Unattended wagon balance system based on visual model

Abstract

The invention discloses an unattended wagon balance system based on a visual model. The system comprises a hardware layer, a control layer and an interaction layer. Wherein the hardware layer is used for collecting vehicle data of a vehicle and transmitting the vehicle data to the control layer; the control layer is used for receiving the vehicle data transmitted by the hardware layer, analyzing the vehicle data based on a target visual model to obtain a control instruction, and transmitting the control instruction to the interaction layer; the interaction layer is used for receiving the control instruction transmitted by the control layer and transmitting the control instruction to the hardware layer; and the hardware layer is also used for receiving the control instruction transmitted by the interaction layer and performing control through the control instruction. The weighbridge process is automatically completed through the hardware layer, the control layer and the interaction layer, a physical sensor is not needed, interference of environmental factors or human factors is avoided, and therefore the recognition precision is improved, and the weighing result is more accurate.

Description

Technical Field

[0001] This invention relates to the field of image processing technology, and in particular to an unattended weighbridge system based on a visual model. Background Technology

[0002] Weighing and measurement are core components of internal logistics management and trade settlement within enterprises, and their accuracy and impartiality are paramount. Traditional weighbridge processes heavily rely on manual operation, resulting in inefficiency, high risks of human error, and potential management loopholes. Therefore, an unattended weighbridge system integrating automated equipment and software can address these issues by achieving unmanned, intelligent, and efficient weighing processes.

[0003] In related technologies, unattended weighbridge systems include the weighbridge itself, a vehicle identification system (such as an RFID reader or license plate recognition camera), a barrier gate, infrared beam detectors, monitoring equipment, a voice prompt system, and a central data processing and storage server. Its basic workflow is as follows: vehicles enter the weighing area using their identification tags; the system automatically identifies the vehicle information, guides the vehicle to a complete stop on the weighbridge for weighing, then collects data, saves records, and activates the barrier gate to allow passage, all under video surveillance. However, the above solution relies on physical sensors (such as infrared beam detectors), which are susceptible to environmental factors (such as rain, snow, or fog) or human interference, leading to decreased identification accuracy or even misjudgments, resulting in inaccurate weighing results. Summary of the Invention

[0004] The present invention proposes an unattended weighbridge system based on a visual model to solve the technical problem of inaccurate weighing results in related technologies.

[0005] To address this, the present invention proposes an unattended weighbridge system based on a visual model, which can automatically complete the weighing process through a hardware layer, a control layer, and an interaction layer. The system analyzes vehicle data based on the target visual model to obtain control commands, eliminating the need for physical sensors and avoiding interference from environmental or human factors. This improves recognition accuracy and makes the weighing results more accurate.

[0006] To achieve the above objectives, this invention proposes, in one aspect, an unattended weighbridge system based on a visual model. The system includes a hardware layer, a control layer, and an interaction layer, wherein... The hardware layer is used to collect vehicle data and transmit the vehicle data to the control layer. The control layer is used to receive the vehicle data transmitted by the hardware layer, analyze the vehicle data based on the target visual model to obtain control commands, and transmit the control commands to the interaction layer. The interaction layer is used to receive the control instructions transmitted by the control layer and transmit the control instructions to the hardware layer; The hardware layer is also used to receive the control commands transmitted by the interaction layer and to perform control through the control commands.

[0007] The unattended weighbridge system based on a vision model according to embodiments of the present invention may also have the following additional technical features: In one embodiment of the present invention, the hardware layer includes a license plate recognition camera, multiple location recognition cameras, a barrier gate assembly, a weighbridge, a display screen, and a voice broadcasting system, wherein, The license plate recognition camera includes a first license plate recognition camera and a second license plate recognition camera installed at the entrance of the weighbridge, used to capture a first image of the vehicle; The plurality of position recognition cameras include a first position recognition camera installed at the entrance of the weighbridge and a second position recognition camera installed at the exit of the weighbridge, used to capture multiple second images of the vehicle; The barrier assembly includes a front barrier and a rear barrier, which are installed at the entrance and exit of the weighbridge, respectively, to control the vehicles to enter and exit the weighbridge. The weighbridge is used to acquire the weighing data of the vehicle; The display screen is installed on the column next to the weighbridge and is used to display prompts and weighing data; The voice broadcasting system is installed on the column next to the weighbridge and is used to broadcast voice prompts.

[0008] In one embodiment of the present invention, the control layer includes a host computer, a barrier gate control board, and an AI box, wherein, The AI ​​box is connected to the license plate recognition camera and the location recognition camera. Based on the target visual model, it analyzes the first image and / or the multiple second images to obtain the corresponding recognition result, and transmits the recognition result to the host computer. The host computer is used to generate the control command based on the recognition result, acquire the weighing data transmitted by the weighbridge, and calculate and store the weighing data. The barrier gate control board is used to receive the barrier gate control command in the control instructions, and control the barrier gate component based on the barrier gate control command.

