Image data processing method and system for container identification recognition

Abstract

The present invention discloses an image data processing method and system for container identification recognition. The method includes: acquiring a container image of a target container with a visible identification; determining an identification character region in the container image based on an image segmentation algorithm; determining a correctly arranged image corresponding to the identification character region based on an arrangement correction algorithm according to the identification character region; and performing character recognition on the correctly arranged image to obtain identification information corresponding to the target container. It can be seen that the present invention can improve the recognition accuracy and efficiency of container identification in complex environments, reduce the correction cost caused by recognition errors, and improve the working efficiency of overall cargo handling.

Description

Technical Field

[0001] The present invention relates to the technical field of data processing, and in particular to an image data processing method and system for container identification. Background Art

[0002] In some specific transportation scenarios, such as logistics cargo transportation or port cargo transportation, it is necessary to identify the container's logo. It is a popular practice to obtain the container's parameter information by performing image recognition on the visual logo on the outside of the container. For example, the image information of the container is obtained by a camera and the printed information of the box type or box number is recognized, which can improve the efficiency of cargo handling. However, most of the existing identification recognition methods do not take into account the problem that the character arrangement of the visual logo may be abnormal. Generally, they can only process visual logos that are normally arranged horizontally. Therefore, for identification recognition scenarios in specific complex environments where the arrangement is abnormal, the recognition efficiency is low and the recognition accuracy is poor. It can be seen that the existing technology has defects that need to be solved urgently. Summary of the invention

[0003] The technical problem to be solved by the present invention is to provide an image data processing method and system for container identification, which can improve the recognition accuracy and efficiency of container identification in complex environments, reduce the correction costs caused by recognition errors, and improve the overall work efficiency of cargo processing.

[0004] In order to solve the above technical problems, the first aspect of the present invention discloses an image data processing method for container identification, the method comprising:

[0005] Obtain a container image with a visual indicator of the presence of the target container;

[0006] Determine the identification character area in the container image based on an image segmentation algorithm;

[0007] According to the identification character area, based on an arrangement correction algorithm, determining a correct arrangement image corresponding to the identification character area;

[0008] Character recognition is performed on the correctly arranged image to obtain identification information corresponding to the target container.

[0009] As an optional implementation, in the first aspect of the present invention, determining the identification character region in the container image based on an image segmentation algorithm includes:

[0010] The container image is input into a trained image segmentation neural network to obtain an output polygonal identification character area; the image segmentation neural network is based on the YOLOv8 algorithm architecture.

[0011] As an alternative embodiment, in the first aspect of the present invention, determining the correct arrangement image corresponding to the identification character region based on the arrangement correction algorithm according to the identification character region includes:

[0012] Calculate the minimum rectangle of all polygon vertices including the identification character region to obtain an identification character rectangle;

[0013] Determine the arrangement mode information corresponding to the plurality of identification character images according to the lengths of different sides of the identification character rectangle;

[0014] Determine the correct arrangement image corresponding to the identification character region according to the arrangement mode information.

[0015] As an alternative embodiment, in the first aspect of the present invention, determining the arrangement mode information corresponding to the identification character region according to the lengths of different sides of the identification character rectangle includes:

[0016] Judge whether the length of the horizontal side of the identification character rectangle is greater than the length of the vertical side to obtain a first judgment result;

[0017] If the first judgment result is yes, determine that the arrangement mode information of the identification character region is horizontal arrangement;

[0018] If the second judgment result is no, determine that the arrangement mode information of the identification character region is vertical arrangement.

[0019] As an alternative embodiment, in the first aspect of the present invention, determining the correct arrangement image corresponding to the identification character region according to the arrangement mode information includes:

[0020] When the arrangement direction information is horizontal arrangement, perform at least one rotation process with a first angle value on the identification character rectangle to obtain a plurality of first images before and after rotation; the first angle value is 180 degrees;

[0021] Input each of the first images into a trained first character recognition algorithm model to obtain the recognition confidence corresponding to each of the first images;

[0022] Determine the first image with the highest recognition confidence as the correct arrangement image;

[0023] When the arrangement direction information is vertical arrangement, horizontally arrange each identification character image in the identification character rectangle according to the order of vertical arrangement, and perform multiple rotation processes with a second angle value to obtain a plurality of second images; the second angle value is 90 degrees;

[0024] Input each of the second images into the trained second character recognition algorithm model to obtain the recognition confidence corresponding to each of the second images;

[0025] Determine the second image with the highest recognition confidence as the correctly arranged image.

[0026] As an optional implementation manner, in the first aspect of the present invention, the horizontal arrangement of each identification character image in the identification character rectangle based on the order of the vertical arrangement manner, and performing multiple rotation processes with the second angle value to obtain multiple second images, includes:

[0027] Input the identification character rectangle into the trained vertical arrangement manner prediction classifier to obtain the predicted vertical arrangement manner output; the predicted vertical arrangement manner is a single-line vertical arrangement manner, an upward-aligned multi-line vertical arrangement manner, a downward-aligned multi-line vertical arrangement manner, or a centered multi-line vertical arrangement manner;

[0028] Horizontally arrange all the identification character images in the identification character rectangle based on the reading order corresponding to the predicted vertical arrangement manner to obtain an arranged image;

[0029] Perform multiple rotation processes with the second angle value on the arranged image to obtain multiple second images.

[0030] As an optional implementation manner, in the first aspect of the present invention, the character recognition of the correctly arranged image to obtain the identification information corresponding to the target container includes:

[0031] Process the correctly arranged image according to the trained character segmentation algorithm model to obtain multiple character images;

[0032] Based on multiple character recognition models, perform recognition on each of the character images to obtain multiple model recognition results corresponding to each of the character images;

[0033] Based on the voting algorithm, determine the character recognition result corresponding to each of the character images according to the multiple model recognition results corresponding to each of the character images;

[0034] Determine the character recognition results corresponding to all the character images as the identification information corresponding to the target container.

