Logistics package tracking and counting method, device, equipment and storage medium

Abstract

The present invention discloses a logistics package tracking and counting method, the method comprising: obtaining a logistics package image, annotating the image, and constructing a logistics package data set; constructing a deep learning network model, and introducing an FC attention mechanism; inputting the logistics package data set into a logistics package deep learning model based on YOLOV11 for training, constructing a loss function, and obtaining a logistics package recognition model; inputting the image of the logistics package into the trained recognition model, identifying and locating the logistics package in the image, and then counting the number of the logistics packages according to the image. The present invention can realize efficient and accurate tracking of packages, improve the accuracy of the logistics package statistics method and the work efficiency of the package statistics, reduce the logistics transportation cost, and thus improve the satisfaction of customer service quality by constructing an intelligent counting system integrating the YOLOV11 model, the FC attention mechanism and the MPDIoU loss function.

Description

Technical Field

[0001] The present invention relates to the field of logistics technology, and in particular to a logistics package tracking and counting method, device, equipment and storage medium. Background Art

[0002] With the acceleration of globalization and the rapid development of e-commerce, the modern logistics industry is facing greater and greater challenges. How to improve logistics transportation efficiency, reduce costs, optimize route planning, and improve customer service quality have become hot topics in the industry. In the process of solving these problems, artificial intelligence technology, especially deep learning, is playing an increasingly important role. The research on intelligent counting of package tracking in logistics transportation center assembly lines is of great significance to improving logistics efficiency, reducing labor costs, and improving customer satisfaction.

[0003] The existing logistics package route tracking and quantity counting methods are inefficient and inaccurate. In order to improve the accuracy of the logistics package counting method, improve the work efficiency of package counting, reduce logistics transportation costs, and thereby improve customer service quality satisfaction, it is necessary to invent a logistics package tracking and counting method, device, equipment and storage medium. Summary of the invention

[0004] The main purpose of the present invention is to solve the technical problems of low efficiency and low accuracy in route tracking and quantity counting of logistics packages in the prior art.

[0005] A first aspect of the present invention provides a logistics package tracking and counting method, the method comprising: Obtaining logistics package images, annotating the images, and constructing a logistics package dataset; Build a deep learning network model and introduce the FC attention mechanism; The logistics package data set is input into the logistics package deep learning model based on YOLOV11 for training, and the loss function is constructed to obtain the logistics package recognition model; The image of the logistics package is input into the trained recognition model to identify and locate the logistics package in the image, and then the number of the logistics package is counted according to the image.

[0006] Optionally, in a first implementation of the first aspect of the present invention, acquiring a logistics package image, annotating the image, and constructing a logistics package dataset includes: Set up cameras at the starting point, midpoint, and end point of the logistics package assembly line; Collecting images of the logistics package at the starting point, the middle point and the end point of the logistics package assembly line; Acquire the image of the logistics package, clean the image, and delete low-quality images; Label the logistics packages in the image, mark the location and type of the logistics packages on the image, and each image corresponds to a txt file containing the labeling information; All annotated images and corresponding annotation files form a logistics package dataset.

[0007] Optionally, in a second implementation of the first aspect of the present invention, constructing a deep learning network model and adding an FC attention mechanism includes: Build deep learning network models; Use the YOLOV11 model as the deep learning network model; Adding FC attention mechanism to the deep learning network model; Among them, the input of the backbone network of the deep learning network model is the preprocessed image, and the output is the feature map; the input of the neck network of the deep learning network model is the feature map, and the output is the adaptive feature fusion map.

[0008] Optionally, in a third implementation of the first aspect of the present invention, the adding of the FC attention mechanism includes: The FC attention mechanism is placed on the enhanced feature extraction network, and the attention mechanism is added to the three effective feature layers extracted by the backbone network; Add FC attention mechanism to the upsampling results in the pyramid network.

[0009] Optionally, in a fourth implementation of the first aspect of the present invention, the logistics package data set is input into a logistics package deep learning model based on YOLOV11 for training, a loss function is constructed, and a logistics package recognition model is obtained, including: Divide the logistics package dataset into training set, validation set and test set; Input the training set to train the YOLOV11-based logistics package deep learning model; Input the validation set to adjust the training parameters of the deep learning model; Input the test set to test the generalization ability of the deep learning model; Construct a loss function of the deep learning model and update the model to obtain a logistics package recognition model based on YOLOV11.

