Port container truck double-authentication method, weighing detection method, system, equipment and medium

Abstract

The invention provides a port container truck dual-authentication method, a weighing detection method, a system, equipment and a medium, and is applied to the technical field of port container truck weighing. A camera device is deployed behind a wagon balance area to capture a panoramic image of container truck weighing, and the wagon balance area and a rear wheel area of a container truck are identified from the panoramic image; through tracking and detecting the motion trail of the rear wheel of the container truck and analyzing the position relation between the rear wheel area and the wagon balance area, whether the container truck is completely loaded in the weighing process or not is accurately judged, automatic container truck loading detection is achieved, the weighing accuracy is improved, errors caused by manual detection are reduced, the robustness of the system is enhanced, and the working efficiency is improved. The cheating behavior is effectively prevented, and the safety and efficiency of port operation are ensured.

Description

Technical Field

[0001] The present application relates to the technical field of port container truck weighing, and specifically to a port container truck dual authentication method and weighing detection method, system, equipment and medium. Background Art

[0002] At present, container terminals are usually equipped with weighing scales to ensure that container trucks can be accurately weighed when entering and leaving the port, so as to achieve precise measurement of cargo weight, thereby ensuring the efficiency and transparency of logistics management, preventing overloading, and improving the safety and efficiency of port operations.

[0003] However, the accuracy of weighing data is affected by the wheel edge pressing and improper wheel parking position of weighing trucks, and some people even cheat in weighing by taking advantage of such situations, which brings challenges to port management. In order to supervise the accuracy of weighing trucks, the existing port weighing truck detection mainly relies on manual detection and traditional algorithms using sensors to realize detection (such as a weighbridge anti-cheating method based on curve similarity algorithm disclosed in CN117606600A), which has problems such as low efficiency, poor robustness and low reliability.

[0004] Based on this, a new method for weighing container trucks at ports is needed. Summary of the invention

[0005] In view of this, the embodiments of this specification provide a port container truck dual authentication method and a weighing detection method, system, equipment, and medium, which are applied to the field of port container truck weighing technology. A camera device is deployed behind the weighing area to capture a panoramic image of the container truck being weighed, and a deep learning model is used to perform image preprocessing, semantic segmentation, and target detection. The weighing area and the rear wheel area of ​​the container truck are automatically identified, and the movement trajectory of the rear wheels of the container truck is tracked to determine whether the container truck is completely weighed. An intelligent system combining software and hardware is used to perform intelligent identification and real-time monitoring, thereby realizing an automated weighing detection process, ensuring that the port container truck is completely weighed, and eliminating fraud, thereby improving weighing accuracy, ensuring data reliability, and improving the standardization and safety of port operations.

[0006] The embodiments of this specification provide the following technical solutions: The embodiment of this specification provides a method for detecting a container truck weighing at a port based on deep learning, including: The camera device behind the weighbridge area is used to capture the dynamic panoramic view of the container truck weighing process, and obtain the weighing image including the rear wheels of the port container truck and the weighbridge area; Performing semantic segmentation on the weighing image to generate a rear wheel contour mask image, and then performing target detection to obtain the rear wheel area of ​​the container truck; Tracking and detecting the rear wheels of the container truck to obtain the motion trajectory of the rear wheels of the container truck; According to the movement track of the rear wheels of the container truck, the positional relationship between the rear wheel area and the scale area is judged to determine whether the container truck is completely weighed during weighing.

[0007] The embodiment of this specification also provides a port container truck weighing detection system, including: a collection module, an image processing module and a weighing detection module; The acquisition module is used to obtain images of container trucks being weighed at the port, wherein the images of container trucks being weighed include: a panoramic image behind the weighbridge area when the container trucks are being weighed; The image processing module is used to automatically identify the weighbridge area from the weighbridge image; it is also used to perform semantic segmentation on the weighbridge image to generate a rear wheel contour mask image, and then perform target detection to obtain the rear wheel area of ​​the container truck; The weighing detection module is used to track and detect the rear wheels of the container truck to obtain the movement trajectory of the rear wheels of the container truck; According to the movement track of the rear wheels of the container truck, the positional relationship between the rear wheel area and the scale area is judged to determine whether the container truck is completely weighed during weighing.

