Automatic receiving method and system for waste steel and iron

Abstract

The invention relates to the technical field of image recognition, and particularly provides an automatic scrap steel receiving method and system, and the method comprises the steps: collecting the image and weight information of a scrap steel carrying vehicle; identifying license plate information of the scrap steel carrier loader based on the image, and segmenting a carriage image from the image; identifying scrap steel individuals and impurity individuals from the carriage image by using a pre-training instance segmentation model deployed in a pluggable manner; generating a scrap steel grade based on a preset grade generation rule and the quantity of the scrap steel individuals and the impurity individuals; and recording the weight information, the license plate information and the scrap steel grade as a receiving event. According to the invention, accurate identification of the grade of the scrap steel, automatic receiving and data recording can be realized, and the automation degree of scrap steel recovery is improved.

Description

Technical Field

[0001] The invention belongs to the field of automation technology, and in particular relates to an automatic receiving method and system for scrap steel. Background Art

[0002] Scrap steel recycling is a complex and meticulous process, in which scrap steel grade assessment and unloading processing are key links. Due to the diversity of scrap steel types, one type may contain up to dozens of different scrap steel varieties, which differ in material, shape, size, etc., making scrap steel grade assessment particularly complicated.

[0003] Traditional image consistency matching methods are unable to cope with such diversity and often have difficulty accurately identifying and distinguishing different types of scrap steel, resulting in inaccurate grade assessment, which not only affects the recycling efficiency of scrap steel, but may also cause economic losses to recycling companies.

[0004] What is even more difficult is that the entire scrap steel recycling process currently still requires manual participation. From the receipt, classification, and grading of scrap steel to unloading and processing, each link requires manual operation, and the degree of automation is low. This not only increases labor costs, but also reduces recycling efficiency. At the same time, manual operation also poses certain safety risks. Summary of the invention

[0005] In view of the above-mentioned deficiencies in the prior art, the present invention provides a method and system for automatically receiving scrap steel to solve the above-mentioned technical problems.

[0006] In a first aspect, the present invention provides a method for automatically receiving scrap steel, comprising: Collect images and weight information of scrap steel carriers; Recognize the license plate information of the scrap steel transport vehicle based on the image, and segment the carriage image from the image; Using a pluggable pre-trained instance segmentation model to identify scrap steel individuals and impurity individuals from the carriage image; Generate scrap steel grades based on preset grade generation rules and the number of scrap steel units and impurity units; The weight information, license plate information and scrap steel grade are recorded as a receipt event.

[0007] In an optional embodiment, the method further comprises: Confirming that the grade of the scrap steel reaches the set grade threshold, and controlling the electromagnetic chuck to perform the unloading action; During the unloading process, the position and movement of the electromagnetic chuck are tracked based on the monitoring video using a motion recognition model, and a motion control instruction for the electromagnetic chuck is generated based on the position and movement.

[0008] In an optional implementation, the action recognition model includes: Feature extraction layer, used to extract contour features, texture features and feature point coordinates of each image; A classifier, used for judging whether it is an electromagnetic chuck according to the contour features and texture features; A feature fusion layer is used to integrate the feature point coordinates of the electromagnetic chuck of adjacent image frames into a data sequence according to the output of the classifier; A bidirectional long short-term memory network layer is used to identify the current action of the electromagnetic chuck based on the data sequence; The output layer is used to output the current action and the latest feature point coordinates recognized by the bidirectional long short-term memory network layer.

[0009] In an optional embodiment, collecting images and weight information of the scrap steel carrier includes: The camera and pressure sensor deployed on the gantry are used to collect images and weight information of the scrap steel carrier passing through the gantry.

[0010] In an optional embodiment, the scrap steel grade is generated based on a preset grade generation rule and the number of individual scrap steels and individual impurities, including: The ratio of the number of impurity individuals to the number of scrap steel individuals is used as an evaluation coefficient, and multiple scrap steel grades and the evaluation coefficient range corresponding to each scrap steel grade are set; The ratio of the number of identified impurity entities to the number of scrap steel entities is calculated, and the target scrap steel grade is determined according to the evaluation coefficient range corresponding to the ratio.

[0011] In a second aspect, the present invention provides an automatic receiving system for scrap steel, comprising: Data acquisition module, used to collect images and weight information of scrap steel carriers; A first processing module, configured to recognize the license plate information of the scrap steel transport vehicle based on the image, and segment a carriage image from the image; A second processing module is used to identify scrap steel individuals and impurity individuals from the carriage image using a pluggable pre-trained instance segmentation model; A third processing module is used to generate scrap steel grades based on a preset grade generation rule and the number of scrap steel units and impurity units; The data storage module is used to record the weight information, license plate information and scrap steel grade as a receiving event.