[0009] In one embodiment of the present invention, when a vehicle enters the recognition area, a first image of the vehicle is captured by the first license plate recognition camera or the second license plate recognition camera; The first image is transmitted to the AI ​​box, and the first image is analyzed by the target visual model to obtain the first recognition result; If the host computer determines that the first recognition result is successful, it sends a front barrier lifting command to the barrier control board, and at the same time broadcasts the first voice prompt information through the voice broadcast system and displays the first prompt information on the display screen. If the host computer determines that the first recognition result is a recognition failure, it will broadcast a second voice prompt through the voice broadcasting system and display the second prompt on the display screen.

[0010] In one embodiment of the present invention, in response to the gate control panel receiving the front gate lifting command, the front gate is opened, the vehicle enters the weighbridge area, and multiple second images are captured in real time by the position recognition camera; The second image is transmitted to the AI ​​box, and the second image is analyzed by the target visual model to obtain a second recognition result; If the host computer determines that the second identification result is accurate, it sends a front barrier lowering command to the barrier control board, and at the same time broadcasts a third voice prompt through the voice broadcast system and displays the third prompt on the display screen. If the host computer determines that the first recognition result is an abnormal location, it will broadcast a fourth voice prompt through the voice broadcasting system and display the fourth prompt on the display screen.

[0011] In one embodiment of the present invention, the first identification result further includes license plate information, wherein, In response to the gate control panel receiving the command to lower the front gate arm, the front gate arm is lowered. The host computer sends a weighing command to the weighbridge, and the weighbridge starts weighing to obtain the first weighing data. The host computer determines the weighing type of the vehicle and associates and stores the first weighing data, the license plate information, and the weighing type in the database; The host computer broadcasts the fifth voice prompt message through the voice broadcast system and displays the fifth prompt message and the first weighing data on the display screen.

[0012] In one embodiment of the present invention, the host computer sends a rear barrier lifting command to the barrier control board. In response to the barrier control board receiving the rear barrier lifting command, the rear barrier is opened and the vehicle leaves the weighbridge. The AI ​​box detects the vehicle's position in real time through the location recognition camera. When the vehicle leaves the weighbridge, the AI ​​box sends the complete weighing result to the host computer. In response to the host computer receiving the complete weighing result sent by the AI ​​box, the host computer sends a rear gate lowering command to the gate control board, and broadcasts the sixth voice prompt through the voice broadcast system, and displays the sixth prompt on the display screen.

[0013] In one embodiment of the present invention, when the vehicle returns to be weighed after loading goods, the above steps are repeated, and the host computer matches the first weighing data of the vehicle in the database with the license plate information; After the weighbridge completes the weighing, the host computer obtains the second weighing data of the vehicle. The host computer obtains the corresponding net weight data through the first weighing data and the second weighing data of the vehicle; The host computer associates and stores the first weighing data, the second weighing data, and the net weight data in the database.

[0014] In one embodiment of the present invention, the target visual model includes a first visual model, wherein the first visual model includes a first image preprocessing module, a localization module, a character segmentation module, a first feature extraction module, a character recognition module, and an output module, and the recognition result includes a first recognition result; the step of analyzing the first image through the target visual model to obtain the first recognition result includes: analyzing the first image through the first visual model to obtain the first recognition result; The step of analyzing the first image using the first visual model to obtain the first recognition result includes: The first image is preprocessed in multiple dimensions by the first image preprocessing module to obtain the preprocessed first image. The positioning module locates the target location area of ​​the license plate from the preprocessed first image and extracts the target area image corresponding to the target location area from the preprocessed first image. The target region image is segmented using the character segmentation module to obtain a single character image; The first feature extraction module extracts features from the single character image to obtain the corresponding feature vector; The character recognition module identifies the character corresponding to the feature vector to obtain the target character corresponding to the first image. If the character recognition model recognizes the target character, the output module will output the license plate information corresponding to the target character and the recognition result to indicate successful recognition.

[0015] In one embodiment of the present invention, the target visual model includes a second visual model, wherein the second visual model includes a second image preprocessing module, a second feature extraction module, and a judgment module, and the recognition result includes a second recognition result; the step of analyzing the second image through the target visual model to obtain the second recognition result includes: analyzing the second image through the second visual model to obtain the second recognition result; The step of analyzing the second image using the second visual model to obtain the second recognition result includes: The second image preprocessing module preprocesses the multiple second images to obtain preprocessed second images; The second feature extraction module extracts features from the preprocessed second image to obtain the pixel coordinates of the key feature points in the image coordinate system of the preprocessed second image. The judgment module determines the relative position of the vehicle and the weighbridge weighing surface based on the pixel coordinates of the key feature points in the image coordinate system, and obtains the second recognition result.

[0016] This invention discloses a vision-based unattended weighbridge system. The system comprises a hardware layer, a control layer, and an interaction layer. The hardware layer collects vehicle data and transmits it to the control layer. The control layer receives the vehicle data from the hardware layer, analyzes the data based on a target vision model to obtain control commands, and transmits these commands to the interaction layer. The interaction layer receives the control commands from the control layer and transmits them to the hardware layer. The hardware layer also receives the control commands from the interaction layer and performs control operations accordingly. Thus, the weighing process can be automatically completed through the hardware, control, and interaction layers. The analysis of vehicle data based on a target vision model to obtain control commands eliminates the need for physical sensors, avoiding interference from environmental or human factors, thereby improving recognition accuracy and resulting in more accurate weighing results.