[0008] Preferably, the port container truck weighing status detection system further comprises: a decision-making judgment module; The decision-making module is used to make a judgment based on the detection result of the weighing detection module. If the container truck is overweight or not fully weighed, the corresponding information will be highlighted and / or a voice warning will be given; Otherwise, the truck weight is displayed and / or the truck weight is announced by voice.

[0009] The embodiment of this specification also provides a port container truck information dual authentication method, including: Use a camera device to capture a panoramic image of the container truck being weighed behind the weighbridge area to obtain images of the container truck being weighed at the port; Automatically identify the rear license plate of the container truck from the weighing image, obtain the container truck information according to the rear license plate, and check whether the container truck information is consistent with the information preset in the operation task; and verifying the container truck information using a vehicle identification device; When the container truck information passes the double authentication, it is inspected using the port container truck weighing inspection method as described above, and the container truck operation information and weighing data are saved.

[0010] The embodiments of the present specification also provide an electronic device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above-mentioned port container truck weighing detection method when executing the computer program.

[0011] The embodiments of the present specification also provide a computer-readable storage medium having a computer program stored thereon, and when the computer program is executed by a processor, the steps of the above-mentioned method for detecting the weighing of container trucks at a port are implemented.

[0012] Compared with the prior art, the at least one technical solution adopted in the embodiments of this specification can achieve the following beneficial effects: By deploying a camera behind the weighbridge area to capture a panoramic image of the container truck being weighed, the weighbridge area and the rear wheel area of ​​the container truck are identified from the panoramic image, and by tracking and detecting the movement trajectory of the rear wheels of the container truck, the positional relationship between the rear wheel area and the weighbridge area is analyzed, so as to accurately determine whether the container truck is fully weighed during weighing, and realize automated container truck weighing detection. This not only improves the accuracy of weighing and reduces the error of manual detection, but also enhances the robustness of the system, effectively prevents cheating, and ensures the safety and efficiency of port operations. BRIEF DESCRIPTION OF THE DRAWINGS

[0013] In order to more clearly illustrate the technical solutions of the embodiments of the present application, the drawings required for use in the embodiments will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0014] Figure 1 This is a schematic diagram of the port container truck weighing detection based on deep learning in this application; Figure 2 This is a schematic diagram of an idle weighbridge in an embodiment of the present application; Figure 3 It is a flow chart of the port container truck weighing detection method based on deep learning in this application; Figure 4 This is a diagram of the weighing test on a container truck in an embodiment of the present application; Figure 5 It is a diagram of an idle weighbridge in an embodiment of the present application; Figure 6 This is a diagram of the installation position of the camera device in the embodiment of the present application; In the figure: 1. container truck; 2. weighbridge; 3. camera device; 4. gate machine; 5. weighbridge edge line. DETAILED DESCRIPTION

[0015] The embodiments of the present application are described in detail below with reference to the accompanying drawings.

[0016] The following describes the implementation methods of the present application through specific examples, and those skilled in the art can easily understand other advantages and effects of the present application from the contents disclosed in this specification. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. The present application can also be implemented or applied through other different specific implementation methods, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present application. It should be noted that, in the absence of conflict, the following embodiments and the features in the embodiments can be combined with each other. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in the field without making creative work belong to the scope of protection of the present application.

[0017] It should be noted that various aspects of the embodiments within the scope of the appended claims are described below. It should be apparent that the aspects described herein can be embodied in a wide variety of forms, and any specific structure and / or function described herein is merely illustrative. Based on the present application, it should be understood by those skilled in the art that an aspect described herein can be implemented independently of any other aspect, and two or more of these aspects can be combined in various ways. For example, any number and aspect described herein can be used to implement the device and / or practice the method. In addition, other structures and / or functionalities other than one or more of the aspects described herein can be used to implement this device and / or practice this method.