[0035] As an optional implementation manner, in the first aspect of the present invention, the method further includes:

[0036] Based on the preset character type position rule, screen out the characters with incorrect positions in the identification information;

[0037] Determine the correct character corresponding to each mispositioned character according to the preset character confusion correspondence relationship;

[0038] Replace all the mispositioned characters in the identification information with the corresponding correct characters to obtain the corrected identification information.

[0039] A second aspect of the embodiments of the present invention discloses an image data processing system for container identification recognition, and the system includes:

[0040] An acquisition module, configured to acquire a container image with a visual identification of a target container;

[0041] A segmentation module, configured to determine an identification character region in the container image based on an image segmentation algorithm;

[0042] A determination module, configured to determine a correct arrangement image corresponding to the identification character region based on an arrangement correction algorithm according to the identification character region;

[0043] An identification module, configured to perform character recognition on the correct arrangement image to obtain identification information corresponding to the target container.

[0044] As an optional implementation manner, in the second aspect of the present invention, the specific manner in which the segmentation module determines the identification character region in the container image based on the image segmentation algorithm includes:

[0045] Input the container image into a trained image segmentation neural network to obtain an output identification character region in the form of a polygon; the image segmentation neural network is based on the YOLOv8 algorithm architecture.

[0046] As an optional implementation manner, in the second aspect of the present invention, the specific manner in which the determination module determines a correct arrangement image corresponding to the identification character region based on the arrangement correction algorithm according to the identification character region includes:

[0047] Calculate the minimum rectangle including all the polygon vertices of the identification character region to obtain an identification character rectangle;

[0048] Determine arrangement manner information corresponding to the plurality of identification character images according to the lengths of different sides of the identification character rectangle;

[0049] Determine a correct arrangement image corresponding to the identification character region according to the arrangement manner information.

[0050] As an optional implementation manner, in the second aspect of the present invention, the specific manner in which the determination module determines arrangement manner information corresponding to the identification character region according to the lengths of different sides of the identification character rectangle includes:

[0051] Judge whether the horizontal side length of the identified character rectangle is greater than the vertical side length to obtain a first judgment result;

[0052] If the first judgment result is yes, determine that the arrangement mode information of the identified character area is horizontal arrangement;

[0053] If the second judgment result is no, determine that the arrangement mode information of the identified character area is vertical arrangement.

[0054] As an optional implementation manner, in the second aspect of the present invention, the specific manner in which the determination module determines the correct arrangement image corresponding to the identified character area according to the arrangement mode information includes:

[0055] When the arrangement direction information is horizontal arrangement, perform at least one rotation process with a first angle value on the identified character rectangle to obtain multiple first images before and after rotation; the first angle value is 180 degrees;

[0056] Input each of the first images into a trained first character recognition algorithm model to obtain the recognition confidence corresponding to each of the first images;

[0057] Determine the first image with the highest recognition confidence as the correct arrangement image;

[0058] When the arrangement direction information is vertical arrangement, horizontally arrange each identified character image in the identified character rectangle according to the order of the vertical arrangement method, and perform multiple rotation processes with a second angle value to obtain multiple second images; the second angle value is 90 degrees;

[0059] Input each of the second images into a trained second character recognition algorithm model to obtain the recognition confidence corresponding to each of the second images;

[0060] Determine the second image with the highest recognition confidence as the correct arrangement image.

[0061] As an optional implementation manner, in the second aspect of the present invention, the specific manner in which the determination module horizontally arranges each identified character image in the identified character rectangle according to the order of the vertical arrangement method and performs multiple rotation processes with a second angle value to obtain multiple second images includes:

[0062] Input the identified character rectangle into a trained vertical arrangement method prediction classifier to obtain the predicted vertical arrangement method output; the predicted vertical arrangement method is a single-line vertical arrangement method, an upward-aligned multi-line vertical arrangement method, a downward-aligned multi-line vertical arrangement method, or a centered-aligned multi-line vertical arrangement method;

[0063] Horizontally arrange all the identification character images in the identification character rectangle according to the reading order corresponding to the predicted vertical arrangement method to obtain an arranged image;

[0064] Perform rotation processing on the arranged image at a second angle value multiple times to obtain multiple second images.

[0065] As an optional implementation manner, in the second aspect of the present invention, the specific manner in which the recognition module performs character recognition on the correctly arranged image to obtain the identification information corresponding to the target container includes:

[0066] Process the correctly arranged image according to the trained character segmentation algorithm model to obtain multiple character images;

[0067] Based on multiple character recognition models, perform recognition on each of the character images to obtain multiple model recognition results corresponding to each of the character images;

[0068] Based on the voting algorithm, determine the character recognition result corresponding to each of the character images according to the multiple model recognition results corresponding to each of the character images;

[0069] Determine the character recognition results corresponding to all the character images as the identification information corresponding to the target container.

[0070] As an optional implementation manner, in the second aspect of the present invention, the system is further configured to perform the following steps:

[0071] Based on the preset character type position rule, screen out the mispositioned characters in the identification information;

[0072] According to the preset character confusion correspondence, determine the correct character corresponding to each of the mispositioned characters;

[0073] Replace all the mispositioned characters in the identification information with the corresponding correct characters to obtain the corrected identification information.

[0074] The third aspect of the present invention discloses another image data processing system for container identification, and the system includes:

[0075] A memory storing executable program code;

[0076] A processor coupled to the memory;

[0077] The processor calls the executable program code stored in the memory and executes some or all of the steps in the image data processing method for container identification disclosed in the first aspect of the present invention.

[0078] The fourth aspect of the present invention discloses a computer storage medium storing computer instructions, which are used to execute some or all of the steps in the image data processing method for container identification mark recognition disclosed in the first aspect of the present invention when being called.

[0079] Compared with the prior art, the embodiments of the present invention have the following beneficial effects:

[0080] The present invention can determine the identification character image in the container image according to the image segmentation algorithm, and then determine the corresponding correct arrangement image based on the arrangement correction algorithm, so as to perform character recognition on the correct arrangement image to obtain accurate identification information, thereby being able to improve the recognition accuracy and efficiency of container identification marks in complex environments, reduce the correction cost caused by recognition errors, and improve the working efficiency of overall cargo handling. BRIEF DESCRIPTION OF THE DRAWINGS

[0081] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0082] Figure 1 FIG. is a schematic flowchart of an image data processing method for container identification mark recognition disclosed in an embodiment of the present invention.