[0010] Optionally, in a fifth implementation of the first aspect of the present invention, the loss function of the YOLOV11-based logistics package identification model is an MPDIoU loss function.

[0011] Optionally, in a sixth implementation of the first aspect of the present invention, inputting the image of the logistics package into the trained recognition model, identifying and locating the logistics package in the image, and then counting the number of the logistics packages according to the image, includes: Input the image of the logistics package into the logistics package recognition model based on YOLOV11; The YOLOV11-based logistics package recognition model recognizes the logistics package in the image; Positioning the logistics package; The number of the logistics packages is counted according to the logistics package image.

[0012] A second aspect of the present invention provides a logistics package tracking and counting device, comprising: A logistics package image acquisition module is used to acquire logistics package images, annotate the images, and construct a logistics package dataset; Build a deep learning network model module, which is used to build a deep learning network model and introduce the FC attention mechanism; The logistics package recognition module is used to input the logistics package data set into the logistics package deep learning model based on YOLOV11 for training, construct a loss function, and obtain a logistics package recognition model; The tracking and statistics module is used to input the image of the logistics package into the trained recognition model, identify and locate the logistics package in the image, and count the number of the logistics package according to the image.

[0013] A third aspect of the present invention provides a computer device, the computer device comprising a memory and at least one processor, the memory storing instructions; The at least one processor calls the instructions in the memory to enable the computer device to execute the logistics package tracking and counting method as described above.

[0014] A fourth aspect of the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein the computer program is loaded by a processor to execute the logistics package tracking and counting method as described above.

[0015] The logistics package tracking and counting method of the present invention, by constructing an intelligent counting system that integrates the YOLOV11 model, adds the FC attention mechanism and the MPDIoU loss function, can achieve efficient and accurate tracking of packages, improve the accuracy of the logistics package statistics method and the work efficiency of package statistics, reduce logistics transportation costs, and thus improve customer service quality satisfaction. BRIEF DESCRIPTION OF THE DRAWINGS

[0016] Figure 1A flow chart of a logistics package tracking and counting method provided by an embodiment of the present invention; Figure 2 A flow chart of obtaining a logistics package image in the logistics package tracking and counting method provided in an embodiment of the present invention; Figure 3 A flowchart of constructing a deep learning network model in the logistics package tracking and counting method provided in an embodiment of the present invention; Figure 4 A flow chart of adding FC attention mechanism to the logistics package tracking and counting method provided by an embodiment of the present invention; Figure 5 A flowchart of training a logistics package deep learning model in the logistics package tracking and counting method provided in an embodiment of the present invention; Figure 6 A flow chart of logistics package identification and statistics in the logistics package tracking and counting method provided in an embodiment of the present invention; Figure 7 A schematic diagram of the structure of a logistics package tracking and counting device provided by an embodiment of the present invention; Figure 8 Another structural schematic diagram of a logistics package tracking and counting device provided by an embodiment of the present invention; Fig. 9 A schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. DETAILED DESCRIPTION

[0017] The present invention provides a logistics package tracking and counting method, which includes: acquiring a logistics package image, annotating the image, and constructing a logistics package data set; constructing a deep learning network model and introducing an FC attention mechanism; inputting the logistics package data set into a logistics package deep learning model based on YOLOV11 for training, constructing a loss function, and obtaining a logistics package recognition model; inputting the image of the logistics package into the trained recognition model, identifying and locating the logistics package in the image, and then counting the number of the logistics packages according to the image.

[0018] The logistics package tracking and counting method of the present invention, by constructing an intelligent counting system that integrates the YOLOV11 model, adds the FC attention mechanism and the MPDIoU loss function, can achieve efficient and accurate tracking of packages, improve the accuracy of the logistics package statistics method and the work efficiency of package statistics, reduce logistics transportation costs, and thus improve customer service quality satisfaction.

[0019] The terms "first", "second", "third", "fourth", etc. (if any) in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments described herein can be implemented in an order other than that illustrated or described herein. In addition, the terms "including" or "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0020] For ease of understanding, the specific process of the embodiment of the present invention is described below. Figure 1 , an embodiment of the present invention provides a logistics package tracking and counting method, the method comprising: 101. Obtain a logistics package image, annotate the image, and construct a logistics package dataset; 102. Build a deep learning network model and introduce the FC attention mechanism; 103. Input the logistics package data set into the logistics package deep learning model based on YOLOV11 for training, construct a loss function, and obtain a logistics package recognition model; 104. Input the image of the logistics package into the trained recognition model, identify and locate the logistics package in the image, and then count the number of the logistics packages based on the image.