[0018] It should also be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present application. The drawings only show components related to the present application rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.

[0019] Additionally, in the following description, specific details are provided to facilitate a thorough understanding of the examples. However, it will be understood by those skilled in the art that the examples can be practiced without these specific details.

[0020] In the daily operation of container terminals, ensuring accurate weighing of container trucks is a key link in logistics management. However, due to behaviors such as irregular wheel parking and wheel edge pressing, the accuracy of weighing is often affected, which poses a challenge to port management.

[0021] At present, the existing port container truck weighing inspection mainly adopts manual inspection and traditional algorithms to prevent cheating, but these methods have obvious limitations: manual inspection is not only inefficient and unreliable, but also easily interfered by human factors; and traditional algorithms, such as the anti-cheating method of weighbridge based on curve similarity algorithm disclosed in patent CN117606600A, its accuracy and robustness cannot meet the current demand for high-precision weighing.

[0022] In view of this, the inventors discovered through research and improvement exploration that the existing detection schemes cannot achieve real-time monitoring and analysis of the wheel motion trajectory for container trucks that dynamically enter the weighing scale. In addition, detecting irregular wheel parking and edge pressing behaviors requires complex algorithms to analyze the relative position between the vehicle and the weighing scale, and the existing algorithms cannot effectively handle these complex situations. Therefore, the existing port container truck weighing detection scheme cannot effectively identify irregular behaviors of container trucks during the weighing process, resulting in damage to the accuracy of the weighing data, which not only affects the efficiency and transparency of logistics management, but also leads to safety issues and economic losses.

[0023] Port container trucks usually consist of two parts: a tractor head and a trailer (trailer or container semi-trailer). According to the size and weight of the goods, a suitable trailer can be selected for transportation. However, due to the narrow space in the weighbridge area and the large size of the container truck, especially when the trailer is loaded with a container, the overall length will increase and cannot accommodate the entire container truck body, making it impossible for the camera device to capture a complete panoramic image of the container truck at one time.

[0024] Based on this, the embodiment of this specification proposes a port container truck weighing detection method based on deep learning: the overall idea is: automatically identify the weighing scale area from the acquired port container truck weighing image, and use the deep learning model to perform semantic segmentation on the image, generate a mask map of the rear wheel contour of the container truck, to identify and isolate the wheel area, which can reduce the amount of data that needs to be processed by the subsequent target detection model, thereby improving the processing speed of the overall system, and accurately locate the rear wheel area of ​​the container truck through target detection technology, and further track the movement trajectory of the rear wheel of the container truck. By analyzing the relative position relationship between the wheel and the weighing scale, it is determined whether the container truck is completely weighed, thereby improving the weighing accuracy, reducing manual intervention, enhancing the robustness of the system, and effectively preventing cheating, ensuring the standardization and safety of port operations.

[0025] The technical solutions provided by various embodiments of the present application are described below in conjunction with the accompanying drawings.

[0026] like Figure 3 As shown, the embodiment of this specification provides a method for detecting container truck weighing at a port based on deep learning, including: Step S1, using a camera device behind the weighbridge area to capture a dynamic panoramic view of the weighing process of the container truck, and obtaining a weighing image containing the rear wheels of the port container truck and the weighbridge area.

[0027] Specifically, Figure 1 As shown in FIG. 1 , a camera device is arranged behind the weighbridge to shoot the rear wheel position of the weighbridge to obtain a panoramic image of the rear wheel position of the weighbridge. Usually, the weighbridge is provided with a warning line or other ground markings that are different from the surrounding environment, such as Figure 2 and Figure 5 The weighbridge edge line 5 is set at the edge of the weighbridge, and the captured panoramic image includes the mark.

