Label information identification method, system and device and computer storage medium

Abstract

The invention discloses a label information identification method, system and device and a computer storage medium, and relates to the technical field of image detection, and the method comprises the steps: obtaining a to-be-identified target image; inputting the target image into a trained label detection model to obtain label information output by the label detection model; according to the label information, a digital image is extracted from the target image, and the digital image comprises an image carrying digital information in the label; and inputting the digital image into the trained label identification model to obtain digital identification information output by the label identification model. According to the method, the label detection model is used for performing label detection on the target image, so that the label information in the target image can be extracted, then the digital information in the label can be exposed, and a subsequent label identification model can conveniently identify the digital image to obtain the digital identification information in the label; the label digital information in the image is identified by means of the label detection model and the label identification model, the implementation is convenient, and the accuracy is high.

Description

Technical Field

[0001] The present application relates to the field of image detection technology, and more specifically, to a label information recognition method, system, device and computer storage medium. Background Art

[0002] Currently, in order to facilitate the management of items, tags can be set for items, and the relevant information of the items can be recorded with the help of tags. In this way, the information of the items can be obtained by simply checking the tags. However, there are numbers in the tags, and due to specific needs, the numbers are expressed in different forms in the tags, which makes it difficult to accurately identify the numbers in the tags, affecting the management of the items.

[0003] In summary, how to accurately identify digital information in a label is a problem that needs to be solved urgently by those skilled in the art. Summary of the invention

[0004] The purpose of this application is to provide a tag information recognition method, which can solve the technical problem of how to accurately recognize digital information in a tag to a certain extent. This application also provides a tag information recognition system, an electronic device and a computer-readable storage medium.

[0005] In order to achieve the above objectives, this application provides the following technical solutions:

[0006] A tag information recognition method, comprising:

[0007] Acquire a target image to be identified;

[0008] Inputting the target image into a trained label detection model to obtain label information output by the label detection model;

[0009] Extracting a digital image from the target image according to the tag information, the digital image including an image carrying the digital information in the tag;

[0010] The digital image is input into a trained label recognition model to obtain digital recognition information output by the label recognition model.

[0011] Preferably, the label information includes a detection frame of the label and corner point information of the detection frame;

[0012] The step of extracting a digital image from the target image according to the tag information, wherein the digital image includes an image carrying the digital information in the tag, comprises:

[0013] Extracting a label image from the target image according to the detection frame and the corner point information;

[0014] Performing perspective transformation and size adjustment on the label image to obtain a transformed image;

[0015] Each digital region in the transformed image is extracted and resized to obtain a corresponding digital image.

[0016] Preferably, after inputting the digital image into a trained label recognition model to obtain the digital recognition information output by the label recognition model, the method further comprises:

[0017] The digital recognition information is detected, and a digital recognition result of the label in the target image is determined according to the detection result.

[0018] Preferably, the digital identification information includes the numerical value at each position and the predicted probability value of the numerical value at each position;

[0019] The detecting the digital identification information and determining the digital identification result of the label in the target image according to the detection result includes:

[0020] Determining the value of the digits in the digital identification information;

[0021] Detecting whether the digital digit value complies with a preset digit rule;

[0022] In response to the digital digit value not complying with the digit rule, a digital recognition result is obtained indicating that the digits in the label are inaccurate.

[0023] Preferably, after detecting whether the digital digit value meets the preset digit rule, the method further includes:

[0024] In response to the digital place value complying with the place rule, generating an average value of the predicted probability values ​​of the values ​​at each position;

[0025] Detecting whether the predicted probability values ​​of the numerical values ​​at each position are all equal to the average value, and whether the average value is greater than a first set value;

[0026] In response to the predicted probability values ​​of the numerical values ​​at each position being equal to the average value, and the average value being greater than the first set value, the numerical values ​​at each position are combined according to the relationship between the digital areas to obtain a digital recognition result.

