Three-dimensional warehouse checking method, system and equipment combining 5G transmission and machine vision

Abstract

The invention relates to the field of three-dimensional warehouse checking, in particular to a 5G transmission and machine vision combined three-dimensional warehouse checking method, system and equipment, which can be understood that according to the technical scheme provided by the invention, a high-definition industrial camera is deployed on a stacker, an advanced vision algorithm is combined, and comprehensive image acquisition and analysis are performed on goods in a warehouse; according to the method, the detailed state of a goods stack can be captured, goods distribution and counting can be automatically recognized through an algorithm, the method adapts to various stacking rules, the real-time performance and accuracy of the inventory result are ensured, various challenges in stereoscopic warehouse inventory can be effectively coped with through intelligent data processing and analysis, and the inventory efficiency is improved. And a new economic, efficient and safe automatic inventory management way is provided for enterprises.

Description

Technical Field

[0001] The present application relates to the field of vertical warehouse inventory, and in particular to a vertical warehouse inventory method, system, and equipment combining 5G transmission with machine vision. Background Art

[0002] The development of the three-dimensional warehouse field continues to benefit from the rapid growth of globalization and e-commerce, which has led to a growing demand for efficient and automated warehousing solutions. The application of automated three-dimensional warehouses can effectively reduce land occupation and labor costs, and is a key factor in improving warehousing and logistics efficiency. The development of automated three-dimensional warehouses can effectively solve the situation of large-scale occupation of land and manpower in the warehousing industry, and realize the automation and intelligence of warehousing, reduce warehousing operation and management costs, and improve logistics efficiency.

[0003] Inventory in a three-dimensional warehouse refers to the process of comprehensively counting and checking the material inventory in the three-dimensional warehouse. Usually, information such as the current quantity of materials is recorded and compared with the actual inventory to ensure the accuracy and completeness of the inventory.

[0004] The existing forms of inventory counting in stereoscopic warehouses mainly include manual inventory counting and RFID inventory counting. Among them, manual inventory counting is a relatively common means of inventory counting, which is to count the number of inventory items one by one manually and record them in a spreadsheet. In order to ensure the accuracy of the inventory counting and the safety of the manual inventory counting process, it is necessary to temporarily stop the receipt and delivery of materials during the inventory counting. This method is time-consuming and labor-intensive, which seriously affects the operational efficiency of the stereoscopic warehouse. There are great safety hazards when personnel count in the warehouse. Due to manual participation in the inventory counting, it is easy to have wrong counting and missing counting, which makes it difficult to match the actual objects with the book; secondly, RFID inventory counting technology installs electronic tags on the inventory goods and counts by scanning the tags. This method is efficient and fast. However, in large-scale stereoscopic warehouse scenarios, the deployment cost of electronic tags remains high, which to a certain extent reduces the market's enthusiasm for using RFID technology. And the current methods are mostly used in material box scenarios.

[0005] That is, the existing manual inventory counting technology is not only inefficient and costly, but also accompanied by personnel safety risks, especially in the three-dimensional warehouse (vertical warehouse) environment, these disadvantages are more prominent. On the other hand, although RFID technology provides an automated solution, its high initial investment, signal stability issues that are easily disturbed by complex industrial environments, and the limitation of only providing tag information rather than intuitive inventory status have greatly limited its application efficiency and accuracy. Therefore, the field of vertical warehouse inventory counting urgently needs an innovative technology that is more cost-effective, safer and more accurate. Summary of the invention

[0006] In view of this, the purpose of the present invention is to provide a vertical warehouse inventory method, system, and equipment that combines 5G transmission with machine vision, so as to solve the problems in the prior art that manual inventory is not only inefficient and costly, but also has signal stability problems that RFID technology is easily disturbed by complex industrial environments, and can only provide tag information rather than intuitive inventory status.

[0007] According to a first aspect of an embodiment of the present invention, a method for inventory counting in a vertical warehouse combining 5G transmission and machine vision is provided, including:

[0008] Receive the issued inventory count requirement task instructions;

[0009] Based on the issued inventory counting demand task instruction, a preset industrial camera is used to capture image information of the current status of the storage location;

[0010] The storage location status image information is transmitted in real time to a preset server end by using a preset 5G terminal. The server end receives the storage location status image information, and uses a preset target detection algorithm to perform in-depth analysis on the storage location status image information to obtain accurate target detection results of the front end face, upper end face and front end face of the cargo stack;

[0011] Utilizing the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet, the quantity of goods in the cargo stack to be counted is calculated by a preset counting algorithm;

[0012] The quantity of goods in the stack to be calculated is fed back in real time to the preset system for storing the disk warehouse data. After receiving the quantity of goods in the stack to be calculated, the preset system for storing the disk warehouse data compares the quantity of goods in the stack to be calculated with the data of the corresponding stack stored in the system. If there is a difference, the corresponding storage location is marked, and the automatic correction process of the inventory record is triggered through manual review or directly.