[0017] Additional aspects and advantages of the invention will be set forth in part in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. Attached Figure Description

[0018] The above and / or additional aspects and advantages of the present invention will become apparent and readily understood from the following description of the embodiments taken in conjunction with the accompanying drawings, wherein: Figure 1 This is a schematic diagram of the structure of an unattended weighbridge system based on a vision model according to an embodiment of the present invention; Figure 2 This is a schematic diagram of the hardware layer structure according to an embodiment of the present invention; Figure 3 This invention relates to a weighbridge control method for an unattended weighbridge system based on a vision model, according to an embodiment of the present invention. Detailed Implementation

[0019] It should be noted that, unless otherwise specified, the embodiments and features described in the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0020] To enable those skilled in the art to better understand the present invention, the technical solutions of the present invention will be clearly and completely described below with reference to the accompanying drawings of the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort should fall within the scope of protection of the present invention.

[0021] The following description, with reference to the accompanying drawings, describes an unattended weighbridge system based on a vision model according to an embodiment of the present invention.

[0022] Figure 1 This is a schematic diagram of the structure of the unattended weighbridge system based on a vision model according to an embodiment of the present invention.

[0023] like Figure 1 As shown, the system includes a hardware layer, a control layer, and an interaction layer. The hardware layer is used to collect vehicle data and transmit it to the control layer. The control layer is used to receive vehicle data transmitted from the hardware layer, analyze the vehicle data based on the target vision model to obtain control commands, and transmit the control commands to the interaction layer. The interaction layer is used to receive control commands transmitted from the control layer and transmit the control commands to the hardware layer. The hardware layer is also used to receive control commands transmitted from the interaction layer and to perform control through these commands.

[0024] In one embodiment of the present invention, the hardware layer may include a license plate recognition camera, multiple location recognition cameras, a barrier gate assembly, a weighbridge, a display screen, and a voice broadcasting system. Figure 2 This is a schematic diagram of a hardware layer structure proposed in an embodiment of the present invention. Figure 2As shown, the license plate recognition cameras include a first license plate recognition camera 2 and a second license plate recognition camera 8 installed at the weighbridge entrance, used to capture a first image of the vehicle; multiple location recognition cameras include a first location recognition camera 4 installed at the weighbridge entrance and a second location recognition camera 7 installed at the weighbridge exit, used to capture multiple second images of the vehicle; the barrier gate assembly includes a front barrier gate 1 and a rear barrier gate 10, respectively installed at the weighbridge entrance and exit, used to control vehicles to enter and exit the weighbridge; the weighbridge is used to acquire the weighing data of the vehicle; displays 3 and 9 are installed on the pillars beside the weighbridge, used to display prompts and weighing data; voice broadcasting systems 5 and 6 are installed on the pillars beside the weighbridge, used to broadcast voice prompts.

[0025] In one embodiment of the present invention, the first license plate recognition camera 2 and the second license plate recognition camera 8 are oriented towards the direction in which vehicles enter the weighbridge, and are used to capture license plate information of vehicles entering the weighbridge, providing data support for obtaining license plate information. Furthermore, in one embodiment of the present invention, the weighbridge entrance can also be the weighbridge exit, and the weighbridge exit can also be the weighbridge entrance.

[0026] Furthermore, in one embodiment of the present invention, when a vehicle enters the recognition area, image acquisition by the first license plate recognition camera 2 or the second license plate recognition camera 8 is automatically triggered, and after the first license plate recognition camera 2 or the second license plate recognition camera 8 acquires the first image, the first image can be transmitted to the control layer through the Ethernet interface.

[0027] Furthermore, in one embodiment of the present invention, the first position recognition camera 4 and the second position recognition camera 7 are opposite to the first license plate recognition camera 2 and the second license plate recognition camera 8, respectively, and face the inside of the weighbridge. They are used to recognize multiple second images of the vehicle outline and the weighbridge boundary after the vehicle is put on the weighbridge, and transmit the second images to the control layer through the Ethernet interface.

[0028] Furthermore, in one embodiment of the present invention, the barrier gate assembly includes a front barrier gate and a rear barrier gate, which are respectively installed at the entrance and exit of the weighing area, and adopt an electric lifting structure with a lifting / lowering response time of no more than 3 seconds.

[0029] Furthermore, in one embodiment of the present invention, the aforementioned weighbridge is a weighing device, which can be a digital electronic weighbridge, and the acquired vehicle weighing data is transmitted to the control layer in real time via an RS485 interface.

[0030] Furthermore, in one embodiment of the present invention, the above-mentioned display screen can be a high-definition LED display screen, installed in a position visible to the vehicle driver—the pillar next to the weighbridge, and supports the display of Chinese characters, numbers, and symbols.