[0028] like Figure 1 and 4 As shown, when the container truck 1 enters the weighing scale 2 for weighing, the camera device 3 installed behind the weighing scale 2 is responsible for capturing a panoramic image including the rear wheels of the container truck and the weighing scale logo to obtain a target image. The target image and the panoramic image are obtained based on the camera device at the same fixed position. The target image includes a weighing image including the rear wheels of the port container truck, which is crucial for verifying whether the container truck has completely entered the weighing scale area. If the rear wheels of the container truck are not completely weighed, that is, part of the wheels are not pressed on the sensor of the weighing scale, then the measured weight will be lighter than the actual weight, resulting in an inaccurate weighing result. Therefore, by analyzing these images, the accuracy and integrity of the weighing process can be ensured.

[0029] In the example, only one photo taking point can be set up behind the scale to reduce hardware costs. Compared with solutions that require dedicated hardware equipment, this embodiment is based on standard computer vision hardware, such as cameras and GPU servers, which reduces equipment costs and maintenance costs.

[0030] The camera and lens integrated device can be set up at the head of the crossing safety island (such as Figure 6 As shown in the figure, valid images of the port container truck being weighed are collected from the dynamic video stream, including the images of the port container truck being weighed with its rear wheels and the weighbridge area.

[0031] like Figure 1 and Figure 2 As shown, pictures of weighbridges and trucks entering and leaving container scales at different locations, angles, lighting conditions, weather conditions, etc. can be collected, and pictures of idle weighbridges can be collected from multiple angles as original data sets. Through a large amount of training data, the deep learning model of the present invention has a strong adaptability to various environmental changes, such as lighting, angle, occlusion, etc., to ensure the stability and consistency of the recognition results.

[0032] In order to ensure accurate image capture in low-light environments, a fill light device can be configured near each lane camera to meet OCR image recognition requirements.

[0033] Step S2: semantically segment the weighing image to generate a rear wheel contour mask image, and then perform target detection to obtain the rear wheel area of ​​the container truck.

[0034] Specifically, a deep learning model is used to perform semantic segmentation on the captured images of container trucks being weighed, generate a mask map of the rear wheel outline, separate different objects in the image (including the rear wheels of the container truck), and assign a category label to each pixel.

[0035] The generated mask map is then fed into an object detection model such as RetinaNet or a trained wheel region detection model to accurately identify and locate the rear wheel region of the truck.

[0036] Step S3: Track and detect the rear wheels of the container truck to obtain the motion trajectory of the rear wheels of the container truck.

[0037] Specifically, a Kalman filter-based tracker is used to track the detected rear wheels of the container truck. By tracking the motion trajectory of the rear wheels of the container truck and analyzing the positional relationship between the motion trajectory and the weighing area, the motion trajectory of each target is predicted and managed.

[0038] A Kalman filter-based tracker can also be used to track the detected truck license plates. By tracking the license plate recognition and distinguishing different trucks, the license plate information can be associated with the weighing data of the truck, making it convenient to record and query the weighing history and logistics information of a specific truck.

[0039] Step S4: judging the positional relationship between the rear wheel area and the weighbridge area according to the movement trajectory of the rear wheel of the container truck, and determining whether the container truck is completely on the scale when weighing.

[0040] Specifically, according to the movement trajectory of the rear wheels of the container truck and the relative position of the weighbridge area, it is judged whether the rear wheels of the container truck are completely located in the weighbridge area during weighing, thereby determining whether the container truck is completely weighed.

[0041] In some embodiments, before semantic segmentation of the image, the image is labeled to obtain labeled training samples; The recognition model is trained using the labeled training samples to guide the recognition model to learn the feature representations of the edge of the weighbridge and the rear wheels of the truck; The recognition model includes: introducing an MSCF module into the encoder part of the EfficientSAM model (Efficient Segment Anything Model) to fuse feature maps of different levels; And / or, an AEM module (Attention Enhancement Module) is integrated into the decoder part of the EfficientSAM model to highlight the information of the edge of the weighbridge and the rear wheels of the container truck through an adaptive weight distribution mechanism.