[0021] The logistics package tracking and counting method of the present invention, by constructing an intelligent counting system that integrates the YOLOV11 model, adds the FC attention mechanism and the MPDIoU loss function, can achieve efficient and accurate tracking of packages, improve the accuracy of the logistics package statistics method and the work efficiency of package statistics, reduce logistics transportation costs, and thus improve customer service quality satisfaction.

[0022] See also Figure 2 The second embodiment of the logistics package tracking and counting method in the embodiment of the present invention is based on the above embodiment, and the steps of obtaining a logistics package image, annotating the image, and constructing a logistics package data set include: 201. Install cameras at the starting point, midpoint and end point of the logistics package assembly line; In some embodiments, cameras are installed at the starting point, midpoint and end point of the logistics package assembly line to obtain images of the package at different stages, and the parameters of the cameras at the starting point, midpoint and end point of the assembly line, such as resolution, frame rate, etc., are set to meet the requirements of image quality and data volume.

[0023] 202. Collect images of the logistics package at the starting point, the middle point and the end point of the logistics package assembly line; In some embodiments, image samples monitored by cameras at the starting point, midpoint, and end point of the pipeline are collected regularly, and the images are adjusted in terms of brightness, contrast, color, etc. to improve image quality and readability.

[0024] 203. Acquire the image of the logistics package, clean the image, and delete low-quality images; In some embodiments, according to the input requirements of the model, the image is cropped and scaled to ensure that the image size is consistent; the image is denoised using an image denoising algorithm to remove noise interference in the image and improve the image quality.

[0025] 204. Label the logistics packages in the image, label the location and type of the logistics packages on the image, and each image corresponds to a txt file containing labeling information; 205. All the annotated images and the corresponding annotation files form a logistics package dataset.

[0026] In some embodiments, an image annotation tool LabelImg may be used to annotate the package in the image.

[0027] The high-quality image samples and precise annotations collected by the embodiments of the present invention can provide more reliable training data for the machine learning model, which helps to improve the accuracy and generalization ability of the model; by processing and annotating images under different conditions, the model can adapt to more diverse actual situations; automated image recognition and processing can reduce the need for manual intervention, thereby reducing labor costs and other related expenses; sufficient annotated data can speed up the training and testing of the model.

[0028] See also Figure 3 In the third embodiment of the logistics package tracking and counting method in the embodiment of the present invention, the deep learning network model is constructed and the FC attention mechanism is added, including: 301. Build a deep learning network model; 302. Use the YOLOV11 model as the deep learning network model; 303. Adding FC attention mechanism to the deep learning network model; Among them, the input of the backbone network of the deep learning network model is the preprocessed image, and the output is the feature map; the input of the neck network of the deep learning network model is the feature map, and the output is the adaptive feature fusion map.

[0029] The YOLOV11 model of the embodiment of the present invention adopts an improved backbone and neck architecture to enhance the feature extraction capability to achieve more accurate target detection and complex task performance; optimized for efficiency and speed: the sophisticated architecture design and optimized training process provide faster processing speed while maintaining the best balance between accuracy and performance, which is particularly suitable for the high-intensity and high-density logistics and parcel transportation industry.

[0030] See also Figure 4 In the fourth embodiment of the logistics package tracking and counting method in the embodiment of the present invention, the FC attention mechanism is added, including: 401. The FC attention mechanism is placed on the enhanced feature extraction network, and the attention mechanism is added to the three effective feature layers extracted by the backbone network; 402. Add FC attention mechanism to the upsampling results in the pyramid network.

[0031] The embodiment of the present invention introduces the FC attention mechanism, which is plug-and-play and can be easily integrated into the existing deep learning model to enhance the feature extraction capability of the model. Through this attention mechanism, the network can more accurately identify and track packages on the assembly line of the logistics transportation center, thereby improving the accuracy and efficiency of package tracking.