[0031] Furthermore, in one embodiment of the present invention, the above-mentioned voice broadcasting system may consist of a tweeter and an audio control module, with adjustable broadcast volume (ensuring that the driver in the vehicle can hear clearly) and support for TTS text-to-speech function.

[0032] Furthermore, in one embodiment of the present invention, the aforementioned control layer may include a host computer, a barrier gate control board, and an AI box. The AI ​​box is connected to a license plate recognition camera and a location recognition camera, analyzes the acquired first image and / or multiple second images based on a target visual model to obtain corresponding recognition results, and transmits the recognition results to the host computer. The host computer is used to generate control commands based on the recognition results, acquire the weighing data transmitted by the weighbridge, and calculate and store the weighing data. The barrier gate control board is used to receive barrier gate control commands in the control commands and control the barrier gate components based on the barrier gate control commands.

[0033] In one embodiment of the present invention, the host computer can be an industrial control computer (CPU not less than i5, memory not less than 8GB), equipped with a Windows 10 embedded system, communicating with the AI ​​box and the weighbridge via Ethernet, and connecting with the gate control board and the voice broadcasting system via serial port.

[0034] Furthermore, in one embodiment of the present invention, the aforementioned barrier gate control board can employ a PLC controller to receive switch control commands ("1" for raising the barrier, "0" for lowering the barrier) sent by a host computer. It then controls the forward and reverse rotation of the barrier gate motor via relays to achieve the raising and lowering actions of the front and rear barrier gates. Simultaneously, the control board feeds back the current barrier gate status ("open" / "closed") to the host computer to ensure that the commands are executed correctly.

[0035] Furthermore, in one embodiment of the present invention, the AI ​​box can be directly physically connected to a weighbridge, a license plate recognition camera, and a location recognition camera to analyze the collected first image and / or second image to obtain the corresponding recognition result, and upload the recognition result to a host computer, thereby realizing a computing mode of "local real-time response + cloud collaborative management".

[0036] Furthermore, in one embodiment of the present invention, the target visual model includes a first visual model and a second visual model. Also, in one embodiment of the present invention, a first recognition result can be obtained by analyzing the first image using the first visual model. In one embodiment of the present invention, the first visual model may include a first image preprocessing module, a localization module, a character segmentation module, a first feature extraction module, a character recognition module, and an output module.

[0037] Furthermore, in one embodiment of the present invention, the method for obtaining a first recognition result by analyzing a first image using a first visual model may include the following steps: Step 1: Perform multi-dimensional preprocessing on the first image using the first image preprocessing module to obtain the preprocessed first image.

[0038] In one embodiment of the present invention, the method for performing multi-dimensional preprocessing on the first image to obtain the preprocessed first image may include: sequentially performing preprocessing on the first image such as grayscale conversion, image enhancement, filtering and denoising, edge sharpening and geometric correction to obtain the preprocessed first image.

[0039] Furthermore, in one embodiment of the present invention, the above-described image preprocessing method can refer to the prior art, and will not be described in detail in the embodiments of the present invention.

[0040] Step 2: Locate the target location area of ​​the license plate from the preprocessed first image using the positioning module, and extract the target area image corresponding to the target location area from the preprocessed first image.

[0041] In one embodiment of the present invention, the localization model can locate the target location region of the license plate from the preprocessed first image using a trained object detection model, and extract the target region image corresponding to the target location region in the preprocessed first image. In one embodiment of the present invention, the trained object detection model can be an SSD (Special Target Detection Model).

[0042] Step 3: The target region image is segmented using the character segmentation module to obtain individual character images.

[0043] In one embodiment of the present invention, the character segmentation module can perform horizontal projection (segmenting lines) and vertical projection (segmenting each character) on the target region image, and use the troughs formed by the gaps between characters to cut the image to obtain a single character image.

[0044] Step 4: Extract features from a single character image using the first feature extraction module to obtain the corresponding feature vector.

[0045] In one embodiment of the present invention, the first feature extraction module described above can extract features from a single character image using a convolutional neural network (CNN) to obtain the corresponding feature vector. In one embodiment of the present invention, the convolutional neural network is trained.

[0046] Step 5: The character recognition module identifies the character corresponding to the feature vector to obtain the target character corresponding to the first image.

[0047] In one embodiment of the present invention, the character recognition module can identify the character corresponding to the feature vector through a ResNet network to obtain the target character corresponding to the first image.

[0048] Step 6: If the character recognition model recognizes the target character, the output module will output the license plate information corresponding to the target character and the recognition result to indicate successful recognition.

[0049] In one embodiment of the present invention, if the recognition result in the character recognition model does not conform to the existing license plate rules, such as the license plate being dirty or obscuring some characters, the output module outputs the recognition result as recognition failure.

[0050] Furthermore, in one embodiment of the present invention, the aforementioned second visual model may include a second image preprocessing module, a second feature extraction module, and a judgment module. In one embodiment of the present invention, the method for analyzing a second image using the second visual model to obtain a second recognition result may include the following steps: Step 101: The second image preprocessing module preprocesses multiple second images to obtain preprocessed second images.