[0042] Specifically, the collected weighing images are manually annotated to generate corresponding XML files, and then organized into a unified format data set.

[0043] The improved EfficientSAM model is used for pre-training, and the pixel-level annotation information is used to guide the model to learn the feature representation of the edge of the scale and the rear wheels of the truck. Here, the MSCF module (Multi-scale Context Fusion) is introduced in the encoder part of EfficientSAM, and feature maps of different levels are fused to improve the model's perception of small-sized or complex contour structures such as the edge of the scale.

[0044] AEM can also be integrated into the decoder to highlight the information of key areas (such as the front and rear wheels of the truck) through an adaptive weight allocation mechanism, suppress background noise, and improve recognition accuracy.

[0045] In some embodiments, before semantic segmentation of the weighing image, the features of the rear wheels of the container truck extracted by the trained recognition model and the labeled training samples are used to train the RetinaNet target detection model to obtain a wheel area detection model for identifying the rear wheels of the container truck.

[0046] Specifically, based on the improved RetinaNet target detection algorithm, a certain number of truck images are collected, and the wheel areas are marked as training samples. The detection model is trained to obtain a trained wheel area detection model.

[0047] To meet the needs of identifying the edge and rear wheels of the scale, adjust the anchor box ratio and size of RetinaNet to adapt to the characteristics of the target object. Perform end-to-end training on the collected dataset. Regularly evaluate model performance on the validation set, monitor algorithm performance through indicators such as accuracy, recall rate, and F1 score, and perform iterative tuning based on feedback results.

[0048] In some embodiments, when semantic segmentation is performed on the upper scale image, the upper scale image is input into an EfficientSAM model for semantic segmentation to generate a rear wheel contour mask image; The rear wheel contour mask image is input into the wheel area detection model for target detection to obtain the position of the rear wheel of the container truck.

[0049] Specifically, semantic segmentation can directly use the general version of the EfficientSAM model, which needs to be fine-tuned; it can also be trained for specific environments and wheels based on its pre-trained model.

[0050] Compared with directly performing target detection on the original image, by first using EfficientSAM to filter the background and non-target areas, the computational burden of the RetinaNet model can be reduced and the inference speed can be improved.

[0051] In some embodiments, the framed camera stream images are passed into a trained model to locate and identify the edge of the weighbridge and the rear wheels of the vehicle.

[0052] In some embodiments, judging the positional relationship between the rear wheel area and the scale area according to the movement trajectory of the rear wheel of the container truck, and determining whether the container truck is completely weighed during weighing, includes: Using a drawing function to draw the lines of the rear wheels of the container truck and the side lines of the weighbridge; Use vector cross product to determine whether the line of the rear wheel of the container truck and the side line of the weighbridge are on the same side; If it is on the same side, measure the distance between the line of the rear wheel of the container truck and the edge line of the scale to determine whether the distance is less than a preset threshold. If so, it means that the container truck is not fully weighed, otherwise, the container truck is weighed correctly.

[0053] Specifically, the drawing function draws lines of the detected wheels and the edge points of the weighbridge on the image to determine the relative positions of the truck and the weighbridge, and uses the vector cross product to determine whether the two line segments are on the same side, thereby determining whether the truck is correctly parked on the weighbridge. If the distance is less than the threshold, it is considered that they overlap and the truck is not completely on the weighbridge; otherwise, it is correctly on the weighbridge.

[0054] In some embodiments, the automatically identifying the weighbridge area from the weighbridge image comprises: Canny edge detection is used to identify the edge lines of the weighbridge, and the edge lines of the weighbridge are used to form a closed figure; wherein the closed figure is the weighbridge area, and the vertices of the closed figure correspond to the end points of the edge lines of the weighbridge, so that the closed figure can be used accurately to identify the location of the weighbridge, and the position of the weighbridge area can be further obtained, which is conducive to position comparison between the wheel area and the weighbridge area.