[0032] See also Figure 5 , the fifth embodiment of the logistics package tracking and counting method in the embodiment of the present invention, the logistics package data set is input into the logistics package deep learning model based on YOLOV11 for training, the loss function is constructed, and the logistics package recognition model is obtained, including: 501. Divide the logistics package data set into a training set, a validation set and a test set; 502. Input the training set to train the YOLOV11-based logistics package deep learning model; 503. Input the verification set to adjust the training parameters of the deep learning model; 504. Input the test set to test the generalization ability of the deep learning model; 505. Construct a loss function of the deep learning model and update the model to obtain a logistics package recognition model based on YOLOV11.

[0033] In some embodiments, the loss function of the YOLOV11-based logistics package identification model is an MPDIoU loss function.

[0034] In some embodiments, the MPDIoU loss function is integrated into the YOLOV11 model to replace the traditional bounding box regression loss function. During the training process of the model, the MPDIoU loss function is used as the optimization target to update the parameters of the model. When training the YOLOV11 model, the MPDIoU loss function is used to optimize the model and observe the performance of the MPDIoU loss function during the training process, such as the change in loss value, convergence speed, etc.

[0035] In some embodiments, a validation set is used to evaluate the YOLOV11 model containing the MPDIoU loss function, and the detection accuracy, recall rate, F1 value and other indicators of the model are calculated. The impact of the MPDIoU loss function on the model performance is analyzed, such as the improvement of the bounding box regression accuracy, the impact on the model complexity and computational complexity, etc.

[0036] In some embodiments, model iteration can also be performed. During the training process, the model parameters are saved every certain number of training rounds to observe the training process of the model, such as the change in loss value, improvement in accuracy, etc. According to the performance of the model on the validation set, the model is adjusted and optimized to improve the performance of the model.

[0037] The YOLOV11 model of the embodiment of the present invention can adapt to packages of various sizes, shapes and types after proper training, and can maintain a high detection accuracy, thereby enhancing the adaptability of the system to different environmental conditions; the YOLOV11 model can accurately locate and classify packages in complex backgrounds, reduce the false detection rate and missed detection rate, and improve the accuracy of logistics operations.

[0038] See also Figure 6 In a sixth embodiment of the method for tracking and counting logistics packages in the embodiments of the present invention, the image of the logistics package is input into the trained recognition model, the logistics package in the image is recognized and located, and then the number of the logistics packages is counted according to the image, including: 601. Input the image of the logistics package into the logistics package recognition model based on YOLOV11; 602. The YOLOV11-based logistics package recognition model recognizes the logistics package in the image; 603. Positioning the logistics package; 604. Count the number of the logistics packages according to the logistics package images.

[0039] The embodiments of the present invention are very suitable for applications in logistics scenarios that require real-time processing of large amounts of image data, especially logistics package classification and counting; the model can help achieve a high degree of automation in the real-time processing of packages, from automated package detection, counting to tracking, reducing the need for manual intervention and improving the efficiency and reliability of the entire logistics system; accurate package tracking and management can reduce the occurrence of loss or delays, improve customer satisfaction, and help logistics service providers provide better services.

[0040] The above describes the method for tracking and counting logistics packages in the embodiment of the present invention. The following describes the device for tracking and counting logistics packages in the embodiment of the present invention. Figure 7 In one embodiment of the present invention, a device for tracking and counting logistics packages includes: The logistics package image acquisition module 701 is used to acquire logistics package images, annotate the images, and construct a logistics package dataset; A deep learning network model building module 702 is used to build a deep learning network model and introduce an FC attention mechanism; The logistics package recognition module 703 is used to input the logistics package data set into the logistics package deep learning model based on YOLOV11 for training, construct a loss function, and obtain a logistics package recognition model; The tracking and statistics module 704 is used to input the image of the logistics package into the trained recognition model, identify and locate the logistics package in the image, and count the number of the logistics packages based on the image.

[0041] The logistics package tracking and counting device of the present invention, by constructing an intelligent counting system that integrates the YOLOV11 model, adds the FC attention mechanism and the MPDIoU loss function, can achieve efficient and accurate tracking of packages, improve the accuracy of the logistics package statistics method and the work efficiency of package statistics, reduce logistics transportation costs, and thus improve customer service quality satisfaction.

