A Storage Cabinet Cluster Management Method Based on Internet of Things Technology
By acquiring three-view images of items through cameras and detecting dimensions using the YOLO model, and combining these with storage cabinet status parameters to calculate a comprehensive score, the problem of low utilization and high reuse rate in storage cabinet cluster management is solved, achieving efficient utilization of storage cabinets and optimized user experience.
Patent Information
- Application Number
- CN202411235578.5
- Authority / Receiving Office
- CN · China
- Patent Type
- Patents(China)
- Current Assignee / Owner
- Filing Date
- 2024-09-04
- Publication Date
- 2024-11-19
- Estimated Expiration
- 2044-09-04
AI Technical Summary
Existing storage cabinet cluster management systems suffer from low storage cabinet utilization and high reuse rates, lacking intelligent analysis and optimization, resulting in resource waste and shortened storage cabinet lifespan.
The system acquires three-view images of the items to be stored using a camera, uses a YOLO deep learning model to detect the size of the items, selects the best storage cabinet based on the size matching strategy, calculates a comprehensive score based on the storage cabinet status parameters, and selects the storage cabinet with the highest comprehensive score for opening.
It improves the utilization efficiency of storage cabinets, enables precise management of the location and size of stored items, optimizes the operating status and user experience of storage cabinets, and reduces measurement errors and manual intervention in traditional methods.
Smart Images

Figure CN118736728B_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the technical field of Internet of Things, and in particular to a storage cabinet cluster management method based on Internet of Things technology control. Background Art
[0002] With the development of Internet of Things technology and intelligent control technology, storage cabinet cluster management systems have gradually been widely used in various application scenarios, such as logistics distribution, smart express cabinets, smart storage cabinets, etc. However, the existing storage cabinet cluster management system has some problems and shortcomings in practical applications: Low storage cabinet utilization: The existing system often lacks intelligent analysis and optimization of the storage cabinet space, and the size of the storage cabinet is manually selected, resulting in low storage cabinet space utilization and serious waste of resources; High reuse: After the existing storage cabinet cluster management system manually selects the size of the storage cabinet, it is often allocated according to a fixed sequence, resulting in high reuse, which affects the service life of the storage cabinet. Summary of the invention
[0003] Based on the above-mentioned shortcomings of the prior art, the purpose of the present invention is to provide a storage cabinet cluster management method based on Internet of Things technology control to solve the above-mentioned technical problems.
[0004] To achieve the above object, the present invention provides the following technical solution: a storage cabinet cluster management method based on Internet of Things technology control, comprising:
[0005] Obtain three-view images of the object to be stored through a camera, and input the three-view images into a real-time size detection model to obtain the length, width and height of the object to be stored;
[0006] The optimal size of the storage cabinet is selected according to the length, width and height of the items to be stored according to the preset size matching strategy;
[0007] Obtain status parameters of all free storage cabinets that meet the optimal size, and calculate the comprehensive score of the storage cabinet according to the status parameters;
[0008] The comprehensive scores are sorted in descending order, and the storage cabinet with the highest comprehensive score is selected for the cabinet opening operation.
[0009] The present invention is further configured such that the construction logic of the real-time size detection model includes:
[0010] Get the history containing the three-view images of the stored objects, use the annotation tools to annotate the objects in the images, and draw the bounding boxes;
[0011] According to the preset model input image size, the three-view images of the history containing the stored items are scaled to a uniform size, and the scaled three-view images of the history containing the stored items are divided into a test set and a validation set;
[0012] Select the YOLO deep learning model, download the pre-trained model weights, configure the model parameters, use the labeled data as input, and use the bounding box pixel coordinates and confidence as output to train the YOLO deep learning model. The model parameters include the input image size and learning rate.
[0013] Use the validation dataset to evaluate model performance and prevent overfitting.