[0013] Furthermore, the preset target detection algorithm is used to perform in-depth analysis on the image information of the storage location status to obtain accurate target detection results of the front end face, the upper end face and the front end face of the pallet, including:

[0014] Marking the boundary box of the front end face of the cargo stack, the upper end face of the cargo stack, and the front end face of the pallet of the storage location status image information to obtain the first image information;

[0015] The first image information is input into the trained deep target detection model to output the results of accurate target detection of the front end face, upper end face and front end face of the cargo stack.

[0016] Furthermore, the deep object detection model includes:

[0017] YOLOv8 object detection model.

[0018] Furthermore, the method of using the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet to calculate the quantity of goods in the cargo stack to be calculated by a preset counting algorithm includes:

[0019] Filter the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet using a preset confidence threshold, retain the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet that are higher than the preset confidence, and obtain the second image information;

[0020] Performing an image boundary overflow check on the second image information, and if the second image information is not completely within the preset image range, adjusting the position or viewing angle of the industrial camera so that the second image information is completely within the preset image range, thereby obtaining third image information;

[0021] Performing effective quantity preliminary verification on the front end face, upper end face and front end face of the pallet in the third image information to obtain different cargo position status information; wherein the different cargo position status information includes: full stack, half stack, empty stack, empty cargo position, and only pallet storage information;

[0022] Perform secondary identification on the detection frames of the front end face, the upper end face and the front end face of the pallet of the third image information, remove redundant information of non-local stack positions, and obtain fourth image information; wherein the fourth image information carries status information of different cargo positions after removing redundant information of non-local stack positions;

[0023] Dividing the fourth image information into layers, and identifying the number of valid layers in the fourth image information and the position and range information of each layer;

[0024] Using the result of the layer division, determining the full layer state of the fourth image information to obtain full layer number information;

[0025] The quantity of goods in the stack to be calculated is calculated using the full layer number information in the fourth image information, the preset standard quantity of goods in each layer, and the different cargo position status information after removing the redundant information of non-primary stack positions in the fourth image information.

[0026] Furthermore, the dividing the fourth image information into layers, and identifying the number of valid layers in the fourth image information and the position and range information of each layer, includes:

[0027] Calculating the vertical distance between each frame in the fourth image information, and distinguishing different cargo levels in combination with a preset same-layer height threshold;

[0028] By utilizing the different cargo levels, the detection frames belonging to the same level are sorted from large to small according to the vertical distance, and the number of valid layers in the fourth image information and the position and range information of each layer are identified.

[0029] Furthermore, the step of determining the full layer status of the fourth image information by using the result of the layer division to obtain the full layer number information includes:

[0030] Using the result of hierarchical division, the length of the front end surface coverage of each layer is obtained;

[0031] Using the detection information of the front face of the pallet and the length covered by the front face of each layer, the actual length of each layer and the standard length of the pallet are obtained;

[0032] If the difference between the actual length of each layer and the standard length of the pallet is within the preset full layer determination threshold range, the layer is determined to be a full layer, and the full layer number information is obtained.

[0033] According to a second aspect of an embodiment of the present invention, a vertical warehouse inventory system combining 5G transmission and machine vision is provided, including:

[0034] The acquisition module is used to obtain the issued inventory counting requirement task instructions;

[0035] A first processing module is used to capture the image information of the current status of the storage location using a preset industrial camera based on the issued inventory counting demand task instruction;

[0036] The second processing module is used to transmit the storage location status image information to a preset server in real time based on the use of a preset 5G terminal. The server receives the storage location status image information and uses a preset target detection algorithm to perform in-depth analysis on the storage location status image information to obtain the result of accurate target detection of the front end face, upper end face and front end face of the cargo stack;

[0037] A third processing module is used to calculate the quantity of goods in the cargo stack to be calculated by a preset counting algorithm based on the result of the precise target detection of the front end face, the upper end face and the front end face of the pallet;

[0038] The fourth processing module is used to feed back the quantity of goods in the to-be-calculated stack to the preset system for storing the disk warehouse data in real time. After receiving the quantity of goods in the to-be-calculated stack, the preset system for storing the disk warehouse data compares the quantity of goods in the to-be-calculated stack with the data of the corresponding stack stored in the system. If there is a difference, the corresponding storage location is marked, and the automatic correction process of the inventory record is triggered manually or directly.

[0039] According to a third aspect of an embodiment of the present invention, a vertical warehouse inventory counting device combining 5G transmission and machine vision is provided, including:

[0040] A memory having an executable program stored therein;

[0041] A processor is used to execute the executable program in the memory to implement the steps of any one of the above methods.