[0042] See also Figure 8 Another embodiment of the logistics package tracking and counting device in the embodiment of the present invention includes: In some embodiments, the logistics package image acquisition module 701 is used to acquire logistics package images, annotate the images, and construct a logistics package dataset; In some embodiments, the module 701 for acquiring logistics package images includes: The camera installation unit 7011 is used to set cameras at the starting point, midpoint and end point of the logistics package assembly line; In some embodiments, cameras are installed at the starting point, midpoint and end point of the logistics package assembly line to obtain images of the package at different stages, and the parameters of the cameras at the starting point, midpoint and end point of the assembly line, such as resolution, frame rate, etc., are set to meet the requirements of image quality and data volume.

[0043] A collection unit 7012 is used to collect images of the logistics package at the starting point, the middle point and the end point of the logistics package assembly line; In some embodiments, image samples monitored by cameras at the starting point, midpoint, and end point of the pipeline are collected regularly, and the images are adjusted in terms of brightness, contrast, color, etc. to improve image quality and readability.

[0044] An acquisition unit 7013 is used to acquire the image of the logistics package, clean the image, and delete low-quality images; In some embodiments, according to the input requirements of the model, the image is cropped and scaled to ensure that the image size is consistent; the image is denoised using an image denoising algorithm to remove noise interference in the image and improve the image quality.

[0045] The labeling unit 7014 is used to label the logistics packages in the image, label the location and type of the logistics packages on the image, and each image corresponds to a txt file containing labeling information; The composition unit 7015 is used to combine all the annotated images and the corresponding annotation files into a logistics package dataset.

[0046] In some embodiments, an image annotation tool LabelImg may be used to annotate the package in the image.

[0047] The high-quality image samples and precise annotations collected by the embodiments of the present invention can provide more reliable training data for the machine learning model, which helps to improve the accuracy and generalization ability of the model; by processing and annotating images under different conditions, the model can adapt to more diverse actual situations; automated image recognition and processing can reduce the need for manual intervention, thereby reducing labor costs and other related expenses; sufficient annotated data can speed up the training and testing of the model.

[0048] In some embodiments, the deep learning network model building module 702 is used to build a deep learning network model and introduce the FC attention mechanism; In some embodiments, the deep learning network model building module 702 includes: A construction unit 7021, used for constructing a deep learning network model; YOLOV11 model unit 7022, used for using the YOLOV11 model as the deep learning network model; An adding unit 7023 is used to add a FC attention mechanism to the deep learning network model; Among them, the input of the backbone network of the deep learning network model is the preprocessed image, and the output is the feature map; the input of the neck network of the deep learning network model is the feature map, and the output is the adaptive feature fusion map.

[0049] The YOLOV11 model of the embodiment of the present invention adopts an improved backbone and neck architecture to enhance the feature extraction capability to achieve more accurate target detection and complex task performance; optimized for efficiency and speed: the sophisticated architecture design and optimized training process provide faster processing speed while maintaining the best balance between accuracy and performance, which is particularly suitable for the high-intensity and high-density logistics and parcel transportation industry.

[0050] In some embodiments, the logistics package identification module 703 is used to input the logistics package data set into the logistics package deep learning model based on YOLOV11 for training, construct a loss function, and obtain a logistics package identification model; In some embodiments, the logistics package identification module 703 includes: A division unit 7031 is used to divide the logistics package data set into a training set, a validation set and a test set; The first input unit 7032 is used to input the training set to train the YOLOV11-based logistics package deep learning model; The second input unit 7033 is used to input the verification set to adjust the training parameters of the deep learning model; The third input unit 7034 is used to input the test set to test the generalization ability of the deep learning model; The updating unit 7035 is used to construct a loss function of the deep learning model and update the model to obtain a logistics package recognition model based on YOLOV11.

[0051] In some embodiments, the loss function of the YOLOV11-based logistics package identification model is an MPDIoU loss function.

[0052] In some embodiments, the MPDIoU loss function is integrated into the YOLOV11 model to replace the traditional bounding box regression loss function. During the training process of the model, the MPDIoU loss function is used as the optimization target to update the parameters of the model. When training the YOLOV11 model, the MPDIoU loss function is used to optimize the model and observe the performance of the MPDIoU loss function during the training process, such as the change in loss value, convergence speed, etc.