[0027] Preferably, after detecting whether the predicted probability values ​​of the numerical values ​​at each position are all greater than or equal to the average value, and whether the average value is greater than a first set value, the method further includes:

[0028] In response to the predicted probability values ​​of the numerical values ​​at each position being equal to the average value and the average value being less than or equal to the first set value, or in response to the existence of a predicted probability value not equal to the average value, the predicted probability value less than the average value is selected as the probability value to be detected;

[0029] Detecting whether all probability values ​​to be detected are greater than or equal to a second set value;

[0030] In response to all the probability values ​​to be detected being greater than or equal to the second set value, combining the values ​​at each position according to the relationship between the digital regions to obtain a digital recognition result;

[0031] In response to the existence of a to-be-detected probability value that is less than the second set value, a number recognition result is obtained that indicates that the number in the label is inaccurate.

[0032] Preferably, after performing perspective transformation and resizing on the label image to obtain a transformed image, the method further comprises:

[0033] Extracting the barcode in the transformed image to obtain an initial barcode image;

[0034] Processing the initial barcode image to obtain a target barcode image;

[0035] The target barcode image is recognized to obtain a barcode recognition result of the label in the target image.

[0036] A label information recognition system, comprising:

[0037] A first acquisition module, used to acquire a target image to be identified;

[0038] A first detection module, used to input the target image into a trained label detection model to obtain label information output by the label detection model;

[0039] A first extraction module, configured to extract a digital image from the target image according to the tag information, wherein the digital image includes an image carrying the digital information in the tag;

[0040] The first recognition module is used to input the digital image into a trained label recognition model to obtain digital recognition information output by the label recognition model.

[0041] An electronic device, comprising:

[0042] Memory for storing computer programs;

[0043] A processor is used to implement the steps of any of the above-mentioned tag information identification methods when executing the computer program.

[0044] A computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of any of the above-mentioned tag information identification methods are implemented.

[0045] The present application provides a label information recognition method, which includes obtaining a target image to be recognized; inputting the target image into a trained label detection model to obtain label information output by the label detection model; extracting a digital image from the target image according to the label information, wherein the digital image includes an image carrying digital information in the label; inputting the digital image into a trained label recognition model to obtain digital recognition information output by the label recognition model. The present application uses a label detection model to perform label detection on the target image, thereby extracting label information in the target image, and then extracting a digital image carrying digital information in the label from the target image, so that the digital information in the label can be exposed, which is convenient for the subsequent label recognition model to recognize the digital image and obtain the digital recognition information in the label, and realizes the recognition of label digital information in the image by means of a label detection model and a label recognition model, which is convenient for implementation, and compared with directly extracting label digital information from the image, the label detection model and the label recognition model belong to a neural network model, which can recognize digital information in the label from multiple angles with higher accuracy. The present application provides a label information recognition system, an electronic device, and a computer-readable storage medium, which also solves corresponding technical problems. BRIEF DESCRIPTION OF THE DRAWINGS

[0046] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are only embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on the provided drawings without paying any creative work.

[0047] Figure 1 A flowchart of a tag information identification method provided in an embodiment of the present application;

[0048] Figure 2 A schematic diagram of a label;

[0049] Figure 3 It is a schematic diagram for detecting digital identification information;

[0050] Figure 4 This is a schematic diagram of the architecture of the label recognition model;

[0051] Figure 5 Flowchart for price tag recognition for robots;

[0052] Figure 6A structural diagram of a label information identification system provided in an embodiment of the present application;

[0053] Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application;

[0054] Figure 8 Another structural schematic diagram of an electronic device provided in an embodiment of the present application. DETAILED DESCRIPTION

[0055] The following will be combined with the drawings in the embodiments of the present application to clearly and completely describe the technical solutions in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0056] See also Figure 1 , Figure 1 A flowchart of a tag information identification method provided in an embodiment of the present application.

[0057] A tag information identification method provided in an embodiment of the present application may include the following steps:

[0058] Step S101: Acquire a target image to be identified.

[0059] In actual applications, the target image to be identified can be obtained first. The source of the target image can be flexibly determined according to the application scenario. For example, when the target image carries a label, it can be an image obtained by photographing the price tag of the item, or it can be an image obtained by photographing the production date label of the item, etc. For another example, when the target image does not necessarily carry a label, the target image can be an image obtained by photographing the item during the label generation process of the item, etc.