[0042] The technical solution provided by the embodiments of the present invention may have the following beneficial effects:

[0043] It can be understood that the technical solution provided by the present invention obtains the issued inventory counting demand task instruction; based on the issued inventory counting demand task instruction, uses a preset industrial camera to capture the current image information of the storage location; then, uses a preset 5G terminal to transmit the current image information of the storage location in real time to a preset server end, the server end receives the current image information of the storage location, and uses a preset target detection algorithm to perform in-depth analysis on the current image information of the storage location to obtain the result of accurate target detection of the front end face, the upper end face and the front end face of the pallet; uses the result of accurate target detection of the front end face, the upper end face and the front end face of the pallet to calculate the number of goods in the stack to be calculated through a preset counting algorithm; finally, the number of goods in the stack to be calculated is fed back to the preset system for storing pallet warehouse data in real time, and after receiving the number of goods in the stack to be calculated, the preset system for storing pallet warehouse data compares the number of goods in the stack to be calculated with the data of the corresponding stack stored in the system, and if there is a difference, the corresponding storage location is marked, and the automatic correction process of the inventory record is triggered manually or directly. It can be understood that the technical solution provided by the present invention, by deploying high-definition industrial cameras on the stacker and combining with advanced visual algorithms, can comprehensively collect and analyze images of the goods in the warehouse, which can not only capture the detailed status of the cargo stack, but also automatically identify the distribution and counting of goods through algorithms, and adapt to various stacking rules to ensure the real-time and accuracy of the inventory results. Through intelligent data processing and analysis, the present invention can effectively cope with various challenges in the vertical warehouse inventory and provide enterprises with an economical, efficient and safe new way of automated inventory management; further, by integrating machine vision technology, it can replace traditional manual inventory, greatly improve the inventory speed, reduce human intervention, and reduce long-term operating costs; adopt a non-invasive visual inventory method to avoid direct entry of personnel into dangerous vertical warehouse areas, significantly improving the safety of operations; use high-definition cameras with advanced image recognition algorithms to directly observe and analyze the actual inventory, overcome the defect that RFID technology cannot intuitively reflect the actual inventory status, and maintain high-accuracy inventory results even in complex environments.

[0044] It is to be understood that the foregoing general description and the following detailed description are exemplary and explanatory only and are not restrictive of the invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings, which are incorporated in and constitute a part of this specification, illustrate embodiments consistent with the invention and, together with the description, serve to explain the principles of the invention.

[0046] Figure 1 It is a schematic diagram of the steps of a vertical warehouse inventory method combining 5G transmission and machine vision according to an exemplary embodiment;

[0047] Figure 2 It is a schematic diagram of a structure flow of a vertical warehouse inventory counting method combining 5G transmission and machine vision according to an exemplary embodiment;

[0048] Figure 3 It is a schematic diagram of the composition of a vertical warehouse inventory counting system combining 5G transmission and machine vision according to an exemplary embodiment;

[0049] Figure 4 It is a schematic diagram of the composition of a vertical warehouse inventory counting device combining 5G transmission and machine vision according to an exemplary embodiment. DETAILED DESCRIPTION

[0050] Exemplary embodiments will be described in detail herein, examples of which are shown in the accompanying drawings. When the following description refers to the drawings, the same numbers in different drawings represent the same or similar elements unless otherwise indicated. The embodiments described in the following exemplary embodiments do not represent all embodiments consistent with the present invention. Instead, they are merely examples of devices and methods consistent with some aspects of the present invention as detailed in the appended claims.

[0051] Embodiment 1

[0052] Figure 1 The figure is a schematic diagram of the steps of a method for counting inventory in a vertical warehouse by combining 5G transmission with machine vision according to an exemplary embodiment, the method comprising:

[0053] S1. Obtain the inventory counting requirement task instruction issued;

[0054] S2. Based on the inventory counting demand task instructions issued, use a preset industrial camera to capture the image information of the storage location status;

[0055] S3. Using a preset 5G terminal to transmit the storage location status image information in real time to a preset server, the server receives the storage location status image information, uses a preset target detection algorithm to perform in-depth analysis on the storage location status image information, and obtains the results of accurate target detection of the front end face, upper end face and front end face of the cargo stack;

[0056] S4. Using the results of precise target detection of the front end surface, upper end surface and front end surface of the cargo stack, the number of cargo to be calculated is calculated by a preset counting algorithm;

[0057] S5. Feedback the quantity of goods in the to-be-calculated stack to the preset system for storing the disk warehouse data in real time. After receiving the quantity of goods in the to-be-calculated stack, the preset system for storing the disk warehouse data compares the quantity of goods in the to-be-calculated stack with the data of the corresponding stack stored in the system. If there is a difference, mark the corresponding storage location, and trigger the automatic correction process of the inventory record through manual review or direct triggering.