[0051] In one embodiment of the present invention, the method for preprocessing the second image to obtain the preprocessed second image may include: performing noise reduction, grayscale conversion and distortion correction on the second image to obtain the preprocessed second image.

[0052] Furthermore, in one embodiment of the present invention, the above-described image preprocessing method can refer to the prior art, and will not be described in detail in the embodiments of the present invention.

[0053] Step 102: The second feature extraction module extracts features from the preprocessed second image to obtain the pixel coordinates of the key feature points in the image coordinate system.

[0054] In one embodiment of the present invention, the method for extracting features from the preprocessed second image to obtain the pixel coordinates of the corresponding key feature points in the image coordinate system may include: extracting features from the preprocessed second image using a deep learning model to obtain the pixel coordinates of the corresponding key feature points in the image coordinate system. In one embodiment of the present invention, the deep learning model may be trained based on the inputs and outputs of the embodiments of the present invention, such as the YOLO model.

[0055] Furthermore, in one embodiment of the present invention, the aforementioned key feature points may include the wheel center point, the front point of the vehicle body on the central axis, and the rear center point.

[0056] Step 103: The judgment module determines the relative positional relationship between the vehicle and the weighbridge weighing surface based on the pixel coordinates of key feature points in the image coordinate system, and obtains the second recognition result.

[0057] In one embodiment of the present invention, the method for determining the relative positional relationship between the vehicle and the weighbridge weighing surface based on the pixel coordinates of key feature points in the image coordinate system to obtain a second recognition result may include the following steps: Step 1031: Obtain the camera's intrinsic and extrinsic parameter matrices.

[0058] In one embodiment of the present invention, the aforementioned intrinsic parameter matrix may include focal length (fx, fy) and principal point coordinates (cx, cy).

[0059] Furthermore, in one embodiment of the present invention, the aforementioned extrinsic parameter matrix may include a rotation matrix R and a translation vector T.

[0060] Step 1032: Based on the camera's intrinsic and extrinsic parameter matrices, convert the pixel coordinates of key feature points in the image coordinate system into coordinate points in the world coordinate system.

[0061] In one embodiment of the present invention, the method for converting the pixel coordinates of key feature points in the image coordinate system to coordinates in the world coordinate system based on the camera's intrinsic and extrinsic parameter matrices may include: back-projecting the pixel coordinates in the image coordinate system based on the intrinsic parameter matrix to obtain coordinate points in the camera coordinate system, and performing point transformation on the coordinates in the camera coordinate system based on the extrinsic parameter matrix to obtain coordinate points in the world coordinate system. Detailed steps of the above method can be found in existing technologies, and will not be elaborated upon in this embodiment.

[0062] Step 1033: Compare the coordinate points in the world coordinate system with the weighing area of ​​the weighbridge to obtain the second identification result.

[0063] In one embodiment of the present invention, if all coordinate points in the world coordinate system are located within the weighbridge weighing area, the second identification result is successful identification; if there are coordinate points in the world coordinate system located outside the weighbridge weighing area, the second identification result is unsuccessful identification.

[0064] Furthermore, in one embodiment of the present invention, the aforementioned interaction layer is an interface between the hardware layer and the control layer, which can realize information transmission and ensure that the driver clearly understands the process status and cooperates in the operation.

[0065] It should be noted that, in one embodiment of the present invention, the above-mentioned unattended weighbridge system based on a visual model can acquire vehicle images through a license plate recognition camera and multiple location recognition cameras, and obtain control commands by analyzing the vehicle images based on the target visual model. It does not rely on physical sensors, avoids interference from environmental or human factors, thereby improving recognition accuracy and making the weighing results more accurate.

[0066] At the same time, it can improve recognition accuracy and make the weighing results more accurate by not relying on physical sensors and avoiding interference from environmental or human factors.

[0067] Figure 3 This invention proposes a method for controlling an unattended weighbridge system based on a vision model, as an embodiment of the present invention. For example... Figure 3 As shown, the weighing method of the above-mentioned unattended weighbridge system based on a vision model may include the following steps: Step S1: The vehicle enters the recognition area, and the first image of the vehicle is captured by the first license plate recognition camera or the second license plate recognition camera.

[0068] In one embodiment of the present invention, the recognition area can be a preset range from the first license plate recognition camera or the second license plate recognition camera. The preset range can be set as needed, such as 5 meters.

[0069] Step S2: The first image is transmitted to the AI ​​box, and the first image is analyzed by the target vision model to obtain the first recognition result.

[0070] In one embodiment of the present invention, after the first image is transmitted to the AI ​​box, the first image can be analyzed by the first visual model to obtain the first recognition result.

[0071] Step S3: If the host computer determines that the first recognition result is successful, it sends a front barrier lifting command to the barrier control board and simultaneously broadcasts the first voice prompt information through the voice broadcast system and displays the first prompt information on the display screen.