[0014] The present invention is further configured such that, in the YOLO deep learning model training, the calculation logic of the loss function is: , Coordinate loss is used to measure the difference between the predicted bounding box coordinates and the true bounding box coordinates. The calculation logic is: L coord = ω coord ∑ i =0 S 2 ∑ j=0 B 1 ij obj [ ( x i - x ^ i ) 2 + ( y i - y ^ i ) 2 + ( w i - w ^ i ) 2 + ( h i - h ^ i ) 2 ] , Indicates the size of the grid, represents the number of bounding boxes predicted by the grid, is the first indicator function, when The first 1 if the bounding box contains the object, otherwise 0. , are the coordinates of the center of the predicted bounding box, , are the coordinates of the center of the true bounding box, , are the width and height of the predicted bounding box, , are the actual bounding box width and height, is the coordinate loss weight; It is the confidence loss, which is used to measure the difference between the confidence that the predicted bounding box contains the object and the actual situation. The calculation logic is: , is the confidence score of the prediction, which is the probability that the predicted bounding box contains the object multiplied by the accuracy of the bounding box, is the true confidence score, that is, whether the bounding box contains the object. When the bounding box contains the object, is 1, when the bounding box does not contain the object, is 0, is the second indicator function, when The first If the bounding box does not contain the object, it is 1, otherwise it is 0. is the confidence loss weight.
[0015] The present invention is further configured to include:
[0016] Get the output bounding box pixel coordinates and confidence, and convert the bounding box pixel coordinates to the actual size. The conversion logic is: ,in, is the actual coordinate, is the camera intrinsic parameter matrix, are the bounding box pixel coordinates, For depth.
[0017] The present invention is further configured to select the optimal size of the storage cabinet according to the length, width and height of the items to be stored according to a preset size matching strategy, including:
[0018] Get the length, width and height information of each storage cabinet, recorded as ;
[0019] When the length, width and height of the items to be stored meet the constraints, find the minimum target value and set the minimum target value as the optimal size of the storage cabinet, where the constraints are , the target value is .
[0020] The present invention is further configured to obtain status parameters of all idle storage cabinets that meet the optimal size, the status parameters including: cabinet opening response time, cabinet opening success rate and allocation utilization rate; and calculate a comprehensive score of the storage cabinet based on the status parameters, including:
[0021] Normalize the cabinet opening response time, cabinet opening success rate and allocation utilization rate;
[0022] The user experience index is calculated based on the normalized cabinet opening response time and cabinet opening success rate;
[0023] The overall score of the storage cabinet is calculated based on the user experience index and the allocated utilization rate.
[0024] The present invention is further configured to calculate the user experience index according to the normalized cabinet opening response time and cabinet opening success rate, and the calculation logic of the user experience index is: , the calculation logic of the comprehensive score of the storage cabinet is: ,in, is the user experience index, For the success rate of cabinet opening, is the cabinet opening response time, The overall rating of the storage cabinet. The allocated utilization rate of the storage cabinet.
[0025] The present invention provides a storage cabinet cluster management method based on Internet of Things technology control. The method obtains three-view images of items to be stored through a camera, inputs the three-view images into a real-time size detection model, and obtains the length, width and height of the items to be stored; selects the optimal size of the storage cabinet according to the length, width and height of the items to be stored according to a preset size matching strategy; obtains state parameters of all idle storage cabinets that meet the optimal size, and calculates the comprehensive score of the storage cabinet according to the state parameters; sorts the comprehensive scores in descending order, and selects the storage cabinet with the highest comprehensive score for opening the cabinet. The beneficial effects produced include:
[0026] 1. Improve the utilization efficiency of storage cabinets: Use the YOLO deep learning model to detect the three-dimensional images of stored items and draw bounding boxes, obtain the length, width and height information of stored items in real time, accurately predict the size of stored items and their optimal placement in the storage cabinet, and achieve precise management of the location and size of stored items, reducing the measurement errors and manual intervention that may occur in traditional methods, thereby improving the space utilization of storage cabinets;
[0027] 2. Efficient comprehensive status evaluation: Based on the storage cabinet status information, including the cabinet opening response time, cabinet opening success rate and allocation utilization rate, combined with the characteristic data of the actual stored items, the operation status and use effect of the storage cabinet can be comprehensively evaluated, so as to carry out effective maintenance and optimization;
[0028] 3. Improve user experience: Comprehensively considering factors such as user locker opening response time, locker opening success rate and user experience index can optimize the operation process and user experience of storage lockers and enhance user satisfaction.