[0083] Figure 2 FIG. is a schematic structural diagram of an image data processing system for container identification mark recognition disclosed in an embodiment of the present invention.

[0084] Figure 3 FIG. is a schematic structural diagram of another image data processing system for container identification mark recognition disclosed in an embodiment of the present invention. DETAILED DESCRIPTION OF THE EMBODIMENTS

[0085] In order to enable those skilled in the art to better understand the solutions of the present invention, the following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are only some embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts belong to the scope of protection of the present invention.

[0086] In the description, claims and the above-mentioned drawings of the present invention, terms such as "first", "second", etc. are used to distinguish different objects, rather than to describe a specific order. In addition, the terms "comprising" and "having" and any variations thereof are intended to cover non-exclusive inclusion. For example, a process, method, device, product or equipment that includes a series of steps or units is not limited to the listed steps or units, but optionally further includes steps or units not listed, or optionally further includes other steps or units inherent to these processes, methods, products or equipment.

[0087] Referring to "embodiments" herein means that the specific features, structures or characteristics described in connection with the embodiments can be included in at least one embodiment of the present invention. The phrase appears in various places in the specification and does not necessarily refer to the same embodiment, nor is it an independent or alternative embodiment mutually exclusive with other embodiments. Those skilled in the art will explicitly and implicitly understand that the embodiments described herein can be combined with other embodiments.

[0088] The present invention discloses an image data processing method and system for container identification recognition, which can determine the identification character image in the container image according to the image segmentation algorithm, and then determine the corresponding correctly arranged image based on the arrangement correction algorithm, so as to perform character recognition on the correctly arranged image to obtain accurate identification information, thereby improving the recognition accuracy and efficiency of container identification in complex environments, reducing the correction cost caused by recognition errors, and improving the overall work efficiency of goods handling. The following will be described in detail respectively.

[0089] Embodiment 1

[0090] Please refer to Figure 1 , Figure 1 which is a schematic flowchart of an image data processing method for container identification recognition disclosed in an embodiment of the present invention. Among them, Figure 1 the described image data processing method for container identification recognition can be applied to a data processing system / data processing device / data processing server (wherein, the server includes a local processing server or a cloud processing server). As Figure 1 shown, the image data processing method for container identification recognition can include the following operations:

[0091] 101. Obtain a container image of the target container with a visible identification.

[0092] Optionally, the target container can be a shipping container.

[0093] Optionally, the identification information can be text information such as the type and / or serial number of the shipping container.

[0094] 102. Based on the image segmentation algorithm, determine the identification character region in the container image.

[0095] 103. Based on the identification character region and the arrangement correction algorithm, determine the correctly arranged image corresponding to the identification character region.

[0096] 104. Perform character recognition on the correctly arranged image to obtain the identification information corresponding to the target container.

[0097] It can be seen that the above-mentioned invention embodiments can determine the identification character image in the container image according to the image segmentation algorithm, and then determine the corresponding correctly arranged image based on the arrangement correction algorithm, so as to perform character recognition on the correctly arranged image to obtain accurate identification information, thereby improving the recognition accuracy and efficiency of container identification in complex environments, reducing the correction cost caused by recognition errors, and improving the overall work efficiency of goods handling.

[0098] As an optional embodiment, in the above steps, based on the image segmentation algorithm, determining the identification character region in the container image includes:

[0099] Input the container image into the trained image segmentation neural network to obtain the output identification character region in the shape of a polygon.

[0100] Optionally, the image segmentation neural network is based on the YOLOv8 algorithm architecture.

[0101] In a specific implementation, the image segmentation neural network is a YOLOv8 segmentation model, which consists of three networks: Head, Neck, and Backbone. Among them, the Head network outputs through a feature pyramid structure and a polygon neural network; the Neck network mainly performs dimensionality reduction operations on data through dimensionality reduction (such as pooling, etc.) related algorithms; the Backbone network adopts some classic convolutional neural networks, including but not limited to Darknet and CSPDarknet. These networks usually contain multiple convolutional layers and pooling layers for gradually extracting the features of the image. Its structure can be divided into several residual blocks (ResidualBlocks), and each residual block contains multiple convolutional layers internally, and also contains cross-layer connections, which helps the propagation of gradients and the reuse of features.

[0102] Specifically, if the image segmentation neural network does not find the identification character, it returns empty information. If it finds the identification character, it records the vertices of the polygon where the identification character is located for the next operation. Specifically, the advantage of polygon segmentation and positioning is that it can accurately locate irregularly arranged text information.

[0103] It can be seen that through the above optional embodiments, the identification character area in the container image can be accurately recognized by the trained image segmentation neural network, so as to subsequently determine the corresponding correctly arranged image based on the arrangement correction algorithm, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the correction cost caused by recognition errors, and improving the overall working efficiency of goods handling.

[0104] As an optional embodiment, in the above steps, according to the identification character area, based on the arrangement correction algorithm, determining the correctly arranged image corresponding to the identification character area includes:

[0105] Calculating the minimum rectangle of all polygon vertices including the identification character area to obtain the identification character rectangle;

[0106] According to the lengths of different sides of the identification character rectangle, determining the arrangement mode information corresponding to multiple identification character images;

[0107] According to the arrangement mode information, determining the correctly arranged image corresponding to the identification character area.

[0108] It can be seen that through the above optional embodiments, by calculating the minimum rectangle of all polygon vertices including the identification character area, the corresponding arrangement mode information can be determined based on the side lengths, and then the correctly arranged image can be determined based on the arrangement mode information to obtain a more accurate correctly arranged image, so as to subsequently perform character recognition on the correctly arranged image to obtain accurate identification information, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the correction cost caused by recognition errors, and improving the overall working efficiency of goods handling.

[0109] As an optional embodiment, in the above steps, according to the lengths of different sides of the identification character rectangle, determining the arrangement mode information corresponding to the identification character area includes:

[0110] Judging whether the length of the horizontal side of the identification character rectangle is greater than the length of the vertical side to obtain a first judgment result;

[0111] If the first judgment result is yes, determining the arrangement mode information of the identification character area as horizontal arrangement;

[0112] If the second judgment result is no, determining the arrangement mode information of the identification character area as vertical arrangement.