[0012] In an optional embodiment, the system further includes: A determination module, used to confirm that the grade of the scrap steel reaches a set grade threshold, and control the electromagnetic chuck to perform an unloading action; The unloading control module is used to track the position and movement of the electromagnetic chuck based on the monitoring video by using the motion recognition model during the unloading process, and generate a motion control instruction for the electromagnetic chuck based on the position and movement.

[0013] In an optional implementation, the action recognition model includes: Feature extraction layer, used to extract contour features, texture features and feature point coordinates of each image; A classifier, used for judging whether it is an electromagnetic chuck according to the contour features and texture features; A feature fusion layer is used to integrate the feature point coordinates of the electromagnetic chuck of adjacent image frames into a data sequence according to the output of the classifier; A bidirectional long short-term memory network layer is used to identify the current action of the electromagnetic chuck based on the data sequence; The output layer is used to output the current action and the latest feature point coordinates recognized by the bidirectional long short-term memory network layer.

[0014] In an optional embodiment, the data acquisition module includes: The data acquisition unit is used to collect images and weight information of the scrap steel carrier passing through the gantry using a camera and a pressure sensor deployed on the gantry.

[0015] In an optional embodiment, the third processing module includes: A setting unit is used to use the ratio of the number of impurity individuals to the number of scrap steel individuals as an evaluation coefficient, and to set multiple scrap steel grades and the evaluation coefficient range corresponding to each scrap steel grade; The calculation unit is used to calculate the ratio of the number of identified impurity entities to the number of scrap steel entities, and determine the target scrap steel grade according to the evaluation coefficient range corresponding to the ratio.

[0016] The beneficial effect of the present invention is that the automatic scrap steel receiving method and system provided by the present invention can realize accurate identification of scrap steel grades, as well as automatic receiving and data recording, thereby improving the degree of automation of scrap steel recycling.

[0017] In addition, the invention has a reliable design principle, a simple structure and a very broad application prospect. BRIEF DESCRIPTION OF THE DRAWINGS

[0018] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, for ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative work.

[0019] Figure 1is a schematic flow chart of a method according to an embodiment of the present invention.

[0020] Figure 2 is a schematic block diagram of a system according to an embodiment of the present invention. DETAILED DESCRIPTION

[0021] In order to enable those skilled in the art to better understand the technical solutions in the present invention, the technical solutions in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work should fall within the scope of protection of the present invention.

[0022] Unless otherwise defined, all technical and scientific terms used herein have the same meaning as those commonly understood by those skilled in the art of the present invention. The terms used in the specification of the present invention herein are only for the purpose of describing specific embodiments and are not intended to limit the present invention.

[0023] The automatic receiving method for scrap steel provided in the embodiment of the present invention is executed by a computer device, and accordingly, the automatic receiving system for scrap steel runs in the computer device.

[0024] Figure 1 is a schematic flow chart of a method according to an embodiment of the present invention. Figure 1 The execution subject may be an automatic receiving system for scrap steel. According to different requirements, the order of the steps in the flow chart may be changed, and some may be omitted.

[0025] like Figure 1 As shown, the method includes: S1. Collect images and weight information of scrap steel carriers.

[0026] Using high-definition cameras, weighbridges and other hardware equipment, the image information and weight information of the scrap steel carrier are collected simultaneously. The image information needs to clearly show the overall appearance of the vehicle and the loading of scrap steel in the vehicle, while the weight information accurately records the total weight of the carrier, providing basic data for the subsequent calculation of the net weight of the scrap steel.

[0027] S2. Recognize the license plate information of the scrap steel transport vehicle based on the image, and segment the carriage image from the image.

[0028] Using advanced image recognition technology, the license plate information of scrap steel carriers is automatically extracted from the collected images to ensure the accuracy of the information. At the same time, the image segmentation algorithm is used to accurately separate the carriage area from the complete image, providing more concentrated visual input for the subsequent identification of individual scrap steel.

[0029] S3. Using a pluggable pre-trained instance segmentation model, identify scrap steel individuals and impurity individuals from the carriage image.

[0030] A pluggable pre-trained instance segmentation model is introduced. After being trained on a large number of scrap steel samples, the model can efficiently identify and distinguish individual scrap steel and impurity individuals from carriage images, including but not limited to different types of scrap steel blocks, metal fragments, plastic wrappings, etc., providing key data support for subsequent grade assessment.

[0031] S4. Generate scrap steel grades based on preset grade generation rules and the number of scrap steel units and impurity units.