[0072] In one embodiment of the present invention, the first voice prompt and the first prompt can be set as needed. For example, in one embodiment of the present invention, the first voice prompt is "License plate recognition successful, front gate open, please enter the weighing area", and the first prompt is "Please enter".

[0073] Step S4: If the host computer determines that the first recognition result is a recognition failure, it will broadcast the second voice prompt information through the voice broadcasting system and display the second prompt information on the display screen.

[0074] In one embodiment of the present invention, the aforementioned second voice prompt and second prompt can be set as needed. For example, in one embodiment of the present invention, the aforementioned second voice prompt is "License plate recognition failed, please adjust the vehicle position and try again," and the aforementioned second prompt is "Recognition failed, please try again."

[0075] Furthermore, in one embodiment of the present invention, if the host computer determines that the first identification result is a failure, the vehicle is allowed to retry a maximum of a preset number of times. If it still fails, a manual intervention alarm is triggered. In one embodiment of the present invention, the preset number of times can be set as needed, such as 5 times.

[0076] Step S5: In response to the gate control panel receiving the front gate lifting command, the front gate is opened, the vehicle enters the weighbridge area, and multiple second images are captured in real time by the position recognition camera.

[0077] Step S6: The second image is transmitted to the AI ​​box, and the second image is analyzed by the target vision model to obtain the second recognition result.

[0078] In one embodiment of the present invention, after the second image is transmitted to the AI ​​box, the second image can be analyzed by the second visual model to obtain the second recognition result.

[0079] Step S7: If the host computer determines that the second identification result is accurate, it sends a front barrier lowering command to the barrier control board and simultaneously broadcasts a third voice prompt through the voice broadcast system and displays the third prompt on the display screen.

[0080] In one embodiment of the present invention, if the host computer determines that the second identification result is accurate, and at this time both the tires and the vehicle body are within the weighing area of ​​the weighbridge, then a command to lower the front gate is sent to the gate control panel.

[0081] Furthermore, in one embodiment of the present invention, the aforementioned third voice prompt and third prompt can be set as needed. For example, in one embodiment of the present invention, the aforementioned third voice prompt is "Vehicle in position, front gate closed, weighing begins," and the aforementioned third prompt is "Weighing in progress...".

[0082] Step S8: If the host computer determines that the first recognition result is an abnormal position, it will broadcast the third voice prompt information through the voice broadcasting system and display the third prompt information on the display screen.

[0083] In one embodiment of the present invention, if the host computer determines that the first identification result is an abnormal position, the vehicle may be partially on the weighbridge, such as the front wheels or rear wheels exceeding the weighbridge boundary (e.g., the front of the vehicle protruding from the weighbridge body). The third voice prompt information can be broadcast through the voice broadcasting system and displayed on the display screen.

[0084] In one embodiment of the present invention, the aforementioned third voice prompt and third prompt can be set as needed. For example, in one embodiment of the present invention, the aforementioned third voice prompt is "The vehicle is not fully on the weighbridge, please drive forward to the center of the weighbridge," and the aforementioned third prompt is "Please adjust your position."

[0085] Step S9: In response to the gate control panel receiving the command to lower the front gate arm, control the front gate arm to lower.

[0086] In step S10, the host computer sends a weighing command to the weighbridge, and the weighbridge starts weighing to obtain the first weighing data.

[0087] In step S11, the host computer determines the vehicle's weighing type and associates and stores the first weighing data, license plate information, and weighing type in the database.

[0088] In one embodiment of the present invention, the host computer can determine the weighing type of the vehicle by the number of times the vehicle has been weighed. Specifically, in one embodiment of the present invention, if the current vehicle is being weighed for the first time, the weighing type of the vehicle is determined to be gross weight.

[0089] Furthermore, in one embodiment of the present invention, after the host computer determines the weighing type of the vehicle, it associates and stores "first weighing data + license plate information + weighing type + weighing time" in the database.

[0090] In step S12, the host computer broadcasts the fifth voice prompt message through the voice broadcast system and displays the fifth prompt message and the first weighing data on the display screen.

[0091] In one embodiment of the present invention, the fifth voice prompt and the fifth prompt may need to be set. For example, in one embodiment of the present invention, the fifth voice prompt is "Gross weight: 35600kg, weighing complete", and the fifth prompt is "Weighing complete, gross weight: 35600kg, rear gate open, please leave".

[0092] In step S13, the host computer sends a rear barrier lifting command to the barrier control board. Upon receiving the rear barrier lifting command, the barrier control board opens the rear barrier, and the vehicle leaves the weighbridge.

[0093] In step S14, the AI ​​box detects the vehicle's position in real time through the location recognition camera. After the vehicle leaves the weighbridge, the AI ​​box sends the complete weighing result to the host computer.

[0094] In one embodiment of the present invention, the AI ​​box detects the vehicle's position in real time through a location recognition camera. When the vehicle leaves the weighbridge, that is, when the vehicle body and tires are removed from the weighbridge area, the AI ​​box sends the complete weighbridge exit result to the host computer.