[0113] It can be seen that through the above optional embodiments, it is possible to determine the arrangement information of multiple identification character images by judging whether the length of the horizontal side of the identification character rectangle is greater than the length of the vertical side, which is convenient for subsequently determining the correct arrangement image based on the arrangement information, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the calibration cost caused by recognition errors, and improving the overall work efficiency of cargo handling.

[0114] As an optional embodiment, in the above steps, determining the correct arrangement image corresponding to the identification character region according to the arrangement information includes:

[0115] When the arrangement direction information is horizontal arrangement, perform at least one rotation process with a first angle value on the identification character rectangle to obtain multiple first images before and after rotation;

[0116] Optionally, the first angle value is 180 degrees;

[0117] Input each first image into the trained first character recognition algorithm model to obtain the recognition confidence corresponding to each first image;

[0118] Determine the first image with the highest recognition confidence as the correct arrangement image;

[0119] When the arrangement direction information is vertical arrangement, horizontally arrange each identification character image in the identification character rectangle based on the order of vertical arrangement, and perform multiple rotation processes with a second angle value to obtain multiple second images;

[0120] Optionally, the second angle value is 90 degrees;

[0121] Input each second image into the trained second character recognition algorithm model to obtain the recognition confidence corresponding to each second image;

[0122] Determine the second image with the highest recognition confidence as the correct arrangement image.

[0123] Optionally, the first character recognition algorithm model or the second character recognition algorithm model can be a lightweight algorithm model, which is not used to specifically recognize characters, but only to confirm the probability that the characters belong to the correct direction. It can be a neural network model with an ABINet architecture. Specifically, AbiNet uses a lightweight backbone network as its main body. Usually, this backbone network will consist of some convolutional layers with relatively shallow depths, such as MobileNetV2, EfficientNet, etc. Such a choice aims to ensure that the network is lightweight while having a certain feature extraction ability.

[0124] Specifically, AbiNet introduces attention modules to enhance the network's feature representation ability. These attention modules usually include channel attention mechanisms or spatial attention mechanisms, which are used to dynamically adjust the feature responses of different channels or different positions in the feature map to increase the weights of important features.

[0125] Specifically, to better fuse multi-scale features, AbiNet introduces feature fusion modules between features at different levels. These modules can be simple upsampling and downsampling operations, or more complex skip connections or pyramid structures, which are used to effectively fuse features of different resolutions.

[0126] Specifically, the detection head of AbiNet usually consists of some convolutional layers and pooling layers, which are used to convert the extracted feature maps into the results of object detection. These detection heads generate the positions and class predictions of the bounding boxes, and usually adopt some techniques to improve the accuracy and stability of detection, such as the anchor box mechanism in YOLOv3 or the Region Proposal Network (RPN) in Faster R-CNN, etc.

[0127] Specifically, similar to other object detection algorithms, the bounding box predictions generated by AbiNet also need to go through a series of post-processing steps to obtain the final detection results. These post-processing steps include removing bounding boxes with low confidence, applying non-maximum suppression (NMS) to remove overlapping bounding boxes, and performing class label matching, etc.

[0128] In the above specific implementation, the size of the smallest rectangle containing all vertices is calculated through the obtained polygon vertices. If the length of the rectangle is greater than the width, it indicates that the text of the box or box number is horizontally arranged. However, since the text may be flipped by 180 degrees, it is necessary to determine whether it is normal horizontal text or flipped text. Therefore, the system uses ABINet to recognize the original text and the text flipped by 180 degrees respectively, calculates their confidence levels, and takes the one with the higher confidence level as the result of the correct arrangement of the text. If the width of the rectangle is greater than the length, it means that the text appears vertically in the original image. There may be three cases here: (1) The text is originally printed vertically. (2) The text is horizontally arranged but rotated 90 degrees clockwise. (3) The text is horizontally arranged but rotated 90 degrees counterclockwise. Therefore, we need to determine which case it is.

[0129] Specifically, in this solution, the original text is first segmented by characters. The segmentation model used here is the Swin Transformer. Then, based on the coordinates of the segmented characters, we rearrange them into a horizontal form to form a new text image. Additionally, the images rotated 90 degrees clockwise and counterclockwise, a total of 3 images, are given to ABINet for judgment, and the one with the higher confidence is taken as the result of the correct arrangement of the text.

[0130] Specifically, Swin Transformer is a deep neural network structure based on the self-attention mechanism, aiming to address the problem that the computational complexity and memory consumption increase sharply as the model scale grows. It adopts a novel hierarchical decomposition idea, reducing the computational and storage requirements of large Transformer models to an acceptable level.

[0131] Specifically, the input image of Swin Transformer is first segmented into a series of fixed-size image patches (or called patches) through an operation similar to a convolutional operation, and then each patch is converted into a lower-dimensional feature vector, usually achieved through a linear transformation (such as a fully connected layer).

[0132] Specifically, the core design of Swin Transformer adopts multiple Stage hierarchical structures, and each Stage contains a group of Transformer blocks. Different from that, Swin Transformer restricts the cross-layer connection of the attention mechanism within the current Stage instead of global connection. This design of local connection reduces the computational and storage complexity.

[0133] Specifically, Swin Transformer introduces a local self-attention mechanism, which divides the input sequence into multiple local regions. The positions within each region only perform self-attention calculations with adjacent positions instead of global positions. This local self-attention mechanism significantly reduces the computational complexity and preserves the model's representation ability to a certain extent.

[0134] Specifically, each Transformer block of Swin Transformer usually contains basic components such as the multi-head self-attention mechanism, feed-forward neural network, and residual connection. These components interact with each other, enabling the model to capture the global dependencies of the input sequence and generate high-quality feature representations.

[0135] Specifically, Swin Transformer can increase the number of model parameters and representation ability by increasing the depth of the stage and the width of the Transformer blocks in each stage, without causing a significant increase in computing and storage. This enables Swin Transformer to be easily applied to various tasks and datasets of different scales.