[0029] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS
[0030] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following briefly introduces the drawings required for describing the embodiments. Obviously, the drawings described below are only some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without creative work. In the drawings:
[0031] Figure 1 The present invention is a flowchart of a storage cabinet cluster management method based on Internet of Things technology control, showing an exemplary embodiment of the present invention. DETAILED DESCRIPTION
[0032] The following will describe the embodiments of the present invention with reference to the accompanying drawings and preferred embodiments. Those skilled in the art can easily understand other advantages and effects of the present invention from the contents disclosed in this specification. The present invention can also be implemented or applied through other different specific embodiments, and the details in this specification can also be modified or changed in various ways based on different viewpoints and applications without departing from the spirit of the present invention. It should be understood that the preferred embodiments are only for illustrating the present invention, not for limiting the scope of protection of the present invention.
[0033] It should be noted that the illustrations provided in the following embodiments are only schematic illustrations of the basic concept of the present invention, and thus the drawings only show components related to the present invention rather than being drawn according to the number, shape and size of components in actual implementation. In actual implementation, the type, quantity and proportion of each component may be changed arbitrarily, and the component layout may also be more complicated.
[0034] In the following description, numerous details are discussed to provide a more thorough explanation of the embodiments of the present invention. However, it is obvious to those skilled in the art that the embodiments of the present invention can be implemented without these specific details. In other embodiments, well-known structures and devices are shown in the form of block diagrams rather than in detail to avoid making the embodiments of the present invention difficult to understand.
[0035] A storage cabinet cluster management method based on Internet of Things technology control, such as Figure 1 As shown, including:
[0036] Obtain three-view images of the object to be stored through a camera, and input the three-view images into a real-time size detection model to obtain the length, width and height of the object to be stored;
[0037] The optimal size of the storage cabinet is selected according to the length, width and height of the items to be stored according to the preset size matching strategy;
[0038] Obtain status parameters of all free storage cabinets that meet the optimal size, and calculate the comprehensive score of the storage cabinet according to the status parameters;
[0039] The comprehensive scores are sorted in descending order, and the storage cabinet with the highest comprehensive score is selected for the cabinet opening operation.
[0040] Specifically, in order to obtain three views of an object, including a front view, a side view, and a top view, at least three cameras are required. The position and angle of each camera should be set to include:
[0041] Front view camera: Position: placed in front of the object, keeping a proper distance from the object; Angle: perpendicular to the ground, with the camera optical axis parallel to the front of the object; Function: obtain the front view of the object, mainly used to extract the height and width information of the object;
[0042] Side view camera: Position: placed on the side of the object, keeping an appropriate distance from the object; Angle: perpendicular to the ground, with the camera optical axis parallel to the side of the object; Function: obtain the side view of the object, mainly used to extract the depth and height information of the object;
[0043] Top view camera: Position: placed directly above the object, at an appropriate height from the object; Angle: parallel to the ground, with the camera optical axis perpendicular to the ground; Function: obtain the top view of the object, mainly used to extract the width and depth information of the object;
[0044] In order to ensure that the acquired image accurately reflects the current state of the object, it is necessary to synchronously shoot three-view images. The method of synchronous shooting includes a trigger mechanism and time synchronization, which is a prior art and will not be described in detail here. Furthermore, when acquiring images, the parameters are set including: image resolution, setting the camera resolution to ensure that the image is clear and can accurately reflect the details of the object; exposure time, adjusting the camera exposure time to avoid overexposure or underexposure to ensure image quality; white balance, setting the camera white balance to ensure accurate color restoration; image correction is performed on the acquired images, including: distortion correction, using camera calibration parameters to perform distortion correction on the image to remove the distortion caused by the lens; color correction, performing color correction on the image according to the ambient light conditions to ensure color consistency; after correction, three-view fusion is performed, and the steps include: feature point matching, finding common feature points in the three-view images, and aligning the images using image registration technology; coordinate conversion, converting the three-view images into a unified coordinate system according to the relative position and angle of the camera; depth information fusion, fusing the depth information of the front view, side view and top view to generate a three-dimensional model of the object.
[0045] The present invention is further configured such that the construction logic of the real-time size detection model includes:
[0046] The three-view image of the stored object is obtained from the history, and the object in the image is annotated using an annotation tool to draw a bounding box. Specifically, the three-view image of the stored object is opened from the history, and the bounding box of the object is manually drawn to record the boundary coordinates of the object.
[0047] According to the preset model input image size, the historical three-view images containing stored items are scaled to a uniform size, and the scaled historical three-view images containing stored items are divided into a test set and a validation set; specifically, the preset model input image sizes include 416x416 and 600x600 pixels, which are not restricted here, and the aspect ratio of the image is maintained during the scaling process to prevent image distortion.