[0060] Step S102: input the target image into the trained label detection model to obtain label information output by the label detection model.

[0061] In practical applications, after obtaining the target image to be identified, the target image does not necessarily carry a label, and if it carries a label, the position of the label in the target image, the size of the label, and other information are unknown, so the label in the target image needs to be detected first. In this process, the target image needs to be input into a trained label detection model to obtain the label information output by the label detection model. This label information is used to characterize the corresponding information of the label in the target image. The content of the label information can be flexibly determined as needed, such as the position of the label in the target image, the size of the label, the mark of the label, etc. In addition, when there is no label in the target image, the label detection model can output information representing that there is no label in the target image.

[0062] It should be noted that, according to the use requirements, you can flexibly select general target detection models such as the two-stage target detection model Faster R-CNN, SPPNet or the one-stage target detection model YOLOv4 / v5, SSD and RetinaNet as the basic network of the label detection model. Since these general target detection models have been fully trained on large-scale public data sets such as ImageNet and MSCOCO, they have good target detection effects and can converge to higher accuracy more quickly. On this basis, in order to improve the success rate of label detection, after adjusting the output category of the general basic network of the label detection model accordingly, it is necessary to use the pre-labeled label data for targeted training.

[0063] Step S103: extracting a digital image from the target image according to the tag information, where the digital image includes an image carrying the digital information in the tag.

[0064] Step S104: input the digital image into the trained label recognition model to obtain the digital recognition information output by the label recognition model.

[0065] In practical applications, after obtaining the label information in the target image, although the label information can be used to know the label-related information in the target image, considering the diversity of label content, other information in the label will interfere with the recognition of digital information. In order to avoid this situation, the digital image can be extracted from the target image according to the label information, and the digital image includes the image carrying the digital information in the label. Of course, due to operational limitations, the digital image may carry other information in the label, but as long as the proportion of digital information is greater than the proportion of other information; then the digital image is input into the trained label recognition model to obtain the digital recognition information output by the label recognition model, and this digital recognition information is used to record the digital information in the label of the target image.

[0066] In a specific application scenario, the label information may include the detection frame of the label and the corner point information of the detection frame, wherein the corner point information of the detection frame may be the information of four corner points, and the label information may be 13 bits in total, including class, x, y, w, h, px1, py1, px2, py2, px3, py3, px4, py4, etc., wherein class is the category index representing the detection frame, x, y is the center point of the detection frame divided by the width and height of the image, w, h is the width and height of the detection frame divided by the width and height of the image, the key points are respectively the normalized or original upper left, upper right, lower right, and lower left, px, py are the values ​​of the key point coordinates divided by the width and height; accordingly, in the process of extracting a digital image from a target image according to the label information, the digital image includes an image carrying digital information in the label, and the label image can be extracted from the target image according to the detection frame and the corner point information; considering the quality and size of the target image, the extracted label image may be deformed or the size may not meet the requirements, so the label image needs to be perspective transformed (perspective transformation) Transformation) and resize, for example, resize the label image to (456, 226), etc., to obtain a transformation image that meets the requirements; then extract and resize each digital area in the transformation image to obtain a corresponding digital image. Each digital area in the transformation image can be flexibly determined as needed, for example, the digital area can be each area with different digital sizes, etc. Figure 2 Taking the label shown as an example, the area where 12 is located, the area where 34, 6, 291.12 are located can all be used as digital areas, etc. In addition, the size of the digital image can be set to (142, 64), etc.

[0067] It should be noted that perspective transformation is a technique in image processing, which is used to project an image from one perspective to another. It is usually used to correct perspective distortion in an image or extract regions of interest. The principle of perspective transformation is to map the pixel coordinates in the source image to the pixel coordinates in the target image by calculating the perspective transformation matrix (Perspective Transformation Matrix). This matrix is ​​a 3x3 matrix, which can convert the coordinates in the source image to the coordinates in the target image through matrix operations. Assuming that [X,Y,Z] is the transformed point, [x,y,1] is the origin, and the 3*3 matrix is ​​the perspective transformation matrix, the transformation formula for perspective transformation can be:

[0068] .