[0058] In one embodiment, see Figure 2 As described in steps S1-S2, the step of obtaining the issued inventory counting requirement task instruction and capturing the image information of the storage location status using a preset industrial camera based on the issued inventory counting requirement task instruction includes:

[0059] Integration of advanced hardware equipment: High-performance industrial cameras and 5G client terminal devices are integrated on the stacker device. Among them, the industrial camera plays a key role in accurately capturing high-definition images of goods in the warehouse. The deployment of 5G terminals fully utilizes the high speed and low latency advantages of the fifth-generation mobile communication technology to ensure that image data can be transmitted to the remote server instantly and efficiently, laying the foundation for real-time monitoring and analysis.

[0060] Intelligent scheduling: After receiving the task instructions automatically generated by the WMS (automated warehouse management system) according to the inventory counting requirements, the instructions are accurately interpreted by the WCS (warehouse control system), and the PLC (programmable logic controller) accurately drives the stacker to guide it to the designated storage location with the industrial camera. When arriving at the target location, the camera immediately starts the shooting task to capture the current image of the storage location.

[0061] In one embodiment, as described in step S3, the preset 5G terminal is used to transmit the storage location status image information to a preset server in real time, and the server receives the storage location status image information, and uses a preset target detection algorithm to perform in-depth analysis on the storage location status image information to obtain the results of accurate target detection of the front end face, upper end face and front end face of the cargo stack, including:

[0062] S31. Real-time image data transmission: With the help of the 5G terminal on the stacker, the high-resolution image data just collected can be uploaded to the cloud server almost without delay. This ultra-high-speed data transmission capability provides instant data support for subsequent image processing and analysis.

[0063] S32. Deep image analysis and inventory counting: After receiving the image, the server uses advanced target detection algorithms to perform in-depth analysis on the image, focusing on identifying the front end, upper end and front end of the pallet. Through a sophisticated counting algorithm, the system can accurately count the number of goods in the stack, thereby comprehensively obtaining an accurate inventory result for the location.

[0064] As described in step S32, in the specific implementation, the boundary box of the front end face of the cargo stack, the upper end face of the cargo stack, and the front end face of the pallet of the storage location status image information is marked to obtain the first image information;

[0065] The first image information is input into the trained deep target detection model to output the results of accurate target detection of the front end face, upper end face and front end face of the cargo stack.

[0066] Furthermore, the deep object detection model includes:

[0067] YOLOv8 object detection model.

[0068] When implementing it, please refer to the following steps:

[0069] 1. Storage location image data collection and detailed annotation: First, we deploy high-resolution industrial cameras at designated locations of the stacker to capture a large number of images of the stacking location and build a rich and diverse image database. The core of this stage is to ensure that the collected images can fully cover the various states of the stack, including full stacks, non-full stacks, and empty stacks. Subsequently, these images will be manually annotated in detail to accurately mark the bounding boxes of the front end of the stack, the upper end of the stack, and the front end of the pallet. Each annotation must strictly follow high-quality standards to ensure the accuracy and reliability of the data, laying a solid foundation for model learning.

[0070] 2. Customization and optimization training of deep learning models: We selected the advanced YOLOv8 as the target detection algorithm framework, relying on its efficient detection capabilities and real-time performance. We imported the carefully annotated image dataset into this model and carried out detection task training for three specific areas of the cargo stack. This process deeply involves the fine-tuning of model hyperparameters, such as learning rate, batch size, etc., and the use of strategies such as early stopping and learning rate decay to prevent overfitting and accelerate convergence. In addition, based on the initial training feedback, we flexibly adjusted the network structure, such as increasing or decreasing the number of layers, changing the size of the feature map, etc., in order to maximize detection accuracy while maintaining computational efficiency.

[0071] 3. High-precision target detection output processing: After completing the YOLOv8 model training, we extract accurate target detection results for the front end, upper end, and front end of the pallet from the model output. These data are further cleaned and preprocessed, such as removing redundant information and correcting false detections, to provide accurate and orderly input for the subsequent counting algorithm. Through this series of rigorous post-processing steps, data quality is ensured, laying a solid data foundation for the realization of efficient and accurate inventory counting algorithms.