[0055] In some embodiments, after obtaining the image of the container truck being weighed at the port, the image of the container truck being weighed is preprocessed, including: image denoising, image enhancement, and / or data enhancement under simulation of different environments and severe weather conditions.

[0056] Specifically, the collected images are preprocessed by normalization, data cleaning, image enhancement, etc., so as to ensure the accuracy of the subsequent detection algorithm for weighing the container truck.

[0057] The collected images can also be processed for data enhancement. Since there are many unstable factors in the outdoor environment, various data enhancement methods can be used, such as simulating noise on rainy days and foggy days, to achieve models for different environments and severe weather conditions, thereby enhancing the accuracy of subsequent detection and the robustness of the model.

[0058] In some embodiments, the port container truck weighing detection method further includes: When a truck passes the weighbridge, the truck weight is displayed and / or announced by voice; When the container truck is overweight or not fully weighed, the corresponding information is highlighted and / or a voice warning is given.

[0059] During implementation, when a truck passes the weighbridge, the system will collect truck data in real time and clearly display the truck weight information through the weighing display device; at the same time, the voice prompt device will automatically announce the truck weight, realizing an efficient and convenient weighing process; When the container truck is not fully loaded, an alarm will be issued and processing will be carried out; In some embodiments, when a container truck is overweight, a clear prompt is issued through highlighting or voice warning, etc., so that relevant personnel can take measures quickly, thereby enhancing the safety and reliability of the system.

[0060] Based on the same inventive concept, the present application also provides a port container truck weighing detection system, including: a collection module, an image processing module and a weighing detection module; The acquisition module is used to obtain images of container trucks being weighed at the port, wherein the images of container trucks being weighed include: a panoramic image behind the weighbridge area when the container trucks are being weighed; The image processing module is used to automatically identify the weighbridge area from the weighbridge image; it is also used to perform semantic segmentation on the weighbridge image to generate a rear wheel contour mask image, and then perform target detection to obtain the rear wheel area of ​​the container truck; The weighing detection module is used to track and detect the rear wheels of the container truck to obtain the movement trajectory of the rear wheels of the container truck; According to the movement track of the rear wheels of the container truck, the positional relationship between the rear wheel area and the scale area is judged to determine whether the container truck is completely weighed during weighing.

[0061] Furthermore, the port container truck weighing detection system also includes: a decision-making module, the result after the image analysis module is returned to the front end, and the data is connected with the back-end system to realize data transmission. If the weighing is not complete, an alarm is issued and processing is performed.

[0062] This application realizes efficient and reliable port container truck weighing detection through the collaboration of software and hardware, which can timely discover and handle abnormal situations, simplify the multi-step operations in traditional methods, reduce manual intervention and operational complexity, and improve the efficiency, accuracy and stability of the port weighing system.

[0063] This application does not rely on traditional hardware sensors. It adopts a deep learning model combined with advanced image processing technology. It uses a camera to capture video streams or static pictures of the vehicle weighing process to accurately determine whether the vehicle has been fully weighed. It not only avoids the high cost, easy damage and occlusion risks inherent in hardware sensors, but also significantly improves the adaptability and stability of the detection link in complex environments.

[0064] Based on the same inventive concept, the present application also provides a port container truck information dual authentication method, including: Use a camera device to capture a panoramic image of the container truck being weighed behind the weighbridge area to obtain images of the container truck being weighed at the port; Automatically identify the rear license plate of the container truck from the weighing image, obtain the container truck information according to the rear license plate, and check whether the container truck information is consistent with the information preset in the operation task; and verifying the container truck information using a vehicle identification device; When the container truck information passes the double authentication, it is inspected using the port container truck weighing inspection method as described above, and the container truck operation information and weighing data are saved.

[0065] Specifically, a vehicle identification device can be used to collect and compare vehicle information, including vehicle number, container number, container type, etc. A camera device can be used to collect and detect the license plate of the container truck. After the container truck is weighed, the license plate recognition system and the camera system can be used to perform dual authentication of the container truck information.