[0069] In practical applications, after a digital image is input into a trained label recognition model and the digital recognition information is output by the label recognition model, the recognized digital information may not be accurate due to the quality of the target image. For example, when the target image is blurred, the accuracy of the recognized digital information is poor. In order to more accurately recognize the digital information, it is necessary to further verify the digital recognition information, that is, to detect the digital recognition information and determine the digital recognition result of the label in the target image based on the detection result.

[0070] In a specific application scenario, the digital recognition information may include the numerical value at each position and the predicted probability value of the numerical value at each position; accordingly, in the process of detecting the digital recognition information and determining the digital recognition result of the label in the target image according to the detection result, such as Figure 3 As shown, the digit value in the digital identification information can be determined; the digit value is detected whether it meets the preset digit rule, for example, if the digit rule is 4 digits, then when the digit value is greater than 4 digits, it is determined that the digit value does not meet the preset digit rule, etc.; in response to the digit value not meeting the digit rule, a digital identification result representing that the number in the label is inaccurate is obtained. In response to the digit value meeting the digit rule, an average value of the predicted probability values ​​of the values ​​at each position is generated; it is detected whether the predicted probability values ​​of the values ​​at each position are all equal to the average value, and whether the average value is greater than a first set value; in response to the predicted probability values ​​of the values ​​at each position being equal to the average value, and the average value is greater than a first set value, the values ​​at each position are combined according to the relationship between the digital regions to obtain a digital recognition result, still in the form of Figure 2 Take the label shown as an example. Since 12 and 34 in this label are both prices, 12 and 34 need to be combined to obtain 12.34 as the numerical recognition result. In response to the predicted probability values ​​of the numerical values ​​at each position being equal to the average value and the average value being less than or equal to the first set value, or in response to the existence of a predicted probability value that is not equal to the average value, the predicted probability value less than the average value is screened as the probability value to be detected; detect whether all the probability values ​​to be detected are greater than or equal to the second set value; in response to all the probability values ​​to be detected being greater than or equal to the second set value, the numerical values ​​at each position are combined according to the relationship between the digital regions to obtain the digital recognition result; in response to the existence of a probability value to be detected that is less than the second set value, a digital recognition result representing the inaccuracy of the number in the label is obtained. Among them, the first set value and the second set value can be flexibly determined according to the accuracy requirements for digital recognition, for example, the first set value and the second set value can both be 0.7, or the first set value is 0.8, the second set value is 0.7, etc.

[0071] It should be noted that the type of the label recognition model of the present application can be flexibly determined as needed. Considering that the CRNN (Convolutional Recurrent Neural Network) model is a deep learning model that combines convolutional neural network (CNN) and recurrent neural network (RNN), it is mainly used for sequence data recognition tasks. Because it combines the advantages of both, it can effectively process sequence data with spatial and temporal dependencies. Therefore, the label recognition model of the present application can be built based on the CRNN model, and the architecture of the label recognition model can be as follows: Figure 4 As shown, K represents kernel_size (convolution kernel size), P represents padding, S represents stride, BatchNorm2d represents batch normalization layer, ReLU represents activation function, MaxPool2d represents maximum pooling layer, Conv2d represents convolution layer, AdaptiveAvgPool2d represents adaptive average pooling layer, Conv2d_new represents new convolution layer, the output channel of the new convolution layer can be 10, etc., Char represents character, and Prob represents probability.

[0072] In practical applications, after the label image is perspective transformed and resized to obtain the transformed image, considering that there may be a barcode in the label, the barcode in the transformed image can also be extracted to obtain the initial barcode image. The barcode can be in the form of QR code, Data Matrix, Aztec, UPC-A, UPC-E, EAN-8, EAN-13, Code39, Code93, Code128, etc.; the initial barcode image is processed, such as grayscale, histogram equalization, etc. to reduce noise and interference, improve contrast and details, and obtain the target barcode image; the target barcode image is recognized, such as using a barcode decoding library to recognize the target barcode image to obtain the barcode recognition result of the label in the target image.