[0072] In the specific implementation, as described in step S4, the number of goods in the cargo stack to be calculated is calculated by using the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet through a preset counting algorithm, including:

[0073] Filter the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet using a preset confidence threshold, retain the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet that are higher than the preset confidence, and obtain the second image information;

[0074] Performing an image boundary overflow check on the second image information, and if the second image information is not completely within the preset image range, adjusting the position or viewing angle of the industrial camera so that the second image information is completely within the preset image range, thereby obtaining third image information;

[0075] Performing effective quantity preliminary verification on the front end face, upper end face and front end face of the pallet in the third image information to obtain different cargo position status information; wherein the different cargo position status information includes: full stack, half stack, empty stack, empty cargo position, and only pallet storage information;

[0076] Perform secondary identification on the detection frames of the front end face, the upper end face and the front end face of the pallet of the third image information, remove redundant information of non-local stack positions, and obtain fourth image information; wherein the fourth image information carries status information of different cargo positions after removing redundant information of non-local stack positions;

[0077] Dividing the fourth image information into layers, and identifying the number of valid layers in the fourth image information and the position and range information of each layer;

[0078] Using the result of the layer division, determining the full layer state of the fourth image information to obtain full layer number information;

[0079] The quantity of goods in the stack to be calculated is calculated using the full layer number information in the fourth image information, the preset standard quantity of goods in each layer, and the different cargo position status information after removing the redundant information of non-primary stack positions in the fourth image information.

[0080] Furthermore, the dividing the fourth image information into layers, and identifying the number of valid layers in the fourth image information and the position and range information of each layer, includes:

[0081] Calculating the vertical distance between each frame in the fourth image information, and distinguishing different cargo levels in combination with a preset same-layer height threshold;

[0082] By utilizing the different cargo levels, the detection frames belonging to the same level are sorted from large to small according to the vertical distance, and the number of valid layers in the fourth image information and the position and range information of each layer are identified.

[0083] Furthermore, the step of determining the full layer status of the fourth image information by using the result of the layer division to obtain the full layer number information includes:

[0084] Using the result of hierarchical division, the length of the front end surface coverage of each layer is obtained;

[0085] Using the detection information of the front face of the pallet and the length covered by the front face of each layer, the actual length of each layer and the standard length of the pallet are obtained;

[0086] If the difference between the actual length of each layer and the standard length of the pallet is within the preset full layer determination threshold range, the layer is determined to be a full layer, and the full layer number information is obtained.

[0087] For specific implementation, please refer to the following steps:

[0088] 1. Effective screening of target detection results: We set strict confidence thresholds for the detection results of the front end, upper end and front end of the pallet output by the previous deep learning model. Only high-quality detection results with confidence higher than the preset value are retained to filter out uncertain or erroneous recognitions, ensuring that the data entering the counting stage is highly credible.

[0089] 2. Image boundary overflow check: Perform boundary checks on the selected high-confidence detection frames to confirm whether they are completely within the image range. This includes verifying whether the upper end of the cargo stack touches the top or side of the image, whether the front end of the cargo stack exceeds the top, bottom or side of the image, and whether the front end of the pallet exceeds the bottom or side of the image. If the detection frame overflows, the system will prompt to adjust the position or viewing angle of the industrial camera to optimize visual coverage and improve recognition integrity.

[0090] 3. Preliminary verification of effective quantity: Before officially starting the counting algorithm, a preliminary verification is performed based on the number of detections of the front end, upper end, and front end of the pallet to distinguish different cargo position states, including full stacks, half stacks, empty stacks, empty cargo positions, and only pallets stored. This pre-check step is crucial for quickly identifying and classifying the current status of cargo positions, and then adopting targeted counting strategies to further improve the accuracy and efficiency of inventory counting.

[0091] 4. Effective fine screening of key faces of cargo stacks: To eliminate the interference of adjacent cargo stacks in the picture, this step conducts secondary screening of the detection frames of the front face, upper face and front face of the pallet of the initially screened cargo stacks, and eliminates redundant information of non-target stacks. Through advanced image analysis technology, it is ensured that only effective detection frames directly related to the target cargo stack are retained, which improves the pertinence and accuracy of counting.

[0092] 5. Identification of the number of effective layers: After determining the effective detection frame on the front end of the cargo stack, the vertical distance between each frame (Y-axis direction) is calculated, combined with the preset height threshold of the same layer, to distinguish different cargo layers. The detection frames belonging to the same layer are sorted from large to small according to the Y value, so as to clearly define the position and range of each layer.

[0093] 6. Full layer status determination: Based on the layers divided in S9, analyze the length covered by the front end of each layer, and combine it with the detection information of the front end of the pallet to obtain the actual length of each layer and the standard length of the pallet. A full layer determination threshold is introduced. If the difference between the length of a layer and the length of the pallet is within the allowable error range, the layer is considered to be a full layer. Taking into account the actual stacking rules, the upper limit of the number of non-full layers that may exist in the entire stack is two layers. If exactly two non-full layers are identified, the system assumes that the height of these two layers is combined and treated as a full layer to adapt to special stacking situations.