[0048] Select the YOLO deep learning model, download the pre-trained model weights, configure the model parameters, use the labeled data as input, and use the bounding box pixel coordinates and confidence as output to train the YOLO deep learning model, where the model parameters include the input image size and learning rate. The learning rate is an important hyperparameter in machine learning and deep learning model training, which determines the step size of the model to update the parameters at each iteration. It plays a key role in the gradient descent optimization algorithm and affects the speed and effect of model convergence; the present invention is further configured such that, in the YOLO deep learning model training, the calculation logic of the loss function is: , Coordinate loss is used to measure the difference between the predicted bounding box coordinates and the true bounding box coordinates. The calculation logic is: L coord = ω coord ∑ i =0 S 2 ∑ j=0 B 1 ij obj [ ( x i - x ^ i ) 2 + ( y i - y ^ i ) 2 + ( w i - w ^ i ) 2 + ( h i - h ^ i ) 2 ] , Indicates the size of the grid, represents the number of bounding boxes predicted by the grid, is the first indicator function, when The first 1 if the bounding box contains the object, otherwise 0. , are the coordinates of the center of the predicted bounding box, , are the coordinates of the center of the true bounding box, , are the width and height of the predicted bounding box, , are the actual bounding box width and height, is the coordinate loss weight; It is the confidence loss, which is used to measure the difference between the confidence that the predicted bounding box contains the object and the actual situation. The calculation logic is: , is the confidence score of the prediction, which is the probability that the predicted bounding box contains the object multiplied by the accuracy of the bounding box, is the true confidence score, that is, whether the bounding box contains the object. When the bounding box contains the object, is 1, when the bounding box does not contain the object, is 0, is the second indicator function, when The first If the bounding box does not contain the object, it is 1, otherwise it is 0. is the confidence loss weight; specifically, set the coordinate loss weight and confidence loss weight In a feasible embodiment of the present invention, the coordinate loss weight The value of is 5, which is used to balance the various parts of the loss, the confidence loss weight The value is 0.5, which is used to reduce the confidence impact on non-existent objects; through the YOLO model, the center coordinates of the bounding box of each grid are predicted , ,width and height and the confidence score For each grid containing an object, the error between the predicted bounding box coordinates and the true bounding box coordinates is calculated, and a weighted sum is performed to obtain the coordinate loss. For each grid, the error between the predicted confidence score and the true confidence score is calculated. For grids containing objects and grids not containing objects, a weighted sum is performed to obtain the confidence loss. The coordinate loss and the confidence loss are added together to obtain the total loss function. The parameters of the YOLO model are optimized through the back-propagation algorithm to minimize the total loss function.
[0049] Use the validation data set to evaluate the model performance and prevent overfitting. According to the evaluation results of the validation set, adjust the model parameters and training strategies to prevent overfitting. Through the above steps, the YOLO deep learning model can be effectively trained to detect and classify the size of objects using three-view images and annotated data. This process includes multiple links such as data collection and annotation, data preprocessing, model selection and configuration, model training and verification, etc. Each link needs to be carefully designed and implemented to ensure the accuracy and robustness of the model.
[0050] The present invention is further configured to include:
[0051] Get the output bounding box pixel coordinates and confidence, and convert the bounding box pixel coordinates to the actual size. The conversion logic is: ,in, is the actual coordinate, is the camera intrinsic parameter matrix, are the bounding box pixel coordinates, is the depth; specifically, in the output of the YOLO model, the prediction result of each bounding box contains the following information: the center coordinates of the bounding box , represents the position of the bounding box in the image; the width of the bounding box and height , indicating the size of the bounding box; confidence , which represents the probability that the bounding box contains an object and the accuracy of the bounding box. In order to convert the predicted bounding box pixel coordinates into the size in the real world, the camera calibration parameters are needed, the camera intrinsic parameter matrix Contains the camera's internal parameter information, including focal length and principal point coordinates. Usually expressed as: , and The camera is and The focal length of the direction, and is the coordinate of the principal point of the image, usually the center of the image, and the pixel coordinates of the bounding box are obtained from the output of the YOLO model and depth information , depth information It can be obtained through a depth camera or estimated through a monocular vision depth estimation algorithm. This is an existing technology and will not be described in detail here. Through the above steps, the bounding box pixel coordinates and depth information output by the YOLO model can be converted into three-dimensional coordinates in the real world. This step is very critical in practical applications because it can map the information in the two-dimensional image to the three-dimensional space, thereby achieving accurate measurement of the size and position of the object.