[0053] In some embodiments, a validation set is used to evaluate the YOLOV11 model containing the MPDIoU loss function, and the detection accuracy, recall rate, F1 value and other indicators of the model are calculated. The impact of the MPDIoU loss function on the model performance is analyzed, such as the improvement of the bounding box regression accuracy, the impact on the model complexity and computational complexity, etc.

[0054] In some embodiments, model iteration can also be performed. During the training process, the model parameters are saved every certain number of training rounds to observe the training process of the model, such as the change in loss value, improvement in accuracy, etc. According to the performance of the model on the validation set, the model is adjusted and optimized to improve the performance of the model.

[0055] The YOLOV11 model of the embodiment of the present invention can adapt to packages of various sizes, shapes and types after proper training, and can maintain a high detection accuracy, thereby enhancing the adaptability of the system to different environmental conditions; the YOLOV11 model can accurately locate and classify packages in complex backgrounds, reduce the false detection rate and missed detection rate, and improve the accuracy of logistics operations.

[0056] In some embodiments, the tracking statistics module 704 is used to input the image of the logistics package into the trained recognition model, identify and locate the logistics package in the image, and count the number of the logistics packages based on the image.

[0057] In some embodiments, the tracking statistics module 704 includes: The fourth input unit 7041 is used to input the image of the logistics package into the logistics package recognition model based on YOLOV11; An identification unit 7042 is used for the logistics package identification model based on YOLOV11 to identify the logistics package in the image; A positioning unit 7043, used for positioning the logistics package; The statistical unit 7044 is used to count the number of the logistics packages according to the logistics package image.

[0058] The embodiments of the present invention are very suitable for applications in logistics scenarios that require real-time processing of large amounts of image data, especially logistics package classification and counting; the model can help achieve a high degree of automation in the real-time processing of packages, from automated package detection, counting to tracking, reducing the need for manual intervention and improving the efficiency and reliability of the entire logistics system; accurate package tracking and management can reduce the occurrence of loss or delays, improve customer satisfaction, and help logistics service providers provide better services.

[0059] above Figure 7 and Figure 8The logistics package tracking and counting method in the embodiment of the present invention is described in detail from the perspective of modular functional entities, and the electronic device in the embodiment of the present invention is described in detail from the perspective of hardware processing.

[0060] Fig. 9 8 is a schematic diagram of the structure of an electronic device provided by an embodiment of the present invention. The electronic device 800 may have relatively large differences due to different configurations or performances, and may include one or more processors (central processing units, CPU) 810 (for example, one or more processors) and a memory 820, and one or more storage media 830 (for example, one or more mass storage devices) storing application programs 833 or data 832. Among them, the memory 820 and the storage medium 830 may be temporary storage or permanent storage. The program stored in the storage medium 830 may include one or more upper modules (not shown in the figure) for vehicle dispatching at a distribution center, and each module may include a series of instruction operations in the electronic device 800. Furthermore, the processor 810 may be configured to communicate with the storage medium 830 to execute a series of instruction operations in the storage medium 830 on the electronic device 800.

[0061] The electronic device 800 may also include one or more power supplies 840, one or more wired or wireless network interfaces 850, one or more input and output interfaces 850, and / or one or more operating systems 831, such as Windows Serve, Mac OS X, Unix, Linux, FreeBSD, etc. It will be appreciated by those skilled in the art that Fig. 9 The structure of the electronic device shown does not constitute a limitation on the electronic device, and may include more or less components than shown in the figure, or combine some components, or arrange the components differently.

[0062] The present invention also provides a computer-readable storage medium, which may be a non-volatile computer-readable storage medium or a volatile computer-readable storage medium. Instructions are stored in the computer-readable storage medium. When the instructions are executed on a computer, the computer executes the steps of the distribution center vehicle dispatching method.

[0063] Those skilled in the art can clearly understand that, for the convenience and brevity of description, the specific working process of the above-described system, device, or unit can refer to the corresponding process in the aforementioned method embodiment and will not be repeated here.

[0064] If the integrated unit is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention is essentially or the part that contributes to the prior art or the whole or part of the technical solution can be embodied in the form of a software product. The computer software product is stored in a storage medium, including several instructions to enable a computer device (which can be a personal computer, server, or network device, etc.) to perform all or part of the steps of the method described in each embodiment of the present invention. The aforementioned storage medium includes: U disk, mobile hard disk, read-only memory (ROM), random access memory (RAM), disk or optical disk and other media that can store program code.