[0095] In step S15, in response to the host computer receiving the complete weighing result sent by the AI ​​box, the host computer sends a command to the gate control board to lower the gate arm, and broadcasts the sixth voice prompt message through the voice broadcast system, and displays the sixth prompt message on the display screen.

[0096] In one embodiment of the present invention, the sixth voice prompt and the sixth notification message can be set as needed. For example, in one embodiment of the present invention, the sixth voice prompt is "Vehicle has left, rear gate closed", and the sixth notification message is "Process completed, thank you for using".

[0097] Step S16: When the vehicle returns to weigh after loading goods, repeat the above steps. The host computer matches the vehicle's first weighing data in the database with the license plate information.

[0098] In one embodiment of the present invention, when the vehicle returns to be weighed after loading goods, the above steps S1 to S9 are repeated, and the host computer matches the first weighing data of the vehicle in the database with the license plate information.

[0099] In step S17, after the weighbridge completes the weighing, the host computer obtains the second weighing data of the vehicle.

[0100] In one embodiment of the present invention, the host computer classifies the second weighing of the vehicle as tare weight.

[0101] In step S18, the host computer obtains the corresponding net weight data through the first and second weighing data of the vehicle.

[0102] In one embodiment of the present invention, the host computer can determine the net weight of the vehicle by subtracting the second weighing data from the first weighing data.

[0103] For example, in one embodiment of the present invention, assuming the first weighing data is 35,600 kg and the second weighing data is 15,600 kg, the net weight of the vehicle is 35,600 kg - 15,600 kg = 20,000 kg. In step S19, the host computer associates and stores the first weighing data, the second weighing data, and the net weight data into the database.

[0104] In one embodiment of the present invention, the above steps S12 to S15 are repeated to complete the weighing process.

[0105] This invention discloses a vision-based unattended weighbridge system. The system comprises a hardware layer, a control layer, and an interaction layer. The hardware layer collects vehicle data and transmits it to the control layer. The control layer receives the vehicle data from the hardware layer, analyzes the data based on a target vision model to obtain control commands, and transmits these commands to the interaction layer. The interaction layer receives the control commands from the control layer and transmits them to the hardware layer. The hardware layer also receives the control commands from the interaction layer and performs control operations accordingly. Therefore, the weighing process can be automatically completed through the hardware, control, and interaction layers, without relying on physical sensors, avoiding interference from environmental or human factors, thus improving recognition accuracy and resulting in more accurate weighing results.

[0106] In the description of this specification, the references to terms such as "one embodiment," "some embodiments," "example," "specific example," or "some examples," etc., indicate that a specific feature, structure, material, or characteristic described in connection with that embodiment or example is included in at least one embodiment or example of the present invention. In this specification, the illustrative expressions of the above terms do not necessarily refer to the same embodiment or example. Furthermore, the specific features, structures, materials, or characteristics described may be combined in any suitable manner in one or more embodiments or examples. Moreover, without contradiction, those skilled in the art can combine and integrate the different embodiments or examples described in this specification, as well as the features of different embodiments or examples.

[0107] Furthermore, the terms "first" and "second" are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one of that feature. In the description of this invention, "a plurality of" means at least two, such as two, three, etc., unless otherwise explicitly specified.

Claims

1. A visual model-based unattended weighbridge system, characterized in that, The system comprises a hardware layer, a control layer, and an interaction layer, wherein, The hardware layer is used to collect vehicle data and transmit the vehicle data to the control layer. The control layer is used to receive the vehicle data transmitted by the hardware layer, analyze the vehicle data based on the target visual model to obtain control commands, and transmit the control commands to the interaction layer. The interaction layer is used to receive the control instructions transmitted by the control layer and transmit the control instructions to the hardware layer; The hardware layer is also used to receive the control commands transmitted by the interaction layer and to perform control through the control commands.

2. The system according to claim 1, characterized in that, The hardware layer includes a license plate recognition camera, multiple location recognition cameras, a barrier gate assembly, a weighbridge, a display screen, and a voice broadcast system. The license plate recognition camera includes a first license plate recognition camera and a second license plate recognition camera installed at the entrance of the weighbridge, used to capture a first image of the vehicle; The plurality of position recognition cameras include a first position recognition camera installed at the entrance of the weighbridge and a second position recognition camera installed at the exit of the weighbridge, used to capture multiple second images of the vehicle; The barrier assembly includes a front barrier and a rear barrier, which are installed at the entrance and exit of the weighbridge, respectively, to control the vehicles to enter and exit the weighbridge. The weighbridge is used to acquire the weighing data of the vehicle; The display screen is installed on the column next to the weighbridge and is used to display prompts and weighing data; The voice broadcasting system is installed on the column next to the weighbridge and is used to broadcast voice prompts.

3. The system according to claim 2, characterized in that, The control layer includes a host computer, a barrier gate control board, and an AI box. The AI ​​box is connected to the license plate recognition camera and the location recognition camera. Based on the target visual model, it analyzes the first image and / or the multiple second images to obtain the corresponding recognition result, and transmits the recognition result to the host computer. The host computer is used to generate the control command based on the recognition result, acquire the weighing data transmitted by the weighbridge, and calculate and store the weighing data. The barrier gate control board is used to receive the barrier gate control command in the control instructions, and control the barrier gate component based on the barrier gate control command.

