehicle weighing data information, including the vehicle's axle weight, axle number, axle type, and vehicle speed. The recognition detection device 1 is arranged on one side of the lane, and an AI control device 3 is arranged inside the recognition detection device 1, and the AI control device 3 is electrically connected to the ground dynamic weighing sensor 2. A laser tailing device 4 and a video image acquisition camera 5 are arranged on the recognition detection device 1, and the laser tailing device 4 and the video image acquisition camera 5 are electrically connected to the AI control device 3. The video image acquisition camera 5 can provide license plate recognition and multiple vehicle images including the front, rear, and body of the vehicle. photos, and can provide video recording of the vehicle within a few seconds; the laser tailing device 4 is used to provide the starting and ending information of the vehicle; the AI control device 3 processes the data collected from the external devices such as the video image acquisition camera 5, the laser tailing device 7, the ground dynamic weighing sensor 2, etc., the AI control device 3 performs AI recognition on the multi-angle photos, videos and other information of the vehicle, and compares them with the database to determine the specific model of the vehicle, and calculates the weight information such as the vehicle weight and axle weight according to the data obtained by the ground dynamic weighing sensor 2. Through the synchronization protocol and software algorithm, it can ensure that the perfect fusion and accurate one-to-one correspondence of the vehicle model and weight can be achieved under various complex road conditions. At the same time, the processing results can be reported to the lane computer in real time through the standard serial port. AI technology is a prior art and will not be elaborated on here.
[0016] Embodiment 2 of the present utility model: A video vehicle type recognition and dynamic weighing fusion system based on AI, comprising a recognition detection device 1 and a ground dynamic weighing sensor 2, wherein the ground dynamic weighing sensor 2 is laid on the lane, and the ground dynamic weighing sensor 2 provides dynamic vehicle weighing data information, including the vehicle's axle weight, axle number, axle type, and vehicle speed. The recognition detection device 1 is arranged on one side of the lane, and an AI control device 3 is arranged inside the recognition detection device 1, and the AI control device 3 is electrically connected to the ground dynamic weighing sensor 2. The recognition detection device 1 is provided with a laser tailing device 4 and a video image acquisition camera 5, and the laser tailing device 4 and the video image acquisition camera 5 are arranged on the recognition detection device 1. The video image acquisition camera 5 is electrically connected to the AI control device 3. The video image acquisition camera 5 can provide license plate recognition and three vehicle photos including the front, rear and body of the vehicle, and can provide video recording of the vehicle within 5 seconds; the laser tailing device 4 is used to provide vehicle start and end information; the AI control device 3 processes data collected from external devices such as the video image acquisition camera 5, the laser tailing device 7, and the ground dynamic weighing sensor 2. Through the synchronization protocol and software algorithm, it can ensure that the perfect fusion and accurate one-to-one correspondence of the vehicle model and weight can be achieved under various complex road conditions. At the same time, the processing results can be reported to the lane computer in real time through the standard serial port.
[0017] There are three video image acquisition cameras 5, and the three video image acquisition cameras 5 are dispersedly arranged facing the lane direction, and can capture and collect information from multiple angles, ensuring that multi-angle information of the front, rear and body of the vehicle can be collected, thereby ensuring the comprehensiveness and accuracy of the information.
[0018] Embodiment 3 of the present utility model: A video vehicle type recognition and dynamic weighing fusion system based on AI, comprising a recognition detection device 1 and a ground dynamic weighing sensor 2, wherein the ground dynamic weighing sensor 2 is laid on the lane, and the ground dynamic weighing sensor 2 provides dynamic vehicle weighing data information, including the vehicle's axle weight, axle number, axle type, and vehicle speed. The recognition detection device 1 is arranged on one side of the lane, and an AI control device 3 is arranged inside the recognition detection device 1, and the AI control device 3 is electrically connected to the ground dynamic weighing sensor 2. The recognition detection device 1 is provided with a laser tailing device 4 and a video image acquisition camera 5, and the laser tailing device 4 and the video image acquisition camera 5 are arranged on the recognition detection device 1. The video image acquisition camera 5 is electrically connected to the AI control device 3. The video image acquisition camera 5 can provide license plate recognition and three vehicle photos including the front, rear and body of the vehicle, and can provide video recording of the vehicle within 5 seconds; the laser tailing device 4 is used to provide vehicle start and end information; the AI control device 3 processes data collected from external devices such as the video image acquisition camera 5, the laser tailing device 7, and the ground dynamic weighing sensor 2. Through the synchronization protocol and software algorithm, it can ensure that the perfect fusion and accurate one-to-one correspondence of the vehicle model and weight can be achieved under various complex road conditions. At the same time, the processing results can be reported to the lane computer in real time through the standard serial port.
[0019] The identification and detection device 1 is provided with an LED information indicator screen 6, and the LED information indicator screen 6 is electrically connected to the AI control device 3. The LED information indicator screen 6 can assist in vehicle guidance, instructing vehicles with designated license plates not to follow too closely and to pass in an orderly manner, thereby ensuring the system recognition rate.