[0136] It can be seen that through the above optional embodiments, after rotating and arranging the identification character rectangles, the confidence level of the arrangement is determined by identifying based on the character recognition algorithm model, and the arrangement with a higher confidence level is determined as the correct arrangement, so as to obtain a more accurate correct arrangement image, which is convenient for subsequent character recognition of the correct arrangement image to obtain accurate identification information, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the calibration cost caused by recognition errors, and improving the overall work efficiency of goods handling.

[0137] As an optional embodiment, in the above steps, each identification character image in the identification character rectangle is horizontally arranged based on the order of the vertical arrangement method, and rotated multiple times at the second angle value to obtain multiple second images, including:

[0138] Input the identification character rectangle into the trained vertical arrangement method prediction classifier to obtain the predicted vertical arrangement method as the output;

[0139] Optionally, the predicted vertical arrangement method is a single-line vertical arrangement method, an upward-aligned multi-line vertical arrangement method, a downward-aligned multi-line vertical arrangement method, or a centered-aligned multi-line vertical arrangement method;

[0140] Horizontally arrange all the identification character images in the identification character rectangle based on the reading order corresponding to the predicted vertical arrangement method to obtain an arranged image;

[0141] Rotate the arranged image multiple times at the second angle value to obtain multiple second images.

[0142] Specifically, in this embodiment, considering that there may be various situations of different alignment methods for multi-line text in vertical arrangement, a classifier algorithm is introduced to determine its arrangement method in advance, so as to facilitate accurate horizontal arrangement.

[0143] It can be seen that through the above optional embodiments, the vertical arrangement method of the identification characters can be predicted by the trained classifier model, so as to accurately arrange the character images to obtain a more accurate correct arrangement image, which is convenient for subsequent character recognition of the correct arrangement image to obtain accurate identification information, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the calibration cost caused by recognition errors, and improving the overall work efficiency of goods handling.

[0144] As an optional embodiment, in the above steps, character recognition is performed on the correctly arranged image to obtain the identification information corresponding to the target container, including:

[0145] Process the correctly arranged image according to the trained character segmentation algorithm model to obtain multiple character images;

[0146] Based on multiple character recognition models, recognize each character image to obtain multiple model recognition results corresponding to each character image;

[0147] Based on the voting algorithm, determine the character recognition result corresponding to each character image according to the multiple model recognition results corresponding to each character image;

[0148] Determine the character recognition results corresponding to all character images as the identification information corresponding to the target container.

[0149] In the above specific implementation, the correctly arranged text can be segmented character by character through Swin Transformer, and each character can be recognized by using three models, namely trocr, resnet50, and ABINet. Voting is carried out in the way of majority vote to determine the result of character recognition and improve the recognition accuracy.

[0150] Optionally, it is also possible to calculate the weighted calculated value of the recognition confidence corresponding to each model recognition result corresponding to each character image, and determine the model recognition result with the highest weighted calculated value as the character recognition result corresponding to the character image. Specifically, the weighted calculated value is the product of the recognition confidence and the weight, and the weight can be proportional to the historical recognition accuracy of the corresponding character recognition model.

[0151] It can be seen that through the above optional embodiments, it is possible to process the correctly arranged image based on the trained character segmentation algorithm model to obtain multiple character images, then recognize each character image based on multiple character recognition models, and then determine the most accurate character recognition result based on the voting algorithm to calculate the accurate identification information, improve the recognition accuracy and efficiency of container identification in complex environments, reduce the calibration cost caused by recognition errors, and improve the overall work efficiency of cargo handling.

[0152] As an optional embodiment, in the above steps, the method further includes:

[0153] Based on the preset character type position rule, screen out the characters with incorrect positions in the identification information;

[0154] According to the preset character confusion correspondence, determine the correct character corresponding to each character with an incorrect position;

[0155] Replace all mispositioned characters in the identification information with their corresponding correct characters to obtain the corrected identification information.

[0156] Optionally, the preset character type position rule can be set by the operator according to the specific actual situation. For example, according to the naming rule of the container number, since only the first four characters of the container number are English and the rest are all numbers, and for the container type characters, the first, second, and fourth characters are numbers and the third character is English, the character types at specific positions are already determined, and thus a rule can be maintained accordingly. Or, to expand the applicability of the solution of the present invention, a neural network for predicting the character types at different positions in the identification information corresponding to different identification types can be trained for predicting and judging the character type position rule.

[0157] Optionally, the preset character confusion correspondence can be determined by the operator according to the historical records. For example, specific numbers and English are prone to confusion and can be determined as a corresponding relationship. Or, the possible correct characters can be predicted through a further confusion prediction neural network.

[0158] It can be seen that through the above optional embodiments, after screening out the mispositioned characters based on the character type position rule, the correct character corresponding to each mispositioned character can be determined according to the character confusion correspondence to correct and obtain the accurate identification information, improving the recognition accuracy and efficiency of the container identification in a complex environment, reducing the calibration cost caused by recognition errors, and improving the overall working efficiency of the cargo handling.

[0159] Embodiment 2

[0160] Please refer to Figure 2 , Figure 2 which is a schematic structural diagram of an image data processing system for container identification disclosed in an embodiment of the present invention. Among them, Figure 2 the described image data processing system for container identification can be applied to a data processing system / data processing device / data processing server (wherein, the server includes a local processing server or a cloud processing server). As Figure 2 shown, the image data processing system for container identification may include:

[0161] An acquisition module 201, configured to acquire a container image with a visual identification of a target container.

[0162] A segmentation module 202, configured to determine an identification character region in the container image based on an image segmentation algorithm.

[0163] A determination module 203, configured to determine a correct arrangement image corresponding to the identification character region based on an arrangement correction algorithm according to the identification character region.

[0164] An identification module 204, configured to perform character recognition on the correctly arranged image to obtain the identification information corresponding to the target container.

[0165] It can be seen that the above-mentioned invention embodiments can determine the identification character image in the container image according to the image segmentation algorithm, and then determine the corresponding correctly arranged image based on the arrangement correction algorithm, so as to perform character recognition on the correctly arranged image to obtain accurate identification information. Therefore, it can improve the recognition accuracy and efficiency of container identification in complex environments, reduce the correction cost caused by recognition errors, and improve the work efficiency of overall cargo handling.