[0073] The present application provides a label information recognition method, which includes obtaining a target image to be recognized; inputting the target image into a trained label detection model to obtain label information output by the label detection model; extracting a digital image from the target image according to the label information, wherein the digital image includes an image carrying digital information in the label; and inputting the digital image into a trained label recognition model to obtain digital recognition information output by the label recognition model. The present application uses a label detection model to perform label detection on the target image, thereby extracting label information in the target image, and then extracting a digital image carrying digital information in the label from the target image, so that the digital information in the label can be exposed, which is convenient for the subsequent label recognition model to recognize the digital image and obtain the digital recognition information in the label, and realizes the recognition of label digital information in the image by using a label detection model and a label recognition model, which is easy to implement, and compared with directly extracting label digital information from the image, the label detection model and the label recognition model belong to a neural network model, which can recognize digital information in the label from multiple angles with higher accuracy.

[0074] In order to facilitate understanding of the label information recognition solution provided by the present application, the process of robot identifying the price tag is described below. Figure 5 As shown, it is applied to the robot device side and includes the following steps:

[0075] A single captured image is used to generate the target image;

[0076] Input the target image into the trained label detection model to obtain the label information output by the label detection model. The label information includes the label detection box and the corner point information of the detection box.

[0077] Extract the label image from the target image based on the detection box and corner point information;

[0078] Perform perspective transformation and size adjustment on the label image to obtain a transformed image;

[0079] Extract the barcode in the label image to obtain an initial barcode image;

[0080] Processing the initial barcode image to obtain a target barcode image;

[0081] Recognize the target barcode image to obtain the barcode recognition result of the label in the target image;

[0082] Extracting and resizing each digital region in the transformed image to obtain a corresponding digital image;

[0083] Input the digital image into the trained label recognition model, obtain the digital recognition information output by the label recognition model and save it.

[0084] See also Figure 6 , Figure 6 A structural diagram of a label information identification system provided in an embodiment of the present application.

[0085] A label information recognition system provided in an embodiment of the present application may include:

[0086] A first acquisition module 101 is used to acquire a target image to be identified;

[0087] The first detection module 102 is used to input the target image into the trained label detection model to obtain label information output by the label detection model;

[0088] A first extraction module 103 is used to extract a digital image from a target image according to the tag information, wherein the digital image includes an image carrying the digital information in the tag;

[0089] The first recognition module 104 is used to input the digital image into the trained label recognition model to obtain the digital recognition information output by the label recognition model.

[0090] The embodiment of the present application provides a label information recognition system, wherein the label information includes a detection frame of the label and corner point information of the detection frame;

[0091] The first extraction module may include:

[0092] A first extraction unit, used to extract a label image from a target image according to a detection frame and corner point information;

[0093] A first adjustment unit, used for performing perspective transformation and size adjustment on the label image to obtain a transformed image;

[0094] The second adjustment unit is used to extract and resize each digital area in the transformed image to obtain a corresponding digital image.

[0095] The present invention provides a label information recognition system, which further includes:

[0096] The first detection module is used for the first recognition module to input the digital image into the trained label recognition model, obtain the digital recognition information output by the label recognition model, detect the digital recognition information, and determine the digital recognition result of the label in the target image according to the detection result.

[0097] An embodiment of the present application provides a label information recognition system, wherein the digital recognition information includes a value at each position and a predicted probability value of the value at each position;

[0098] The first detection module may include:

[0099] A first determining unit, used to determine the digit value in the digital identification information;

[0100] The first detection unit is used to detect whether the digital digit value meets the preset digit rule; in response to the digital digit value not meeting the digit rule, a digital recognition result representing that the digit in the label is inaccurate is obtained.

[0101] A label information identification system provided by an embodiment of the present application, the first detection unit is also used to: after detecting whether the digital digit value conforms to a preset digit rule, in response to the digital digit value conforming to the digit rule, generate an average value of the predicted probability values ​​of the values ​​at each position; detect whether the predicted probability values ​​of the values ​​at each position are all equal to the average value, and whether the average value is greater than a first set value; in response to the predicted probability values ​​of the values ​​at each position being equal to the average value, and the average value being greater than the first set value, combine the values ​​at each position according to the relationship between the digital regions to obtain a digital recognition result.