[0066] After verifying the identity of the container truck, the container truck signal and the container truck operation task are verified, and then the weighbridge automatically collects the weighed weight of the container truck. When the tare weight exceeds the limit, the driver is given a clear prompt through highlighted display or voice warning, guiding him to take corresponding measures, such as unloading excess cargo, checking the status of the container truck, etc., to ensure the safety and compliance of subsequent operations.

[0067] Once the truck operation information is verified and the tare data collection is completed, the system will automatically save these key information. The lifting mechanism will respond to this signal, and the gate machine 4 will be lifted up, allowing the truck to safely leave the weighbridge area. The entire process is seamlessly connected and efficiently managed.

[0068] In some embodiments, after weighing is completed, the driver will be informed through an integrated voice prompt device that the weighing process has been completed.

[0069] Based on the same inventive concept, the present application also provides an electronic device for a port container truck weighing detection method based on deep learning, or an electronic device based on an integrated circuit layout filling unit insertion method, the electronic device comprising: a processor, a memory and a computer program; wherein The memory is used to store the computer program, and the memory may also be a flash memory. The computer program is, for example, an application program, a functional module, etc. for implementing the above method.

[0070] The processor is used to execute the computer program stored in the memory to implement each step performed by the device in the above method. For details, please refer to the relevant description in the above method embodiment.

[0071] Optionally, the memory can be independent or integrated with the processor.

[0072] When the memory is a device independent of the processor, the device may further include: A bus is used to connect the memory and the processor.

[0073] Based on the same inventive concept, the present application also provides a computer storage medium, which stores computer executable instructions. When the computer executable instructions are executed by a processor, they execute: a port container truck weighing detection method based on deep learning as described in any embodiment of the present application.

[0074] The present invention is tailor-made based on the existing phenomenon of fraud in container truck weighing inspection at ports. It adopts advanced deep learning algorithms and extracts deep features of the edge of the scale and the front and rear wheels of container trucks through methods such as cross-attention codec, thereby significantly improving the accuracy of weighing inspection. Compared with traditional image processing methods, the deep learning model of the present invention can more effectively cope with challenges such as complex backgrounds, lighting changes and occlusions, ensuring the high reliability of the recognition process.

[0075] In addition, the present invention realizes real-time reasoning of deep learning models on high-performance computing hardware, which can quickly process large-scale port container truck images and realize real-time recognition to meet practical application needs.

[0076] In this specification, the same or similar parts between the various embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the embodiments described later, the description is relatively simple, and the relevant parts can be referred to the partial description of the previous embodiments.

[0077] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any changes or substitutions that can be easily thought of by a person skilled in the art within the technical scope disclosed in the present application should be included in the protection scope of the present application. Therefore, the protection scope of the present application shall be subject to the protection scope of the claims.

Claims

1. A method for detecting container truck weighing at a port based on deep learning, characterized in that: include: The camera device behind the weighbridge area is used to capture the dynamic panoramic view of the container truck weighing process, and obtain the weighing image including the rear wheels of the port container truck and the weighbridge area; Performing semantic segmentation on the weighing image to generate a rear wheel contour mask image, and then performing target detection to obtain the rear wheel area of ​​the container truck; Tracking and detecting the rear wheels of the container truck to obtain the motion trajectory of the rear wheels of the container truck; According to the movement track of the rear wheels of the container truck, the positional relationship between the rear wheel area and the scale area is judged to determine whether the container truck is completely weighed during weighing.

2. The method for detecting container trucks on a port scale according to claim 1, characterized in that: Before performing semantic segmentation on the image, the image is labeled to obtain labeled training samples; The recognition model is trained using the labeled training samples to guide the recognition model to learn the feature representations of the edge of the weighbridge and the rear wheels of the truck; The recognition model includes: introducing an MSCF module in the encoder part of the EfficientSAM model to fuse feature maps of different levels; And / or, an AEM module is integrated into the decoder part of the EfficientSAM model to highlight the information of the edge of the weighbridge and the rear wheels of the container truck through an adaptive weight distribution mechanism.