[0032] Based on the preset grade generation rules, the grade of the goods is automatically calculated by comprehensively considering the ratio of the number of individual scrap steel and impurity individuals, the type of scrap steel, the size and other factors. This process not only improves the accuracy of grade assessment, but also greatly shortens the assessment time and improves overall work efficiency.

[0033] S5. Record the weight information, license plate information and scrap steel grade as a receiving event.

[0034] The collected weight information, license plate information and intelligently generated scrap steel grade are integrated into a complete record of receipt events. This record not only reflects the specific situation of each scrap steel recycling in detail, but also provides valuable data support for subsequent inventory management, financial analysis, etc. At the same time, this comprehensive and accurate recording method also lays a solid foundation for the digital transformation of the scrap steel recycling industry.

[0035] In an embodiment of the present invention, based on step S1, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner.

[0036] The camera and pressure sensor deployed on the gantry are used to collect images and weight information of the scrap steel carrier passing through the gantry.

[0037] Specifically, the hardware device configuration includes: Install a gantry or column to support monitoring equipment and automatic weighing equipment.

[0038] Equipped with high-precision weighing sensor to measure the weight of scrap steel.

[0039] Install a license plate recognition system to identify truck information.

[0040] Deploy PLC controllers to automatically control the operation of the entire receiving system.

[0041] Equipped with sufficient computer hardware resources to run intelligent algorithms and models.

[0042] In an embodiment of the present invention, based on step S2, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner.

[0043] S201, License Plate Information Recognition Implementation Solution Image preprocessing: The collected scrap steel carrier images are preprocessed, including denoising, grayscale, binarization and other operations, to improve image quality and reduce the complexity of subsequent processing.

[0044] License plate location: Use deep learning-based target detection algorithms (such as YOLO, SSD, etc.) or traditional image processing algorithms (such as edge detection, morphological processing, etc.) to locate the license plate in the preprocessed image.

[0045] License plate character recognition: The located license plate area is segmented, and then the character sequence on the license plate is recognized using a character recognition algorithm (such as convolutional neural network CNN, recurrent neural network RNN, etc.).

[0046] Result verification: Verify the recognized license plate information, such as checking whether the characters conform to the license plate format specifications and whether there are obvious recognition errors.

[0047] S202, Carriage Image Segmentation Implementation Solution Image preprocessing (shared steps with license plate recognition): Similarly, pre-processing operations such as denoising and grayscale are performed on the collected images to provide high-quality input for subsequent image segmentation.

[0048] Rough positioning of the carriage area: Using shape- or color-based feature extraction methods, the approximate area of ​​the carriage can be initially located. This can be achieved by analyzing the color distribution, edge features, etc. in the image.

[0049] Fine segmentation: Based on the rough positioning, the car compartment area is finely segmented using image segmentation algorithms (such as GrabCut, U-Net, Mask R-CNN, etc.). These algorithms can accurately separate the car compartment area from the background based on pixel-level classification.

[0050] Post-processing: The segmented carriage image is post-processed, such as removing noise and filling holes, to obtain a more complete and clearer image of the carriage area.

[0051] In an embodiment of the present invention, based on step S3, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner.

[0052] S301, Preparation stage Dataset preparation: Collect a large number of scrap steel sample images, which should contain different types of scrap steel pieces, metal fragments, plastic wrappings and other impurity individuals.

[0053] These sample images are annotated to clearly distinguish the boundaries between scrap steel individuals and impurity individuals so that they can be used to train the instance segmentation model.

[0054] Model selection: Select a pre-trained instance segmentation model suitable for scrap steel and impurity individual identification. Commonly used models include MaskR-CNN, YOLACT, etc. These models perform well in object detection and instance segmentation tasks.

[0055] Choose an appropriate model architecture and configuration based on your hardware resources and computing requirements.

[0056] Environment configuration: Build a deep learning framework (such as TensorFlow, PyTorch, etc.) and the corresponding computing environment.

[0057] Install the required dependent libraries and tools to ensure that the model can run and train smoothly.

[0058] S302, model training stage Data preprocessing: The collected scrap steel sample images are preprocessed, including image scaling, normalization, data enhancement and other operations, to improve the generalization ability of the model.

[0059] Model training: The preprocessed data is fed into a pre-trained instance segmentation model for transfer learning and fine-tuning.

[0060] Set appropriate training parameters, such as learning rate, batch size, number of iterations, etc., to ensure that the model can fully learn the characteristics of scrap steel and impurities.

[0061] During the training process, the performance of the model is evaluated regularly, and the training parameters are adjusted and the model structure is optimized based on the evaluation results.

[0062] Model Validation: The trained model is validated using the validation data set to ensure that the model can accurately identify and distinguish individual scrap steel and impurity individuals.