[0166] As an optional embodiment, the specific manner in which the segmentation module determines the identification character region in the container image based on the image segmentation algorithm includes:

[0167] Input the container image into the trained image segmentation neural network to obtain the output identification character region in the form of a polygon; the image segmentation neural network is based on the YOLOv8 algorithm architecture.

[0168] It can be seen that through the above optional embodiment, the identification character region in the container image can be accurately recognized by the trained image segmentation neural network, so as to subsequently determine the corresponding correctly arranged image based on the arrangement correction algorithm, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the correction cost caused by recognition errors, and improving the work efficiency of overall cargo handling.

[0169] As an optional embodiment, the specific manner in which the determination module determines the correctly arranged image corresponding to the identification character region based on the arrangement correction algorithm according to the identification character region includes:

[0170] Calculate the minimum rectangle of all polygon vertices including the identification character region to obtain the identification character rectangle;

[0171] Determine the arrangement mode information corresponding to multiple identification character images according to the lengths of different sides of the identification character rectangle;

[0172] Determine the correctly arranged image corresponding to the identification character region according to the arrangement mode information.

[0173] It can be seen that through the above optional embodiment, the corresponding arrangement mode information can be determined based on the side lengths by calculating the minimum rectangle of all polygon vertices including the identification character region, and then the correctly arranged image can be determined based on the arrangement mode information to obtain a more accurate correctly arranged image, so as to subsequently perform character recognition on the correctly arranged image to obtain accurate identification information, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the correction cost caused by recognition errors, and improving the work efficiency of overall cargo handling.

[0174] As an alternative embodiment, the specific manner in which the determination module determines the arrangement manner information corresponding to the identification character region according to the lengths of different side lengths of the identification character rectangle includes:

[0175] Judge whether the length of the horizontal side of the identification character rectangle is greater than the length of the vertical side to obtain a first judgment result;

[0176] If the first judgment result is yes, determine that the arrangement manner information of the identification character region is horizontal arrangement;

[0177] If the second judgment result is no, determine that the arrangement manner information of the identification character region is vertical arrangement.

[0178] It can be seen that through the above alternative embodiments, it is possible to determine the arrangement manner information of multiple identification character images by judging whether the length of the horizontal side of the identification character rectangle is greater than the length of the vertical side, which is convenient for subsequently determining the correct arrangement image based on the arrangement manner information, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the calibration cost caused by recognition errors, and improving the overall work efficiency of goods handling.

[0179] As an alternative embodiment, the specific manner in which the determination module determines the correct arrangement image corresponding to the identification character region according to the arrangement manner information includes:

[0180] When the arrangement direction information is horizontal arrangement, perform at least one rotation process of the first angle value on the identification character rectangle to obtain multiple first images before and after rotation; optionally, the first angle value is 180 degrees;

[0181] Input each first image into the trained first character recognition algorithm model to obtain the recognition confidence corresponding to each first image;

[0182] Determine the first image with the highest recognition confidence as the correct arrangement image;

[0183] When the arrangement direction information is vertical arrangement, horizontally arrange each identification character image in the identification character rectangle in the order of vertical arrangement and perform multiple rotation processes of the second angle value to obtain multiple second images; optionally, the second angle value is 90 degrees;

[0184] Input each second image into the trained second character recognition algorithm model to obtain the recognition confidence corresponding to each second image;

[0185] Determine the second image with the highest recognition confidence as the correct arrangement image.

[0186] It can be seen that through the above optional embodiments, after rotating and arranging the identification character rectangles, the character recognition algorithm model can be used for recognition to determine the arrangement with a higher confidence level as the correct arrangement, so as to obtain a more accurate correct arrangement image, which is convenient for subsequent character recognition of the correct arrangement image to obtain accurate identification information, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the calibration cost caused by recognition errors, and improving the overall work efficiency of cargo handling.

[0187] As an optional embodiment, the specific manner in which the determination module horizontally arranges each identification character image in the identification character rectangle based on the order of the vertical arrangement method and performs rotation processing at multiple second angle values to obtain multiple second images includes:

[0188] Input the identification character rectangle into the trained vertical arrangement method prediction classifier to obtain the predicted vertical arrangement method output;

[0189] Optionally, the predicted vertical arrangement method is a single-line vertical arrangement method, an upward-aligned multi-line vertical arrangement method, a downward-aligned multi-line vertical arrangement method, or a centered-aligned multi-line vertical arrangement method;

[0190] Horizontally arrange all the identification character images in the identification character rectangle based on the reading order corresponding to the predicted vertical arrangement method to obtain an arrangement image;

[0191] Perform rotation processing at multiple second angle values on the arrangement image to obtain multiple second images.

[0192] It can be seen that through the above optional embodiments, the vertical arrangement method of identification characters can be predicted by the trained classifier model to accurately arrange the character images to obtain a more accurate correct arrangement image, which is convenient for subsequent character recognition of the correct arrangement image to obtain accurate identification information, assisting in improving the recognition accuracy and efficiency of container identification in complex environments, reducing the calibration cost caused by recognition errors, and improving the overall work efficiency of cargo handling.

[0193] As an optional embodiment, the specific manner in which the recognition module performs character recognition on the correct arrangement image to obtain the identification information corresponding to the target container includes:

[0194] Process the correct arrangement image according to the trained character segmentation algorithm model to obtain multiple character images;

[0195] Based on multiple character recognition models, perform recognition on each character image to obtain multiple model recognition results corresponding to each character image;

[0196] Based on the voting algorithm, determine the character recognition result corresponding to each character image according to the multiple model recognition results corresponding to each character image;

[0197] Determine the identification information corresponding to the target container as the character recognition results corresponding to all character images.

[0198] It can be seen that through the above optional embodiments, it is possible to process the correctly arranged image based on the trained character segmentation algorithm model to obtain multiple character images, then identify each character image based on multiple character recognition models, and then determine the most accurate character recognition result based on the voting algorithm to calculate the accurate identification information, improve the recognition accuracy and efficiency of container identification in complex environments, reduce the calibration cost caused by recognition errors, and improve the overall work efficiency of goods handling.