3. The method for detecting container trucks on the port scale according to claim 2 is characterized in that: Before semantic segmentation of the weighing image, the features of the rear wheels of the container truck extracted by the trained recognition model and the labeled training samples are used to train the RetinaNet target detection model to obtain a wheel area detection model for identifying the rear wheels of the container truck.

4. The method for detecting container trucks on the port scale according to claim 3 is characterized in that: When semantic segmentation is performed on the upper scale image, the upper scale image is input into the EfficientSAM model for semantic segmentation to generate a rear wheel contour mask image; The rear wheel contour mask image is input into the wheel area detection model for target detection to obtain the position of the rear wheel of the container truck.

5. The method for detecting container trucks on a port scale according to claim 1, characterized in that: The step of judging the positional relationship between the rear wheel area and the weighbridge area according to the movement trajectory of the rear wheel of the container truck, and determining whether the container truck is completely weighed during weighing, comprises: Using a drawing function to draw the lines of the rear wheels of the container truck and the side lines of the weighbridge; Use vector cross product to determine whether the line of the rear wheel of the container truck and the side line of the weighbridge are on the same side; If it is on the same side, measure the distance between the line of the rear wheel of the container truck and the edge line of the scale to determine whether the distance is less than a preset threshold. If so, it means that the container truck is not fully weighed, otherwise, the container truck is weighed correctly.

6. The method for detecting container trucks on a port scale according to claim 1, characterized in that: The automatic identification of the weighbridge area from the weighbridge image comprises: The Canny edge detection is used to identify the edge lines of the weighbridge, and the edge lines of the weighbridge are used to form a closed figure; wherein the closed figure is the weighbridge area, and the vertices of the closed figure correspond to the end points of the edge lines of the weighbridge.

7. The method for detecting container trucks on a port scale according to claim 1, characterized in that: After obtaining the image of the container truck being weighed at the port, the image of the container truck being weighed is preprocessed, including: image denoising, image enhancement, and / or data enhancement under simulating different environments and severe weather conditions.

8. The method for detecting container trucks on a port scale according to claim 1, characterized in that: The port container truck weighing detection method also includes: When a truck passes the weighbridge, the truck weight is displayed and / or announced by voice; When the container truck is overweight or not fully weighed, the corresponding information is highlighted and / or a voice warning is given.

9. A port container truck weighing detection system, characterized in that: include: Acquisition module, image processing module and weighing detection module; The acquisition module is used to obtain images of container trucks being weighed at the port, and the images of container trucks being weighed include images behind the weighbridge area when the container trucks are weighed; The image processing module is used to automatically identify the weighbridge area from the weighbridge image; it is also used to perform semantic segmentation on the weighbridge image to generate a rear wheel contour mask image, and then perform target detection to obtain the rear wheel area of ​​the container truck; The weighing detection module is used to track and detect the rear wheels of the container truck to obtain the movement trajectory of the rear wheels of the container truck; According to the movement track of the rear wheels of the container truck, the positional relationship between the rear wheel area and the scale area is judged to determine whether the container truck is completely weighed during weighing.

10. A method for double authentication of port container truck information, characterized in that: include: Use a camera device to capture a panoramic image of the container truck being weighed behind the weighbridge area to obtain images of the container truck being weighed at the port; Automatically identify the rear license plate of the container truck from the weighing image, obtain the container truck information according to the rear license plate, and check whether the container truck information is consistent with the information preset in the operation task; and verifying the container truck information using a vehicle identification device; When the container truck information passes the double authentication, the container truck is inspected by the port container truck weighing inspection method as described in any one of claims 1 to 8, and the container truck operation information and weighing data are saved.

11. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the port container truck weighing detection method as described in any one of claims 1-8 are implemented.

12. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the method for detecting container trucks at a port by weighing the container trucks as claimed in any one of claims 1 to 8 are implemented.

Images