[0094] 7. Quantity: The system calculates the exact quantity of goods in the entire stack. This step integrates the previous analysis results, realizes the efficient conversion from local details to the overall total, and ensures the rapid and accurate inventory counting.

[0095] It should be noted that for the system structure Figure 2 The algorithms involved in S4 in the following are described in detail. Since 2D vision is used in this implementation, it cannot provide depth information. We must impose certain shape constraints on the materials in the target stacking position and follow specific stacking rules to set reasonable stacking restrictions. First, the stacking criteria of the visual pallet library are described.

[0096] In the contemporary industrial scene, many manufacturers have widely adopted robotic technology to perform the task of palletizing goods, which has significantly improved the efficiency of the operation and ensured the accuracy and neatness of the palletizing. Robotic palletizing is not only fast, but also can strictly follow the preset sequence and rules to achieve consistency in the pallet shape and optimize the utilization of storage space. In this way, not only human errors are reduced, but also the degree of automation of the production process is enhanced, further improving the safety and standardization of the overall operation.

[0097] The identification objects of this visual inventory follow the efficient palletizing criteria in a typical industrial environment to ensure the accuracy and practicality of the inventory. The specific operation starts with the selected corner position, and then the goods are stacked in strict accordance with the order of columns (or rows). Each column (or row) must be completely filled before the next row is built. Any skipping is strictly prohibited during this period to maintain the stability and consistency of the structure. After the current level is fully stacked, the next layer will be stacked upwards until the entire cargo stack is successfully constructed according to the established specifications.

[0098] In specific implementation, the following beneficial effects may be achieved:

[0099] It can be understood that the technical solution provided by the present invention, by deploying high-definition industrial cameras on the stacker and combining them with advanced visual algorithms, can conduct comprehensive image acquisition and analysis of the goods in the warehouse, which can not only capture the detailed status of the cargo stack, but also automatically identify the distribution and counting of goods through algorithms, and adapt to a variety of stacking rules to ensure the real-time and accuracy of the inventory results. Through intelligent data processing and analysis, the present invention can effectively cope with the various challenges in warehouse inventory and provide enterprises with an economical, efficient and safe new way of automated inventory management.

[0100] 1. Improve inventory efficiency and reduce costs: By integrating machine vision technology to replace traditional manual inventory, the inventory speed is greatly improved, while reducing manual intervention and reducing long-term operating costs.

[0101] 2. Ensure personnel safety and adaptability to the vertical warehouse environment: Adopt non-invasive visual inventory methods to prevent personnel from directly entering dangerous vertical warehouse areas, significantly improving operational safety.

[0102] 3. Enhance inventory accuracy and reliability: Use high-definition cameras with advanced image recognition algorithms to directly observe and analyze physical inventory, overcoming the defect that RFID technology cannot intuitively reflect the actual inventory status, and maintaining highly accurate inventory results even in complex environments.

[0103] 4.5G-enabled terminal transmission technology: Innovatively apply 5G high-speed, low-latency communication technology to terminal devices as a key tool for inventory data transmission, ensuring that image data can be transmitted to remote servers instantly and efficiently, laying the foundation for real-time monitoring and analysis.

[0104] 5. Establishment of standardized cargo stacking principles: A set of scientific and reasonable cargo stacking principles are defined, aiming to achieve accurate identification of cargo stacking structures through visual inspection technology. These principles not only optimize the standardized process of inventory management, but also greatly enhance the accuracy and efficiency of inventory counting.

[0105] 6. In-depth optimization and innovation of the visual inventory algorithm: The full-layer judgment mechanism in the inventory algorithm and the introduction of pallet detection enrich the judgment of the number of layers in the visual inventory counting.

[0106] 7. Preliminary quantity validity prediction mechanism: Before entering the detailed counting process, the system adds a pre-inspection step to quickly diagnose the status of the storage location, including but not limited to full stacks, empty stacks, empty storage locations or pallets stored separately, and gives guiding suggestions accordingly. This mechanism helps to eliminate abnormal situations in advance, ensure the continuity and accuracy of the subsequent counting process, and further improve the overall efficiency of the inventory.