[0052] The present invention is further configured to select the optimal size of the storage cabinet according to the length, width and height of the items to be stored according to a preset size matching strategy, including:
[0053] Get the length, width and height information of each storage cabinet, recorded as ;
[0054] When the length, width and height of the items to be stored meet the constraints, find the minimum target value and set the minimum target value as the optimal size of the storage cabinet, where the constraints are , the target value is , the target value does not directly represent the volume of the remaining space in the storage cabinet, but is an indicator to measure whether the storage cabinet is suitable for storing items. The smaller the target value, the more closely the size of the storage cabinet fits the size of the item to be stored, that is, the higher the space utilization rate of the storage cabinet. Among multiple storage cabinets that meet the conditions, the storage cabinet with a smaller target value is more suitable for storing the item because it can make more efficient use of the storage space. In this case, we hope to select a storage cabinet that fits the size of the item as much as possible to minimize the remaining space and thus improve space utilization. Through the above steps, the most suitable storage cabinet can be selected according to the length, width and height of the item to be stored according to the preset size matching strategy. The process of calculating the target value ensures that the selected storage cabinet can fit the size of the item as much as possible, thereby improving space utilization. It can effectively optimize the allocation of storage cabinets, making the storage of items more efficient and reasonable.
[0055] The present invention is further configured to obtain the status parameters of all idle storage cabinets that meet the optimal size, and the status parameters include: cabinet opening response time, cabinet opening success rate and allocation utilization rate; specifically, the cabinet opening response time means the time from when the user issues a cabinet opening instruction to when the cabinet door is opened. The cabinet opening response time can timely understand the response speed of each storage cabinet, identify and improve the storage cabinets with slower response, and improve the overall operation efficiency; the cabinet opening success rate means the proportion of successful cabinet opening operations, and the cabinet opening success rate directly affects the user satisfaction. By monitoring and optimizing the cabinet opening success rate, the failure rate and user dissatisfaction can be reduced; the allocation utilization rate means the frequency of effective use of the storage cabinet. By analyzing the allocation utilization rate, the utilization rate of the storage cabinet can be identified, the allocation of storage resources can be optimized, and the overall storage space utilization efficiency can be improved; the comprehensive score of the storage cabinet is calculated according to the status parameters, including:
[0056] Normalize the cabinet opening response time, cabinet opening success rate, and allocation utilization rate; normalization is an important step in data preprocessing, and its purpose is to convert data of different dimensions or units to a common scale for fair comparison and processing. Through normalization, the influence of different units or dimensions in the data can be eliminated, so that each parameter can be compared and calculated on the same scale. Different parameters may have different units and dimensions, and direct comparison will lead to unfair results. After normalization, all parameters are on the same scale, eliminating the influence of dimension. At the same time, normalization can avoid numerical instability caused by large differences in data ranges and promote the convergence and stability of the model. Common normalization methods include Min-Max Normalization and Standardization. The cabinet opening response time, cabinet opening success rate and allocation utilization rate are normalized. Since the cabinet opening success rate and allocation utilization rate are already between [0, 1], the Min-Max normalization method can be used to make the cabinet opening response time also compared on the same scale, and map the normalized value of the response time to [0, 1], while maintaining the consistency of the evaluation criteria to ensure fair comparison between different parameters.