[0199] As an optional embodiment, the system is further configured to perform the following steps:

[0200] Based on the preset character type position rule, filter out the mispositioned characters in the identification information;

[0201] According to the preset character confusion correspondence, determine the correct character corresponding to each mispositioned character;

[0202] Replace all mispositioned characters in the identification information with the corresponding correct characters to obtain the corrected identification information.

[0203] It can be seen that through the above optional embodiments, it is possible to filter out mispositioned characters based on the character type position rule, and then determine the correct character corresponding to each mispositioned character according to the character confusion correspondence to correct and obtain accurate identification information, improve the recognition accuracy and efficiency of container identification in complex environments, reduce the calibration cost caused by recognition errors, and improve the overall work efficiency of goods handling.

[0204] Embodiment III

[0205] Please refer to Figure 3 , Figure 3 which is another image data processing system for container identification disclosed in the embodiments of the present invention. Figure 3 The described image data processing system for container identification is applied to a data processing system / data processing device / data processing server (wherein, the server includes a local processing server or a cloud processing server). As Figure 3 shown, the image data processing system for container identification may include:

[0206] A memory 301 storing executable program code;

[0207] A processor 302 coupled to the memory 301;

[0208] Among them, the processor 302 calls the executable program code stored in the memory 301 to execute the steps of the image data processing method for container identification described in the first embodiment.

[0209] Embodiment 4

[0210] An embodiment of the present invention discloses a computer-readable storage medium that stores a computer program for electronic data exchange. Among them, the computer program enables a computer to execute the steps of the image data processing method for container identification described in the first embodiment.

[0211] Embodiment 5

[0212] An embodiment of the present invention discloses a computer program product. The computer program product includes a non-transitory computer-readable storage medium storing a computer program, and the computer program is operable to enable a computer to execute the steps of the image data processing method for container identification described in the first embodiment.

[0213] The above describes specific embodiments of this specification, and other embodiments are within the scope of the appended claims. In some cases, the actions or steps recited in the claims may be performed in a different order than in the embodiments and still achieve the desired result. Additionally, the processes depicted in the drawings do not necessarily have to be performed in the specific order or continuous order shown to achieve the desired result. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0214] The systems, devices, modules, or units illustrated in the above embodiments may be specifically implemented by computer chips or entities, or by products with certain functions. A typical implementation device is a computer. Specifically, the computer may be, for example, a personal computer, a laptop computer, a cellular phone, a camera phone, a smartphone, a personal digital assistant, a media player, a navigation device, an email device, a game console, a tablet computer, a wearable device, or any combination of these devices.

[0215] For convenience of description, when describing the above devices, various units are described separately according to their functions. Of course, when implementing this specification, the functions of each unit may be implemented in the same or multiple software and / or hardware.

[0216] Those skilled in the art should understand that the embodiments of this specification can be provided as a method, a system, or a computer program product. Therefore, the embodiments of this specification can take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware aspects. Moreover, the embodiments of this specification can take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) that contain computer-usable program code.

[0217] This specification is described with reference to the flowcharts and / or block diagrams of methods, apparatuses (systems), and computer program products according to the embodiments of this specification. It should be understood that each flow and / or block in the flowchart and / or block diagram, and the combination of flows and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to the processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing devices to generate a machine, such that the instructions executed by the processor of the computer or other programmable data processing devices generate a means for implementing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0218] These computer program instructions can also be stored in a computer-readable memory that can direct a computer or other programmable data processing device to work in a specific manner, such that the instructions stored in the computer-readable memory generate a manufactured article including an instruction means that implements the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0219] These computer program instructions can also be loaded onto a computer or other programmable data processing device, such that a series of operation steps are executed on the computer or other programmable device to generate a computer-implemented process. Thus, the instructions executed on the computer or other programmable device provide steps for implementing the functions specified in one Figure 1 one flow or multiple flows and / or blocks Figure 1 one block or multiple blocks.

[0220] In a typical configuration, a computing device includes one or more processors (CPUs), an input / output interface, a network interface, and a memory.

[0221] The memory may include non-permanent memory in the form of computer-readable media, random access memory (RAM), and / or non-volatile memory such as read-only memory (ROM) or flash memory (flash RAM). The memory is an example of computer-readable media.

[0222] Computer readable media include permanent and non-permanent, removable and non-removable media that can be implemented by any method or technology to store information. Information can be computer readable instructions, data structures, program modules or other data. Examples of computer storage media include, but are not limited to, phase change memory (PRAM), static random access memory (SRAM), dynamic random access memory (DRAM), other types of random access memory (RAM), read-only memory (ROM), electrically erasable programmable read-only memory (EEPROM), flash memory or other memory technology, compact disk read-only memory (CD-ROM), digital versatile disk (DVD) or other optical storage, magnetic cassettes, magnetic disk storage or other magnetic storage devices or any other non-transmission media that can be used to store information that can be accessed by a computing device. As defined herein, computer readable media does not include temporary computer readable media (transitory media), such as modulated data signals and carrier waves.

[0223] It should also be noted that the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, method, commodity or device including a series of elements includes not only those elements, but also other elements not explicitly listed, or also includes elements inherent to such process, method, commodity or device. In the absence of more restrictions, the elements defined by the sentence "comprises a ..." do not exclude the existence of other identical elements in the process, method, commodity or device including the elements.

[0224] This specification may be described in the general context of computer-executable instructions executed by a computer, such as program modules. Generally, program modules include routines, programs, objects, components, data structures, etc. that perform specific tasks or implement specific abstract data types. This specification may also be practiced in distributed computing environments where tasks are performed by remote processing devices connected through a communication network. In a distributed computing environment, program modules may be located in local and remote computer storage media, including storage devices.