[0107] See also Figure 3 , Figure 3 The present invention is a schematic diagram of a vertical inventory counting system combining 5G transmission and machine vision according to an exemplary embodiment, wherein the system includes:

[0108] The acquisition module 10 is used to obtain the issued inventory counting requirement task instruction;

[0109] The first processing module 20 is used to capture the image information of the current state of the storage location using a preset industrial camera based on the issued inventory counting demand task instruction;

[0110] The second processing module 30 is used to transmit the storage location status image information to a preset server in real time based on the use of a preset 5G terminal, and the server receives the storage location status image information, and uses a preset target detection algorithm to perform in-depth analysis on the storage location status image information to obtain the result of accurate target detection of the front end face, upper end face and front end face of the cargo stack;

[0111] The third processing module 40 is used to calculate the quantity of goods in the cargo stack to be calculated by a preset counting algorithm based on the result of the precise target detection of the front end face, the upper end face and the front end face of the pallet;

[0112] The fourth processing module 50 is used to feed back the quantity of goods in the to-be-calculated stack to the preset system for storing the disk warehouse data in real time. After receiving the quantity of goods in the to-be-calculated stack, the preset system for storing the disk warehouse data compares the quantity of goods in the to-be-calculated stack with the data of the corresponding stack stored in the system. If there is a difference, the corresponding storage location is marked, and the automatic correction process of the inventory record is triggered manually or directly.

[0113] See also Figure 4 , Figure 4 The present invention is a schematic diagram of a vertical warehouse inventory counting device combining 5G transmission and machine vision according to an exemplary embodiment, wherein the device includes:

[0114] A memory 41 on which an executable program is stored;

[0115] The processor 42 is used to execute the executable program in the memory 41 to implement the steps of any one of the above methods.

[0116] It can be understood that the same or similar parts of the above embodiments can be referenced to each other, and the contents not described in detail in some embodiments can refer to the same or similar contents in other embodiments.

[0117] It should be noted that, in the description of the present invention, the terms "first", "second", etc. are only used for descriptive purposes and cannot be understood as indicating or implying relative importance. In addition, in the description of the present invention, unless otherwise specified, the meaning of "plurality" refers to at least two.

[0118] Any process or method description in a flowchart or otherwise described herein may be understood to represent a module, segment or portion of code that includes one or more executable instructions for implementing the steps of a specific logical function or process, and the scope of the preferred embodiments of the present invention includes alternative implementations in which functions may not be performed in the order shown or discussed, including performing functions in a substantially simultaneous manner or in the reverse order depending on the functions involved, which should be understood by those skilled in the art to which the embodiments of the present invention belong.

[0119] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware or a combination thereof. In the above-mentioned embodiments, a plurality of steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented by hardware, as in another embodiment, it can be implemented by any one of the following technologies known in the art or their combination: a discrete logic circuit having a logic gate circuit for implementing a logic function for a data signal, a dedicated integrated circuit having a suitable combination of logic gate circuits, a programmable gate array (PGA), a field programmable gate array (FPGA), etc.

[0120] A person skilled in the art may understand that all or part of the steps in the method for implementing the above-mentioned embodiment may be completed by instructing related hardware through a program, and the program may be stored in a computer-readable storage medium, which, when executed, includes one or a combination of the steps of the method embodiment.

[0121] In addition, each functional unit in each embodiment of the present invention may be integrated into a processing module, or each unit may exist physically separately, or two or more units may be integrated into one module. The above-mentioned integrated module may be implemented in the form of hardware or in the form of a software functional module. If the integrated module is implemented in the form of a software functional module and sold or used as an independent product, it may also be stored in a computer-readable storage medium.

[0122] The storage medium mentioned above can be a read-only memory, a magnetic disk or an optical disk, etc.

[0123] In the description of this specification, the description with reference to the terms "one embodiment", "some embodiments", "examples", "specific examples", or "some examples" means that the specific features, structures, materials or characteristics described in conjunction with the embodiment or example are included in at least one embodiment or example of the present invention. In this specification, the schematic representation of the above terms does not necessarily refer to the same embodiment or example. Moreover, the specific features, structures, materials or characteristics described may be combined in any one or more embodiments or examples in a suitable manner.

[0124] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are exemplary and are not to be construed as limitations of the present invention. A person skilled in the art may change, modify, replace and vary the above embodiments within the scope of the present invention.

Claims

The vertical warehouse inventory method combining 1.5G transmission and machine vision is characterized by: The method comprises: Receive the issued inventory count requirement task instructions; Based on the issued inventory counting demand task instruction, a preset industrial camera is used to capture image information of the current status of the storage location; The storage location status image information is transmitted in real time to a preset server end by using a preset 5G terminal. The server end receives the storage location status image information, and uses a preset target detection algorithm to perform in-depth analysis on the storage location status image information to obtain accurate target detection results of the front end face, upper end face and front end face of the cargo stack; Utilizing the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet, the quantity of goods in the cargo stack to be counted is calculated by a preset counting algorithm; The quantity of goods in the stack to be calculated is fed back in real time to the preset system for storing the disk warehouse data. After receiving the quantity of goods in the stack to be calculated, the preset system for storing the disk warehouse data compares the quantity of goods in the stack to be calculated with the data of the corresponding stack stored in the system. If there is a difference, the corresponding storage location is marked, and the automatic correction process of the inventory record is triggered through manual review or directly.