[0057] The user experience index is calculated according to the normalized cabinet opening response time and the cabinet opening success rate; the present invention is further configured to calculate the user experience index according to the normalized cabinet opening response time and the cabinet opening success rate, and the calculation logic of the user experience index is: Specifically, the exponential transformation makes the influence of parameters more significant, which helps to identify key issues and optimize them in time. The physical meaning of this formula is that the user experience index is exponentially related to the cabinet opening success rate and inversely proportional to the cabinet opening response time. That is, the higher the cabinet opening success rate and the shorter the cabinet opening response time, the greater the user experience index. The user experience index combines the cabinet opening success rate and response time, and can fully reflect the user's feelings about the use of the storage cabinet. By improving the user experience index, user satisfaction and trust can be enhanced;
[0058] The comprehensive score of the storage cabinet is calculated based on the user experience index and the allocation utilization rate. The calculation logic of the comprehensive score of the storage cabinet is: ,in, is the user experience index, For the success rate of cabinet opening, is the cabinet opening response time, The overall rating of the storage cabinet. is the allocation utilization rate of the storage cabinet. Specifically, the physical meaning of this formula is that the comprehensive score of the storage cabinet is jointly affected by the user experience index and the allocation utilization rate. The higher the user experience index, the higher the comprehensive score; the higher the allocation utilization rate, the lower the comprehensive score. This formula uses a logarithmic function to ensure that the score is within a reasonable range. Through the above steps, parameters such as the cabinet opening response time, cabinet opening success rate, and allocation utilization rate can be normalized, and the user experience index and the comprehensive score of the storage cabinet can be calculated. The comprehensive score combines the user experience index and the allocation utilization rate to provide a comprehensive performance indicator to help managers more accurately evaluate the overall status and performance of the storage cabinet. The user experience index and the comprehensive score are automatically calculated by the formula, which can realize the automation and intelligence of storage cabinet management, reduce manual intervention, and improve management efficiency.
[0059] The above embodiments can be implemented in whole or in part by software, hardware, firmware or any other combination. When implemented by software, the above embodiments can be implemented in whole or in part in the form of a computer program product. The computer program product includes one or more computer instructions or computer programs. When the computer instructions or computer programs are loaded or executed on a computer, the process or function described in the embodiment of the present application is generated in whole or in part. The computer can be a general-purpose computer, a special-purpose computer, a computer network, or other programmable device. The computer instructions can be stored in a computer-readable storage medium, or transmitted from one computer-readable storage medium to another computer-readable storage medium. For example, the computer instructions can be transmitted from one website site, computer, server or data center to another website site, computer, server or data center by wired (e.g., infrared, wireless, microwave, etc.). The computer-readable storage medium can be any available medium that a computer can access or a data storage device such as a server or data center that contains one or more available media sets. The available medium can be a magnetic medium (e.g., a floppy disk, a hard disk, a tape), an optical medium (e.g., a DVD), or a semiconductor medium. The semiconductor medium can be a solid-state hard disk.
[0060] It should be understood that the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. A and B can be singular or plural. In addition, the character " / " in this article generally indicates that the associated objects before and after are in an "or" relationship, but it may also indicate an "and / or" relationship. Please refer to the context for specific understanding.
[0061] In this application, "at least one" means one or more, and "more than one" means two or more. "At least one of the following" or similar expressions refers to any combination of these items, including any combination of single or plural items. For example, at least one of a, b, or c can be represented by: a, b, c, ab, ac, bc, or abc, where a, b, c can be single or multiple.
[0062] It should be understood that in the various embodiments of the present application, the size of the serial numbers of the above-mentioned processes does not mean the order of execution. The execution order of each process should be determined by its function and internal logic, and should not constitute any limitation on the implementation process of the embodiments of the present application.
[0063] Those of ordinary skill in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered to be beyond the scope of this application.
[0064] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working processes of the systems, devices and units described above can refer to the corresponding processes in the aforementioned method embodiments and will not be repeated here.
[0065] In the several embodiments provided in the present application, it should be understood that the disclosed system can be implemented in other ways. For example, the device embodiments described above are only schematic. For example, the division of the units is only a logical function division. There may be other division methods in actual implementation, such as multiple units 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 devices or units, which can be electrical, mechanical or other forms.
[0066] The units described as separate components may or may not be physically separated, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed on multiple network units. Some or all of the units may be selected according to actual needs to achieve the purpose of the solution of this embodiment.
[0067] In addition, each functional unit in each embodiment of the present application may be integrated into one processing unit, or each unit may exist physically separately, or two or more units may be integrated into one unit.
[0068] If the functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present application can be essentially or partly embodied in the form of a software product that contributes to the prior art. The computer software product is stored in a storage medium and includes several instructions for a computer device (which can be a personal computer, a server, or a network device, etc.) to perform all or part of the steps of the methods described in the various embodiments of the present application. The aforementioned storage media include: various media that can store program codes, such as USB flash drives, mobile hard disks, read-only memories (ROM), random access memories (RAM), magnetic disks or optical disks.
[0069] The above is only a specific implementation of the present application, but the protection scope of the present application is not limited thereto. Any person skilled in the art who is familiar with the present technical field can easily think of changes or substitutions within the technical scope disclosed in the present application, which should be included in the protection scope of the present application. Therefore, the protection scope of the present application should be based on the protection scope of the claims.