[0225] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the system embodiment, since it is basically similar to the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0226] Finally, it should be noted that: What is disclosed by an image data processing method and system for container identification recognition disclosed in the embodiments of the present invention is only the preferred embodiments of the present invention, and is only used to illustrate the technical solutions of the present invention, rather than limiting it; Although the present invention has been described in detail with reference to the foregoing embodiments, those of ordinary skill in the art should understand that: They can still modify the technical solutions recorded in the foregoing embodiments, or perform equivalent replacements on some of the technical features; And these modifications or replacements do not make the essence of the corresponding technical solutions deviate from the spirit and scope of the technical solutions of the embodiments of the present invention.

Claims

1. An image data processing method for container identification and recognition, characterized in that The method includes: Obtaining a container image with a visual identification of the target container; Determining the identification character region in the container image based on an image segmentation algorithm; Based on the identification character region, determining the correctly arranged image corresponding to the identification character region based on an arrangement correction algorithm, including: Calculating the minimum rectangle of all polygon vertices including the identification character region to obtain an identification character rectangle; Determining arrangement method information corresponding to a plurality of the identification character images according to the lengths of different sides of the identification character rectangle; When the arrangement method information is horizontal arrangement, performing at least one rotation process with a first angle value on the identification character rectangle to obtain a plurality of first images before and after rotation; the first angle value is 180 degrees; Inputting each of the first images into a trained first character recognition algorithm model to obtain the recognition confidence corresponding to each of the first images; Determining the first image with the highest recognition confidence as the correctly arranged image; When the arrangement method information is vertical arrangement, inputting the identification character rectangle into a trained vertical arrangement prediction classifier to obtain the predicted vertical arrangement output; the predicted vertical arrangement is a single-line vertical arrangement, an upward-aligned multi-line vertical arrangement, a downward-aligned multi-line vertical arrangement, or a centered multi-line vertical arrangement; Horizontally arranging all the identification character images in the identification character rectangle based on the reading order corresponding to the predicted vertical arrangement to obtain an arranged image; Performing multiple rotation processes with a second angle value on the arranged image to obtain a plurality of second images; the second angle value is 90 degrees; Inputting each of the second images into a trained second character recognition algorithm model to obtain the recognition confidence corresponding to each of the second images; Determining the second image with the highest recognition confidence as the correctly arranged image; Performing character recognition on the correctly arranged image to obtain the identification information corresponding to the target container, including: Processing the correctly arranged image according to a trained character segmentation algorithm model to obtain a plurality of character images; Based on a plurality of character recognition models, performing recognition on each of the character images to obtain a plurality of model recognition results corresponding to each of the character images; Based on a voting algorithm, determining the character recognition result corresponding to each of the character images according to the plurality of model recognition results corresponding to each of the character images; Determining the character recognition results corresponding to all the character images as the identification information corresponding to the target container.

2. The image data processing method for container identification according to claim 1, characterized in that The determining the identification character region in the container image based on the image segmentation algorithm includes: Inputting the container image into a trained image segmentation neural network to obtain an output identification character region in the form of a polygon; the image segmentation neural network is based on the YOLOv8 algorithm architecture.

3. The image data processing method for container identification recognition according to claim 1, wherein The determining the arrangement method information corresponding to the identification character region according to the lengths of different sides of the identification character rectangle includes: Judging whether the length of the horizontal side of the identification character rectangle is greater than the length of the vertical side to obtain a first judgment result; If the first judgment result is yes, determine that the arrangement information of the identification character area is horizontal arrangement; If the first judgment result is no, determine that the arrangement information of the identification character area is vertical arrangement.

4. The image data processing method for container identification recognition according to claim 1, wherein The method further includes: Based on a preset character type position rule, filter out the mispositioned characters in the identification information; According to a preset character confusion correspondence, determine the correct character corresponding to each mispositioned character; Replace all the mispositioned characters in the identification information with the corresponding correct characters to obtain the corrected identification information.

5. An image data processing system for container identification recognition, characterized in that, The system includes: An acquisition module, configured to acquire a container image with a visual identification of a target container; A segmentation module, configured to determine an identification character area in the container image based on an image segmentation algorithm, including: Calculate the minimum rectangle of all polygon vertices including the identification character area to obtain an identification character rectangle; Determine the arrangement information corresponding to a plurality of the identification character images according to the lengths of different sides of the identification character rectangle; When the arrangement information is horizontal arrangement, perform at least one rotation process with a first angle value on the identification character rectangle to obtain a plurality of first images before and after rotation; the first angle value is 180 degrees; Input each of the first images into a trained first character recognition algorithm model to obtain the recognition confidence corresponding to each of the first images; Determine the first image with the highest recognition confidence as the correctly arranged image; When the arrangement information is vertical arrangement, input the identification character rectangle into a trained vertical arrangement prediction classifier to obtain the predicted vertical arrangement output; the predicted vertical arrangement is a single-line vertical arrangement, an upward-aligned multi-line vertical arrangement, a downward-aligned multi-line vertical arrangement, or a centered multi-line vertical arrangement; Horizontally arrange all the identification character images in the identification character rectangle based on the reading order corresponding to the predicted vertical arrangement to obtain an arranged image; Perform a rotation process with a second angle value on the arranged image multiple times to obtain a plurality of second images; the second angle value is 90 degrees; Input each of the second images into a trained second character recognition algorithm model to obtain the recognition confidence corresponding to each of the second images; Determine the second image with the highest recognition confidence as the correctly arranged image; A determination module, configured to determine the correctly arranged image corresponding to the identification character area based on an arrangement correction algorithm according to the identification character area; An identification module, configured to perform character recognition on the correctly arranged image to obtain the identification information corresponding to the target container, including: Process the correctly arranged image according to a trained character segmentation algorithm model to obtain a plurality of character images; Perform recognition on each of the character images based on a plurality of character recognition models to obtain a plurality of model recognition results corresponding to each of the character images; Based on a voting algorithm, determine the character recognition result corresponding to each of the character images according to the plurality of model recognition results corresponding to each of the character images; Determine the character recognition results corresponding to all the character images as the identification information corresponding to the target container.

6. An image data processing system for container identification recognition, characterized in that The system includes: A memory storing executable program codes; A processor coupled to the memory; The processor calls the executable program codes stored in the memory and executes the image data processing method for container identification according to any one of claims 1-4.

Images