[0063] Based on the verification results, the model is further optimized and adjusted to improve its recognition accuracy and robustness.

[0064] S303, Intelligent Identification Stage Model Reasoning: The preprocessed car image is input into the trained instance segmentation model for inference calculation.

[0065] The model will output the bounding box and category label of each scrap steel and impurity individual to achieve intelligent identification of scrap steel and impurity individuals.

[0066] Result processing: Post-process the recognition results output by the model, such as removing redundant bounding boxes and merging adjacent individuals of the same type.

[0067] Based on the identification results, the number, type and other information of individual scrap steel and impurity individuals are counted to provide key data support for subsequent grade assessment.

[0068] In an embodiment of the present invention, based on step S4, a possible embodiment is given below to illustrate its specific implementation scheme in a non-limiting manner.

[0069] S401, scrap steel grade setting and evaluation coefficient range First, we set four scrap steel grades: Grade A, Grade B, Grade C and Grade D. Each grade corresponds to an evaluation coefficient range, which is determined by the ratio of the number of impurity individuals to the number of scrap steel individuals (hereinafter referred to as "impurity ratio"). The specific settings are as follows:

[0070] S402, Impurity ratio calculation and target scrap grade determination Assume that in a scrap steel recycling process, the following number of scrap steel units and impurity units are identified through image recognition technology: Number of scrap steel units: 1000 Number of impurity entities: 100 Next, we calculate the impurity ratio: Impurity ratio = number of impurities / number of scrap steel = 100 / 1000 = 0.1 or 10% Then, we look for the corresponding evaluation coefficient range based on the calculated impurity ratio to determine the target scrap grade: The impurity ratio is 10%, which falls within the evaluation coefficient range corresponding to grade B (5% - 15%).

[0071] Therefore, the scrap steel grade of this batch of scrap steel is determined to be Grade B.

[0072] In an embodiment of the present invention, based on step S5, a possible embodiment is given below to illustrate its specific implementation in a non-limiting manner.

[0073] Create a delivery event record template that includes the necessary fields and information.

[0074] Fill the integrated data into the template to generate a complete receipt event record.

[0075] The records should include key information such as receipt time, location, vehicle information, scrap steel grade, weight, etc.

[0076] The generated receipt event records are stored in a database or cloud storage platform.

[0077] Ensure data security and accessibility.

[0078] Provides a data query interface to facilitate users to query specific delivery event records according to their needs.

[0079] Use data analysis tools to count and analyze receipt event records to provide data support for inventory management, financial analysis, etc.

[0080] On the basis of the above embodiment, in order to further improve the automation degree of scrap steel recycling, as an implementable method, after obtaining the scrap steel grade, the following steps are performed: 1. Scrap grade confirmation and unloading action triggering Scrap steel grade determination: After the scrap steel is delivered to the designated location, the scrap steel grade is first confirmed. This can be done through the image recognition technology and intelligent grade generation system mentioned earlier.

[0081] The system will automatically determine the grade of scrap steel based on the ratio of the number of identified scrap steel individuals to impurity individuals.

[0082] Level threshold setting: One or more scrap steel grade thresholds are pre-set based on the company's needs and market standards.

[0083] These thresholds will be used to decide when to trigger the unloading action of the electromagnetic chuck.

[0084] Unloading action trigger: When the scrap steel grade reaches or exceeds the set grade threshold, the system triggers the unloading action of the electromagnetic chuck.

[0085] This typically involves sending a signal or command to the electromagnetic chuck's control system to initiate its unloading function.

[0086] 2. Tracking and control of electromagnetic chuck during unloading Surveillance video acquisition: Install high-definition cameras in the unloading area to capture real-time video of the electromagnetic suction cup during the unloading process.

[0087] Ensure the clarity and coverage of surveillance video to accurately track the position and movement of the electromagnetic suction cup.

[0088] Action recognition model application: Use the pre-trained action recognition model to analyze surveillance videos.

[0089] The model can identify the position and actions of the electromagnetic chuck in the video, such as moving, lifting, and unloading.

[0090] Position and motion tracking: Through the motion recognition model, the position and movement of the electromagnetic suction cup during the unloading process are tracked in real time.

[0091] This includes the movement trajectory of the electromagnetic chuck, its current position, and the specific type of action performed.

[0092] Action control command generation: Based on the tracked position and movement of the electromagnetic chuck, the system generates corresponding motion control instructions.

[0093] These instructions may include adjusting the moving speed of the electromagnetic chuck, stopping movement, changing the unloading direction, etc.

[0094] The generation of instructions should be based on real-time video data and motion recognition results to ensure that the electromagnetic suction cup can complete the unloading task in the expected manner.