2. The method according to claim 1, characterized in that The method of using a preset target detection algorithm to deeply analyze the image information of the storage location status to obtain the results of accurate target detection of the front end face, the upper end face and the front end face of the pallet, includes: Marking the boundary box of the front end face of the cargo stack, the upper end face of the cargo stack, and the front end face of the pallet of the storage location status image information to obtain the first image information; The first image information is input into the trained deep target detection model to output the results of accurate target detection of the front end face, upper end face and front end face of the cargo stack.

3. The method according to claim 2, characterized in that The deep object detection model includes: YOLOv8 object detection model.

4. The method according to claim 1, characterized in that: The method uses the result of the precise target detection of the front end face, the upper end face and the front end face of the pallet to calculate the quantity of goods in the cargo stack to be calculated through a preset counting algorithm, including: Filter the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet using a preset confidence threshold, retain the results of the precise target detection of the front end face, the upper end face and the front end face of the pallet that are higher than the preset confidence, and obtain the second image information; Performing an image boundary overflow check on the second image information, and if the second image information is not completely within the preset image range, adjusting the position or viewing angle of the industrial camera so that the second image information is completely within the preset image range, thereby obtaining third image information; Performing effective quantity preliminary verification on the front end face, upper end face and front end face of the pallet in the third image information to obtain different cargo position status information; wherein the different cargo position status information includes: full stack, half stack, empty stack, empty cargo position, and only pallet storage information; Perform secondary identification on the detection frames of the front end face, the upper end face and the front end face of the pallet of the third image information, remove redundant information of non-local stack positions, and obtain fourth image information; wherein the fourth image information carries status information of different cargo positions after removing redundant information of non-local stack positions; Dividing the fourth image information into layers, and identifying the number of valid layers in the fourth image information and the position and range information of each layer; Using the result of the layer division, determining the full layer state of the fourth image information to obtain full layer number information; The quantity of goods in the stack to be calculated is calculated using the full layer number information in the fourth image information, the preset standard quantity of goods in each layer, and the different cargo position status information after removing the redundant information of non-primary stack positions in the fourth image information.

5. The method according to claim 4, characterized in that The step of dividing the fourth image information into layers and identifying the number of valid layers in the fourth image information and the position and range information of each layer includes: Calculating the vertical distance between each frame in the fourth image information, and distinguishing different cargo levels in combination with a preset same-layer height threshold; By utilizing the different cargo levels, the detection frames belonging to the same level are sorted from large to small according to the vertical distance, and the number of valid layers in the fourth image information and the position and range information of each layer are identified.

6. The method according to claim 4, characterized in that The step of determining the full layer state of the fourth image information by using the result of the layer division to obtain the full layer number information includes: Using the result of hierarchical division, the length of the front end surface coverage of each layer is obtained; Using the detection information of the front face of the pallet and the length covered by the front face of each layer, the actual length of each layer and the standard length of the pallet are obtained; If the difference between the actual length of each layer and the standard length of the pallet is within the preset full layer determination threshold range, the layer is determined to be a full layer, and the full layer number information is obtained.

7. A vertical warehouse inventory counting system combining 5G transmission and machine vision, applied to a vertical warehouse inventory counting method combining 5G transmission and machine vision as described in any one of claims 1 to 5, characterized in that: The system comprises: The acquisition module is used to obtain the issued inventory counting requirement task instructions; A first processing module is used to capture the image information of the current state of the storage location using a preset industrial camera based on the issued inventory counting demand task instruction; The second processing module is used to transmit the storage location status image information to a preset server in real time based on the use of a preset 5G terminal. The server receives the storage location status image information and uses a preset target detection algorithm to perform in-depth analysis on the storage location status image information to obtain the result of accurate target detection of the front end face, upper end face and front end face of the cargo stack; A third processing module is used to calculate the quantity of goods in the cargo stack to be calculated by a preset counting algorithm based on the result of the precise target detection of the front end face, the upper end face and the front end face of the pallet; The fourth processing module is used to feed back the quantity of goods in the to-be-calculated stack to the preset system for storing the disk warehouse data in real time. After receiving the quantity of goods in the to-be-calculated stack, the preset system for storing the disk warehouse data compares the quantity of goods in the to-be-calculated stack with the data of the corresponding stack stored in the system. If there is a difference, the corresponding storage location is marked, and the automatic correction process of the inventory record is triggered manually or directly. The vertical warehouse inventory counting equipment that combines 8.5G transmission with machine vision is characterized by: The device comprises: A memory having an executable program stored therein; A processor, configured to execute the executable program in the memory to implement the steps of the method according to any one of claims 1 to 6.

Images