[0065] As described above, the above embodiments are only used to illustrate the technical solutions of the present invention, rather than to limit the same. Although the present invention has been described in detail with reference to the aforementioned embodiments, those skilled in the art should understand that the technical solutions described in the aforementioned embodiments may still be modified, or some of the technical features thereof may be replaced by equivalents. However, these modifications or replacements do not deviate the essence of the corresponding technical solutions from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. A logistics package tracking and counting method, characterized in that: The method comprises: Obtaining logistics package images, annotating the images, and constructing a logistics package dataset; Build a deep learning network model and introduce the FC attention mechanism; The logistics package data set is input into the logistics package deep learning model based on YOLOV11 for training, and the loss function is constructed to obtain the logistics package recognition model; The image of the logistics package is input into the trained recognition model to identify and locate the logistics package in the image, and then the number of the logistics package is counted according to the image.

2. The logistics package tracking and counting method according to claim 1, characterized in that: The obtaining of logistics package images, labeling of the images, and construction of logistics package datasets include: Set up cameras at the starting point, midpoint, and end point of the logistics package assembly line; Collecting images of the logistics package at the starting point, the middle point and the end point of the logistics package assembly line; Acquire the image of the logistics package, clean the image, and delete low-quality images; Label the logistics packages in the image, mark the location and type of the logistics packages on the image, and each image corresponds to a txt file containing the labeling information; All annotated images and corresponding annotation files form a logistics package dataset.

3. The logistics package tracking and counting method according to claim 2, characterized in that: The deep learning network model is constructed and the FC attention mechanism is added, including: Build deep learning network models; Use the YOLOV11 model as the deep learning network model; Adding FC attention mechanism to the deep learning network model; Among them, the input of the backbone network of the deep learning network model is the preprocessed image, and the output is the feature map; the input of the neck network of the deep learning network model is the feature map, and the output is the adaptive feature fusion map.

4. The logistics package tracking and counting method according to claim 4 is characterized in that: The FC attention mechanism is added, including: The FC attention mechanism is placed on the enhanced feature extraction network, and the attention mechanism is added to the three effective feature layers extracted by the backbone network; Add FC attention mechanism to the upsampling results in the pyramid network.

5. The logistics package tracking and counting method according to claim 3 is characterized in that: The logistics package data set is input into the logistics package deep learning model based on YOLOV11 for training, a loss function is constructed, and a logistics package recognition model is obtained, including: Divide the logistics package dataset into training set, validation set and test set; Input the training set to train the YOLOV11-based logistics package deep learning model; Input the validation set to adjust the training parameters of the deep learning model; Input the test set to test the generalization ability of the deep learning model; Construct a loss function of the deep learning model and update the model to obtain a logistics package recognition model based on YOLOV11.

6. The logistics package tracking and counting method according to claim 5, characterized in that: The loss function of the YOLOV11-based logistics package identification model is the MPDIoU loss function.

7. The logistics package tracking and counting method according to claim 5, characterized in that: The step of inputting the image of the logistics package into the trained recognition model, identifying and locating the logistics package in the image, and then counting the number of the logistics packages according to the image includes: Input the image of the logistics package into the logistics package recognition model based on YOLOV11; The YOLOV11-based logistics package recognition model recognizes the logistics package in the image; Positioning the logistics package; The number of the logistics packages is counted according to the logistics package image.

8. A logistics package tracking and counting device, characterized in that: include: A logistics package image acquisition module is used to acquire logistics package images, annotate the images, and construct a logistics package dataset; Build a deep learning network model module, which is used to build a deep learning network model and introduce the FC attention mechanism; The logistics package recognition module is used to input the logistics package data set into the logistics package deep learning model based on YOLOV11 for training, construct a loss function, and obtain a logistics package recognition model; The tracking and statistics module is used to input the image of the logistics package into the trained recognition model, identify and locate the logistics package in the image, and count the number of the logistics package according to the image.

9. A computer device, characterized in that: The computer device includes a memory and at least one processor, wherein instructions are stored in the memory; The at least one processor calls the instructions in the memory to cause the computer device to execute the logistics package tracking and counting method as described in any one of claims 1-7.

10. A computer-readable storage medium, characterized in that: A computer program is stored thereon, and the computer program is loaded by a processor to execute the logistics package tracking and counting method described in any one of claims 1-7.

Images