[0095] Instruction sending and execution: The generated motion control command is sent to the control system of the electromagnetic chuck.

[0096] After receiving the command, the control system controls the electromagnetic chuck to perform corresponding actions according to the command requirements.

[0097] 3. Safety and efficiency considerations Security Monitoring: During the unloading process, the position and movement of the electromagnetic chuck, as well as the safety of the surrounding environment, are continuously monitored.

[0098] If any potential safety risks or abnormal situations are detected, the system should immediately sound an alarm and stop the unloading action.

[0099] Efficiency optimization: Through precise motion recognition and position tracking, the moving path and unloading strategy of the electromagnetic chuck can be optimized.

[0100] This helps reduce unloading time, improve unloading efficiency, and reduce energy consumption.

[0101] Data recording and analysis: During the unloading process, the position, movement and corresponding control instructions of the electromagnetic chuck are recorded.

[0102] The data is subsequently analyzed and evaluated to improve the unloading process and enhance system performance.

[0103] Among them, the action recognition model includes: The feature extraction layer is used to extract the contour features, texture features and feature point coordinates of each image; in order to capture the contour features and texture features of the image, the feature extraction layer uses various feature descriptors. For example, for contour features, algorithms such as Hough transform or Canny edge detection can be used to extract the boundaries of the object. For texture features, methods such as local binary patterns (LBP), gray-level co-occurrence matrix (GLCM) or Gabor filters can be used to describe the details and texture of the object surface. These feature descriptors can quantify the key information in the image and provide strong support for subsequent classification and recognition. Another important task in the feature extraction layer is to detect and extract feature points in the image. These feature points are stable and unique local structures in the image, such as corners, edge points or spots. Common feature point detection algorithms include SIFT (Scale Invariant Feature Transform), SURF (Speeded Robust Features) and ORB (Oriented Fast and Rotationally Short). By detecting these feature points and calculating their coordinates and descriptors, the feature extraction layer can provide accurate location information for subsequent tracking and recognition. As a typical deep learning model, convolutional neural network (CNN) has been widely used in feature extraction tasks. CNN can automatically learn hierarchical feature representations in images through multi-layer convolution and pooling operations. In the feature extraction layer, a pre-trained CNN model can be used to extract high-level features of the image, which have higher discriminative power for subsequent recognition and classification tasks.

[0104] The classifier is used to determine whether it is an electromagnetic chuck based on the contour features and texture features; the classifier is a decision-making unit in the model, which determines whether the current image contains an electromagnetic chuck based on the contour features and texture features provided by the feature extraction layer. The classifier learns the unique visual feature pattern of the electromagnetic chuck by training a large amount of sample data. When receiving a new image, the classifier will match these features with the learned pattern to make a judgment on whether the electromagnetic chuck is included. This step is the basis for subsequent action recognition. Only when the electromagnetic chuck is correctly identified can its behavior be further analyzed.

[0105] The feature fusion layer is used to integrate the feature point coordinates of the electromagnetic chuck in adjacent image frames into a data sequence according to the output of the classifier; the feature fusion layer is the key link connecting adjacent image frames. It integrates the feature point coordinates of the electromagnetic chuck in adjacent image frames into a continuous data sequence according to the output of the classifier. This data sequence not only contains the position change of the electromagnetic chuck in the time series, but also implies dynamic information such as the direction and speed of its movement. By integrating this information, the feature fusion layer provides rich input data for the subsequent bidirectional long short-term memory network layer.

[0106] Among them, the feature fusion layer includes: Adjacent image frame matching: Using feature matching algorithms, feature points in adjacent image frames are matched. This can be achieved by calculating metrics such as Euclidean distance and Hamming distance between feature points. The matching results will be used to determine the position change of the electromagnetic chuck in different image frames.

[0107] Integration of feature point coordinates: According to the output of the classifier (such as support vector machine, neural network, etc.), the feature points belonging to the electromagnetic chuck are screened out. The coordinates of these feature points are integrated according to the time series to form a continuous data sequence. This data sequence will contain the position information of the electromagnetic chuck at different time points.

[0108] Dynamic information extraction: By analyzing the changes in the position of the electromagnetic chuck in the data sequence, dynamic information such as its direction and speed of movement can be extracted. This can be achieved by calculating the difference between adjacent positions and using methods such as Kalman filters for smoothing.

[0109] Feature fusion: The extracted dynamic information such as position information, direction information and speed information are fused to form a feature vector containing rich information.