4. The system according to claim 3, characterized in that, When a vehicle enters the recognition area, a first image of the vehicle is captured by the first license plate recognition camera or the second license plate recognition camera. The first image is transmitted to the AI ​​box, and the first image is analyzed by the target visual model to obtain the first recognition result; If the host computer determines that the first recognition result is successful, it sends a front barrier lifting command to the barrier control board, and at the same time broadcasts the first voice prompt information through the voice broadcast system and displays the first prompt information on the display screen. If the host computer determines that the first recognition result is a recognition failure, it will broadcast a second voice prompt through the voice broadcasting system and display the second prompt on the display screen.

5. The system according to claim 4, characterized in that, In response to the gate control panel receiving the front gate lifting command, the front gate is opened, the vehicle enters the weighbridge area, and multiple second images are captured in real time by the position recognition camera; The second image is transmitted to the AI ​​box, and the second image is analyzed by the target visual model to obtain a second recognition result; If the host computer determines that the second identification result is accurate, it sends a front barrier lowering command to the barrier control board, and at the same time broadcasts a third voice prompt through the voice broadcast system and displays the third prompt on the display screen. If the host computer determines that the first recognition result is an abnormal location, it will broadcast a fourth voice prompt through the voice broadcasting system and display the fourth prompt on the display screen.

6. The system according to claim 5, characterized in that, The first recognition result also includes license plate information, wherein, In response to the gate control panel receiving the command to lower the front gate arm, the front gate arm is lowered. The host computer sends a weighing command to the weighbridge, and the weighbridge starts weighing to obtain the first weighing data. The host computer determines the weighing type of the vehicle and associates and stores the first weighing data, the license plate information, and the weighing type in the database; The host computer broadcasts the fifth voice prompt message through the voice broadcast system and displays the fifth prompt message and the first weighing data on the display screen.

7. The system according to claim 6, characterized in that, The host computer sends a rear barrier lifting command to the barrier control board. In response to the barrier control board receiving the rear barrier lifting command, the rear barrier is opened and the vehicle leaves the weighbridge. The AI ​​box detects the vehicle's position in real time through the location recognition camera. When the vehicle leaves the weighbridge, the AI ​​box sends the complete weighing result to the host computer. In response to the host computer receiving the complete weighing result sent by the AI ​​box, the host computer sends a rear gate lowering command to the gate control board, and broadcasts the sixth voice prompt through the voice broadcast system, and displays the sixth prompt on the display screen.

8. The system according to claim 7, characterized in that, When the vehicle returns to be weighed after loading goods, the above steps are repeated, and the host computer matches the first weighing data of the vehicle in the database with the license plate information. After the weighbridge completes the weighing, the host computer obtains the second weighing data of the vehicle. The host computer obtains the corresponding net weight data through the first weighing data and the second weighing data of the vehicle; The host computer associates and stores the first weighing data, the second weighing data, and the net weight data in the database.

9. The system according to claim 4, characterized in that, The target visual model includes a first visual model, wherein the first visual model includes a first image preprocessing module, a localization module, a character segmentation module, a first feature extraction module, a character recognition module, and an output module, and the recognition result includes a first recognition result; the step of analyzing the first image through the target visual model to obtain the first recognition result includes: analyzing the first image through the first visual model to obtain the first recognition result; The step of analyzing the first image using the first visual model to obtain the first recognition result includes: The first image is preprocessed in multiple dimensions by the first image preprocessing module to obtain the preprocessed first image. The positioning module locates the target location area of ​​the license plate from the preprocessed first image and extracts the target area image corresponding to the target location area from the preprocessed first image. The target region image is segmented using the character segmentation module to obtain a single character image; The first feature extraction module extracts features from the single character image to obtain the corresponding feature vector; The character recognition module identifies the character corresponding to the feature vector to obtain the target character corresponding to the first image. If the character recognition model recognizes the target character, the output module will output the license plate information corresponding to the target character and the recognition result to indicate successful recognition.

10. The system according to claim 5, characterized in that, The target visual model includes a second visual model, wherein the second visual model includes a second image preprocessing module, a second feature extraction module, and a judgment module, and the recognition result includes a second recognition result; the step of analyzing the second image through the target visual model to obtain the second recognition result includes: analyzing the second image through the second visual model to obtain the second recognition result; The step of analyzing the second image using the second visual model to obtain the second recognition result includes: The second image preprocessing module preprocesses the multiple second images to obtain preprocessed second images; The second feature extraction module extracts features from the preprocessed second image to obtain the pixel coordinates of the key feature points in the image coordinate system of the preprocessed second image. The judgment module determines the relative position of the vehicle and the weighbridge weighing surface based on the pixel coordinates of the key feature points in the image coordinate system, and obtains the second recognition result.

Images