Claims
1. A storage cabinet cluster management method based on Internet of Things technology control, characterized in that: include: Obtain three-view images of the object to be stored through a camera, and input the three-view images into a real-time size detection model to obtain the length, width and height of the object to be stored; The construction logic of the real-time size detection model includes: obtaining three-view images of historical storage items, using annotation tools to annotate the items in the image, and drawing a bounding box; scaling the three-view images of historical storage items to a uniform size according to a preset model input image size, and dividing the scaled three-view images of historical storage items into a test set and a validation set; selecting a YOLO deep learning model, downloading pre-trained model weights, configuring model parameters, taking the annotation data as input, taking the bounding box pixel coordinates and confidence as output, and performing YOLO deep learning model training, wherein the model parameters include the input image size and the learning rate; in the YOLO deep learning model training, the calculation logic of the loss function is: L=L coord +L conf ; Among them, L coord is the coordinate loss, which is used to measure the difference between the predicted bounding box coordinates and the true bounding box coordinates. The coordinate loss L coord The calculation logic is: Among them, S represents the size of the grid, B represents the number of bounding boxes predicted by the grid, is the first indicator function, which is 1 when the jth bounding box in the i-th grid contains an object, otherwise it is 0. i ,y i are the coordinates of the center of the predicted bounding box, is the coordinate of the center of the actual bounding box, w i 、h i are the width and height of the predicted bounding box, are the width and height of the true bounding box, ω coord is the coordinate loss weight; L conf It is the confidence loss, which is used to measure the difference between the confidence of the predicted bounding box containing the object and the actual situation. The confidence loss L conf The calculation logic is: Among them, C i is the confidence score of the prediction, which is the probability that the predicted bounding box contains the object multiplied by the accuracy of the bounding box, is the true confidence score, that is, whether the bounding box contains the object. When the bounding box contains the object, is 1, when the bounding box does not contain the object, is 0, is the second indicator function, which is 1 when the jth bounding box in the i-th grid does not contain an object, otherwise it is 0, ω nojb Confidence loss weight; Use validation data set to evaluate model performance and prevent overfitting; The optimal size of the storage cabinet is selected according to the length, width and height of the items to be stored according to the preset size matching strategy; The state parameters of all free storage cabinets that meet the optimal size are obtained, and the comprehensive score of the storage cabinet is calculated according to the state parameters; the state parameters include: cabinet opening response time, cabinet opening success rate and allocation utilization rate; the comprehensive score of the storage cabinet is calculated according to the state parameters, including: normalizing the cabinet opening response time, cabinet opening success rate and allocation utilization rate; calculating the user experience index according to the normalized cabinet opening response time and cabinet opening success rate; calculating the comprehensive score of the storage cabinet according to the user experience index and the allocation utilization rate; the calculation logic of the user experience index is: Jew=e Roc / Tor ; The calculation logic of the comprehensive score of the storage cabinet is: ORsc=ln(1+Iue / (1+Rua)); Among them, Iue is the user experience index, Roc is the cabinet opening success rate, Tor is the cabinet opening response time, ORsc is the comprehensive score of the storage cabinet, and Rua is the allocation utilization rate of the storage cabinet; The comprehensive scores are sorted in descending order, and the storage cabinet with the highest comprehensive score is selected for the cabinet opening operation.
2. According to the storage cabinet cluster management method based on Internet of Things technology control according to claim 1, it is characterized in that: Also includes: Get the output bounding box pixel coordinates and confidence, and convert the bounding box pixel coordinates to the actual size. The conversion logic is: Among them, (X, Y, Z) are the actual coordinates, K is the camera intrinsic parameter matrix, (u, v) are the bounding box pixel coordinates, and d is the depth.
3. According to the storage cabinet cluster management method based on Internet of Things technology control according to claim 1, it is characterized in that: The optimal size of the storage cabinet is selected according to the length, width and height of the items to be stored according to the preset size matching strategy, including: Get the length, width and height information of each storage cabinet, recorded as (L S ,W S ,H S ); When the length, width and height of the items to be stored meet the constraints, find the minimum target value and set the minimum target value as the optimal size of the storage cabinet, where the constraints are The target value is (L S -L i )*(W S -W i )*(H S -H i ).