[0110] The bidirectional long short-term memory network layer is used to identify the current action of the electromagnetic chuck based on the data sequence; the bidirectional long short-term memory network layer (Bi-LSTM) is the core component of the model, which is responsible for identifying the current action of the electromagnetic chuck based on the data sequence provided by the feature fusion layer. Bi-LSTM, with its unique bidirectional structure, can simultaneously consider past and future information, thereby more accurately capturing the movement trajectory and action pattern of the electromagnetic chuck. Through training and learning, this layer can establish a mapping relationship between data sequences and action categories, and realize accurate recognition of the electromagnetic chuck action.

[0111] The output layer is used to output the current action and the latest feature point coordinates recognized by the bidirectional long short-term memory network layer. The output layer can be a fully connected layer that maps the output of the Bi-LSTM to the required action and coordinate representation.

[0112] For a specific example: 1. Equipment and devices (1) License plate recognition technology, automatic weighing technology and PLC control technology are used to achieve automatic measurement of trucks, automatic identification of receiving points, and automatic calculation of net weight and miscellaneous weight.

[0113] (2) At the entrance area of ​​the receiving point, install a light gantry or column with an independent foundation and a height of more than 10 meters, and install a surveillance camera on the gantry or column.

[0114] 2. Algorithms and Systems (1) Unmanned weighing module: It uses high-precision weighing sensors, license plate recognition systems, PLC controllers, and intelligent access control to achieve unmanned automatic weighing of scrap steel materials.

[0115] (2) Multi-category scrap steel sample labeling and training module: Utilize machine learning technology and adopt any visual classification / segmentation algorithm model to classify, label and train scrap steel samples and establish a scrap steel category recognition model.

[0116] (3) Multi-class impurity sample labeling and training module: Label and train impurity samples, use any visual classification / segmentation algorithm model, and establish an impurity recognition model to improve the accuracy of impurity detection.

[0117] (4) Multi-type truck compartment labeling and training module: Through image recognition technology, any visual classification / segmentation algorithm model is used to label and train truck compartments to achieve automatic identification of compartment types.

[0118] (5) Multiple electromagnetic chuck labeling and training module: Label and train the electromagnetic chuck, use any visual classification / segmentation algorithm model, establish a chuck control model, and realize automatic positioning and tracking of the chuck.

[0119] (6) Fully automatic receiving process identification and control module: By establishing a decision tree for the receiving quality inspection process, the module can identify and control the vehicle entry, electromagnetic suction cup unloading, and completion of unloading, thereby realizing automatic identification and control of the receiving process and improving operational efficiency.

[0120] (7) Fully automatic carriage positioning and segmentation module: Utilizes image processing technology and adopts ONNX technology to deploy pluggable algorithm models to locate and segment the carriages, providing accurate target areas for subsequent category identification and impurity detection.

[0121] (8) Fully automatic suction cup tracking and photography module: Through visual tracking technology and the use of ONNX technology, the pluggable deployment algorithm model can achieve real-time tracking and photography of the suction cup to ensure the accuracy of material measurement.

[0122] (9) Fully automatic scrap steel category identification and calculation module: Based on the established scrap steel category identification model, the ONNX technology is used to deploy the algorithm model in a pluggable manner to automatically identify and classify scrap steel samples.

[0123] (10) Fully automatic impurity identification and calculation module: Based on the established impurity identification model, the ONNX technology is used to deploy the algorithm model in a pluggable manner to automatically identify and calculate impurities.

[0124] (11) Voice prompt module: Based on text2speech technology, it can intelligently identify the current operation stage and provide voice broadcast prompts to business personnel and truck drivers in scenarios such as metering, receiving, grading and abnormal business.

[0125] 3. System Integration The data from unmanned measurement, intelligent quality inspection of incoming goods, MES, ERP and other systems are integrated using API to achieve unmanned operation of the entire process of automatic identification of business measurement, automatic identification of business quality inspection, and automatic calculation of settlement vouchers.

[0126] 4. Model Optimization Through quantization, distillation, pruning and other technologies, the resource usage of the ONNX model is minimized, the performance is greatly improved, a single receiving point only occupies 3GB of GPU memory, and the model deployment and operation costs are greatly reduced.

[0127] Advantages of the present invention include: 1. Improve the accuracy of quality determination: Through the application of modeling at the category level and instance segmentation technology, the present invention can more accurately determine the quality of scrap steel.

[0128] 2. Optimization of the entire process: The present invention realizes the unmanned and efficient operation of the entire process of scrap steel receiving, organically combines a variety of automation and intelligent technologies (see attached figure), and improves the overall operation efficiency.

[0129] 3. Real-time and accuracy of data processing: Through the integration of business and metering data, the present invention can process business data in real time and accurately, thereby improving decision-making efficiency.

[0130] 4. Clear and stable images: The gantry or column design with independent foundation ensures the clarity and stability of the monitoring image, which helps to improve the reliability and accuracy of the system.