[0020] Embodiment 4 of the present utility model: A video vehicle type recognition and dynamic weighing fusion system based on AI, comprising a recognition detection device 1 and a ground dynamic weighing sensor 2, wherein the ground dynamic weighing sensor 2 is laid on the lane, and the ground dynamic weighing sensor 2 provides dynamic vehicle weighing data information, including the vehicle's axle weight, axle number, axle type, and vehicle speed. The recognition detection device 1 is arranged on one side of the lane, and an AI control device 3 is arranged inside the recognition detection device 1, and the AI control device 3 is electrically connected to the ground dynamic weighing sensor 2. The recognition detection device 1 is provided with a laser tailing device 4 and a video image acquisition camera 5, and the laser tailing device 4 and the video image acquisition camera 5 are arranged on the recognition detection device 1. The video image acquisition camera 5 is electrically connected to the AI control device 3. The video image acquisition camera 5 can provide license plate recognition and three vehicle photos including the front, rear and body of the vehicle, and can provide video recording of the vehicle within 5 seconds; the laser tailing device 4 is used to provide vehicle start and end information; the AI control device 3 processes data collected from external devices such as the video image acquisition camera 5, the laser tailing device 7, and the ground dynamic weighing sensor 2. Through the synchronization protocol and software algorithm, it can ensure that the perfect fusion and accurate one-to-one correspondence of the vehicle model and weight can be achieved under various complex road conditions. At the same time, the processing results can be reported to the lane computer in real time through the standard serial port.
[0021] The ground dynamic weighing sensor 2 includes a first sensor component 8 and a second sensor component 9, and the first sensor component 8 and the second sensor component 9 are electrically connected to the AI control device 3. The first sensor component 8 and the second sensor component 9 are both composed of two weighing sensors arranged along the same straight line, and the first sensor component 8 and the second sensor component 9 are arranged across the lane and parallel to each other. The AI control device 3 can calculate the axle weight, number of axles, axle type, vehicle speed and other data information of the dynamic passing vehicle through the weight data and information collection time obtained by the four weighing sensors.
[0022] Specifically, a base 7 is provided at the bottom of the identification and detection device 1, and the base 7 is fixed on the ground.
[0023] Specifically, a fill light 10 is disposed at the lower portion of the recognition and detection device 1 near the lane direction, and the fill light 10 can provide fill light when the light is insufficient.
[0024] The working principle of an embodiment of the utility model: the utility model adopts a high-precision laser and ground dynamic weighing sensor dual trigger system. The ground dynamic weighing sensor 2 is laid on the lane. The ground dynamic weighing sensor 2 provides dynamic vehicle weighing data information, including the vehicle's axle weight, axle number, axle type, and vehicle speed. The identification and detection device 1 is set on one side of the lane. The video image acquisition camera 5 can provide license plate recognition and 3 vehicle photos including the front, rear, and body of the vehicle, and can provide video recording of the vehicle within 5 seconds; the laser tailing device 4 is used to provide the vehicle start and end information; the AI control device 3 processes the information from the video image The data collected by external devices such as the acquisition camera 5, the laser tailing device 7, the ground dynamic weighing sensor 2, etc. can ensure the perfect integration and accurate one-to-one correspondence of the vehicle model and weight with high precision under various complex road conditions through the synchronization protocol and software algorithm, and the processing results can be reported to the lane computer in real time through the standard serial port; at night or when the light is dim, the fill light 10 is controlled to turn on to fill light, ensuring that information can be accurately collected when the light is insufficient; when the light is sufficient, the fill light 10 is turned off; the control method of the fill light 10 can adopt custom time control, or adopt the control of adding a light sensor.
Claims
1. An AI-based video vehicle model recognition and dynamic weighing fusion system, characterized in that: The invention comprises an identification detection device (1) and a ground dynamic weighing sensor (2), wherein the ground dynamic weighing sensor (2) is laid on a lane, the identification detection device (1) is arranged on one side of the lane, an AI control device (3) is arranged inside the identification detection device (1), and the AI control device (3) is electrically connected to the ground dynamic weighing sensor (2), and a laser finishing device (4) and a video image acquisition camera (5) are arranged on the identification detection device (1), and the laser finishing device (4) and the video image acquisition camera (5) are electrically connected to the AI control device (3).
2. According to the AI-based video vehicle type recognition and dynamic weighing fusion system of claim 1, it is characterized in that: Three video image acquisition cameras (5) are provided, and the three video image acquisition cameras (5) are dispersedly arranged facing the lane direction.
3. According to the AI-based video vehicle type recognition and dynamic weighing fusion system of claim 1, it is characterized in that: The identification and detection device (1) is provided with an LED information indication screen (6), and the LED information indication screen (6) is electrically connected to the AI control device (3).
4. According to the AI-based video vehicle type recognition and dynamic weighing fusion system of claim 1, it is characterized in that: The bottom of the identification and detection device (1) is provided with a base (7), and the base (7) is fixed on the ground.
5. According to the AI-based video vehicle type recognition and dynamic weighing fusion system of claim 1, it is characterized in that: The ground dynamic weighing sensor (2) comprises a first sensor component (8) and a second sensor component (9), wherein the first sensor component (8) and the second sensor component (9) are arranged across the lane and parallel to each other, and the first sensor component (8) and the second sensor component (9) are electrically connected to the AI control device (3), and the first sensor component (8) and the second sensor component (9) are both composed of two weighing sensors arranged along the same straight line.
6. The AI-based video vehicle type recognition and dynamic weighing fusion system according to claim 1 is characterized in that: A fill light (10) is arranged at the lower part of the identification and detection device (1) close to the lane direction.