[0131] 5. Reduced technology costs: The adoption of the ONNX intelligent model format reduces the embedding cost of intelligent model technology, making the system more economical and efficient.

[0132] 6. Advantages of comprehensive modeling: Comprehensive modeling of scrap steel and related elements improves the adaptability and accuracy of the system and can better handle various complex scrap steel processing scenarios.

[0133] 7. Improved user experience: The voice prompt function improves the user experience of business personnel and truck drivers, and reduces operational misunderstandings and communication costs.

[0134] 8. Model optimization effect: Model optimization technology significantly reduces resource usage and operating costs, making the system more efficient and economical.

[0135] In some embodiments, the automatic steel scrap receiving system may include a plurality of functional modules composed of computer program segments. The computer programs of the various program segments in the automatic steel scrap receiving system may be stored in a memory of a computer device and executed by at least one processor to perform (see Figure 1 Description) Automatic receipt of scrap steel.

[0136] In this embodiment, the automatic scrap steel receiving system can be divided into multiple functional modules according to the functions it performs, such as Figure 2 As shown. The functional modules of the system 200 may include: a data acquisition module 210, a first processing module 220, a second processing module 230, a third processing module 240 and a data storage module 250. The module referred to in the present invention refers to a series of computer program segments that can be executed by at least one processor and can complete fixed functions, which are stored in a memory. In this embodiment, the functions of each module will be described in detail in subsequent embodiments.

[0137] Data acquisition module, used to collect images and weight information of scrap steel carriers; A first processing module, configured to recognize the license plate information of the scrap steel transport vehicle based on the image, and segment a carriage image from the image; A second processing module is used to identify scrap steel individuals and impurity individuals from the carriage image using a pluggable pre-trained instance segmentation model; A third processing module is used to generate scrap steel grades based on a preset grade generation rule and the number of scrap steel units and impurity units; The data storage module is used to record the weight information, license plate information and scrap steel grade as a receiving event.

[0138] Optionally, as an embodiment of the present invention, the system further includes: A determination module, used to confirm that the grade of the scrap steel reaches a set grade threshold, and control the electromagnetic chuck to perform an unloading action; The unloading control module is used to track the position and movement of the electromagnetic chuck based on the monitoring video by using the motion recognition model during the unloading process, and generate a motion control instruction for the electromagnetic chuck based on the position and movement.

[0139] Optionally, as an embodiment of the present invention, the action recognition model includes: Feature extraction layer, used to extract contour features, texture features and feature point coordinates of each image; A classifier, used for judging whether it is an electromagnetic chuck according to the contour features and texture features; A feature fusion layer is used to integrate the feature point coordinates of the electromagnetic chuck of adjacent image frames into a data sequence according to the output of the classifier; A bidirectional long short-term memory network layer is used to identify the current action of the electromagnetic chuck based on the data sequence; The output layer is used to output the current action and the latest feature point coordinates recognized by the bidirectional long short-term memory network layer.

[0140] Optionally, as an embodiment of the present invention, the data acquisition module includes: The data acquisition unit is used to collect images and weight information of the scrap steel carrier passing through the gantry using a camera and a pressure sensor deployed on the gantry.

[0141] Optionally, as an embodiment of the present invention, the third processing module includes: A setting unit is used to use the ratio of the number of impurity individuals to the number of scrap steel individuals as an evaluation coefficient, and to set multiple scrap steel grades and the evaluation coefficient range corresponding to each scrap steel grade; The calculation unit is used to calculate the ratio of the number of identified impurity entities to the number of scrap steel entities, and determine the target scrap steel grade according to the evaluation coefficient range corresponding to the ratio.

[0142] In this specification, the same or similar parts between the various embodiments can be referred to each other. In particular, for the device embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the description in the method embodiment.

[0143] In the several embodiments provided by the present invention, it should be understood that the disclosed systems and methods can be implemented in other ways. For example, the system embodiments described above are only schematic. For example, the division of the modules is only a logical function division. There may be other division methods in actual implementation, such as multiple modules or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of systems or modules, which can be electrical, mechanical or other forms.

[0144] The modules described as separate components may or may not be physically separated, and the components shown as modules may or may not be physical modules, that is, they may be located in one place or distributed on multiple network modules. Some or all of the modules may be selected according to actual needs to achieve the purpose of the solution of this embodiment.

[0145] In addition, each functional module in each embodiment of the present invention may be integrated into one processing module, or each module may exist physically separately, or two or more modules may be integrated into one module.

[0146] Although the present invention has been described in detail with reference to the accompanying drawings and in combination with the preferred embodiments, the present invention is not limited thereto. Without departing from the spirit and essence of the present invention, a person of ordinary skill in the art may make various equivalent modifications or substitutions to the embodiments of the present invention, and these modifications or substitutions shall be within the scope of the present invention. Any person of ordinary skill in the art may easily think of changes or substitutions within the technical scope disclosed by the present invention, and these shall be within the scope of protection of the present invention.

Claims

1. A method for automatically receiving scrap steel, characterized in that: include: Collect images and weight information of scrap steel carriers; Recognize the license plate information of the scrap steel transport vehicle based on the image, and segment the carriage image from the image; Using a pluggable pre-trained instance segmentation model to identify scrap steel individuals and impurity individuals from the carriage image; Generate scrap steel grades based on preset grade generation rules and the number of scrap steel units and impurity units; The weight information, license plate information and scrap steel grade are recorded as a receipt event.

2. The method according to claim 1, characterized in that The method further comprises: Confirming that the grade of the scrap steel reaches the set grade threshold, and controlling the electromagnetic chuck to perform the unloading action; During the unloading process, the position and movement of the electromagnetic chuck are tracked based on the monitoring video using a motion recognition model, and a motion control instruction for the electromagnetic chuck is generated based on the position and movement.

3. The method according to claim 2, characterized in that The action recognition model includes: Feature extraction layer, used to extract contour features, texture features and feature point coordinates of each image; A classifier, used for judging whether it is an electromagnetic chuck according to the contour features and texture features; A feature fusion layer is used to integrate the feature point coordinates of the electromagnetic chuck of adjacent image frames into a data sequence according to the output of the classifier; A bidirectional long short-term memory network layer is used to identify the current action of the electromagnetic chuck based on the data sequence; The output layer is used to output the current action and the latest feature point coordinates recognized by the bidirectional long short-term memory network layer.

4. The method according to claim 1, characterized in that: Collect images and weight information of scrap trucks, including: The camera and pressure sensor deployed on the gantry are used to collect images and weight information of the scrap steel carrier passing through the gantry.

5. The method according to claim 1, characterized in that: Generate scrap grades based on preset grade generation rules and the number of scrap steel units and impurity units, including: The ratio of the number of impurity individuals to the number of scrap steel individuals is used as an evaluation coefficient, and multiple scrap steel grades and the evaluation coefficient range corresponding to each scrap steel grade are set; The ratio of the number of identified impurity entities to the number of scrap steel entities is calculated, and the target scrap steel grade is determined according to the evaluation coefficient range corresponding to the ratio.

6. An automatic receiving system for scrap steel, characterized in that: include: Data acquisition module, used to collect images and weight information of scrap steel carriers; A first processing module, configured to recognize the license plate information of the scrap steel transport vehicle based on the image, and segment a carriage image from the image; A second processing module is used to identify scrap steel individuals and impurity individuals from the carriage image using a pluggable pre-trained instance segmentation model; A third processing module is used to generate scrap steel grades based on a preset grade generation rule and the number of scrap steel units and impurity units; The data storage module is used to record the weight information, license plate information and scrap steel grade as a receiving event.

7. The system according to claim 6, characterized in that The system further comprises: A determination module, used to confirm that the grade of the scrap steel reaches a set grade threshold, and control the electromagnetic chuck to perform an unloading action; The unloading control module is used to track the position and movement of the electromagnetic chuck based on the monitoring video by using the motion recognition model during the unloading process, and generate a motion control instruction for the electromagnetic chuck based on the position and movement.

8. The system according to claim 7, characterized in that The action recognition model includes: Feature extraction layer, used to extract contour features, texture features and feature point coordinates of each image; A classifier, used to determine whether it is an electromagnetic chuck according to the contour features and texture features; A feature fusion layer is used to integrate the feature point coordinates of the electromagnetic chuck of adjacent image frames into a data sequence according to the output of the classifier; A bidirectional long short-term memory network layer is used to identify the current action of the electromagnetic chuck based on the data sequence; The output layer is used to output the current action and the latest feature point coordinates recognized by the bidirectional long short-term memory network layer.

9. The system according to claim 6, characterized in that The data acquisition module comprises: The data acquisition unit is used to collect images and weight information of the scrap steel carrier passing through the gantry using a camera and a pressure sensor deployed on the gantry.

10. The system according to claim 6, characterized in that The third processing module comprises: A setting unit is used to use the ratio of the number of impurity individuals to the number of scrap steel individuals as an evaluation coefficient, and to set multiple scrap steel grades and the evaluation coefficient range corresponding to each scrap steel grade; The calculation unit is used to calculate the ratio of the number of identified impurity entities to the number of scrap steel entities, and determine the target scrap steel grade according to the evaluation coefficient range corresponding to the ratio.

Images