[0102] A label information identification system provided by an embodiment of the present application, wherein the first detection unit is further used to: detect whether the predicted probability values ​​of the numerical values ​​at each position are all greater than or equal to the average value, and whether the average value is greater than a first set value, and in response to the predicted probability values ​​of the numerical values ​​at each position being equal to the average value and the average value being less than or equal to the first set value, or in response to the existence of predicted probability values ​​not equal to the average value, screen the predicted probability values ​​less than the average value as the probability values ​​to be detected; detect whether all the probability values ​​to be detected are greater than or equal to the second set value; in response to all the probability values ​​to be detected being greater than or equal to the second set value, combine the numerical values ​​at each position according to the relationship between the digital regions to obtain a digital recognition result; in response to the existence of a probability value to be detected that is less than the second set value, obtain a digital recognition result representing that the number in the label is inaccurate.

[0103] A label information recognition system provided in an embodiment of the present application may further include:

[0104] A second extraction unit is used for the first adjustment unit to perform perspective transformation and size adjustment on the label image, and after obtaining the transformed image, extract the barcode in the transformed image to obtain an initial barcode image;

[0105] A first processing unit, used for processing the initial barcode image to obtain a target barcode image;

[0106] The first recognition unit is used to recognize the target barcode image and obtain the barcode recognition result of the label in the target image.

[0107] The present application also provides an electronic device and a computer-readable storage medium, both of which have the corresponding effects of the tag information identification method provided in the embodiment of the present application. Figure 7, Figure 7 A schematic diagram of the structure of an electronic device provided in an embodiment of the present application.

[0108] An electronic device provided in an embodiment of the present application includes a memory 201 and a processor 202. The memory 201 stores a computer program. When the processor 202 executes the computer program, the steps of the tag information identification method described in any of the above embodiments are implemented.

[0109] See also Figure 8 , another electronic device provided in the embodiment of the present application may also include: an input port 203 connected to the processor 202, for transmitting commands input from the outside to the processor 202; a display unit 204 connected to the processor 202, for displaying the processing results of the processor 202 to the outside; a communication module 205 connected to the processor 202, for realizing communication between the electronic device and the outside. The display unit 204 can be a display panel, a laser scanning display, etc.; the communication method adopted by the communication module 205 includes but is not limited to mobile high-definition link technology (Mobile High-Definition Link, MHL), Universal Serial Bus (Universal Serial Bus, USB), High-Definition Multimedia Interface (High-DefinitionMultimedia Interface, HDMI), wireless connection: wireless fidelity technology (WIreless Fidelity, WiFi), Bluetooth communication technology, low-power Bluetooth communication technology, and communication technology based on IEEE802.11s.

[0110] An embodiment of the present application provides a computer-readable storage medium, in which a computer program is stored. When the computer program is executed by a processor, the steps of the tag information identification method described in any of the above embodiments are implemented.

[0111] The computer-readable storage medium involved in this application includes random access memory (RAM), memory, read-only memory (ROM), electrically programmable ROM, electrically erasable programmable ROM, registers, hard disks, removable disks, CD-ROMs (Compact Disc Read-Only Memory), or any other form of storage medium known in the technical field.

[0112] For the description of the relevant parts of the tag information identification system, electronic device, and computer-readable storage medium provided in the embodiments of the present application, please refer to the detailed description of the corresponding parts in the tag information identification method provided in the embodiments of the present application, which will not be repeated here. In addition, the parts of the above technical solutions provided in the embodiments of the present application that are consistent with the implementation principles of the corresponding technical solutions in the prior art are not described in detail to avoid excessive elaboration.

[0113] It should also be noted that, in this article, relational terms such as first and second, etc. are only used to distinguish one entity or operation from another entity or operation, and do not necessarily require or imply any such actual relationship or order between these entities or operations. Moreover, the terms "include", "comprise" or any other variants thereof are intended to cover non-exclusive inclusion, so that a process, method, article 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, article or device. In the absence of further restrictions, the elements defined by the sentence "comprise a ..." do not exclude the presence of other identical elements in the process, method, article or device including the elements.

[0114] The above description of the disclosed embodiments enables those skilled in the art to implement or use the present application. Various modifications to these embodiments will be apparent to those skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application will not be limited to the embodiments shown herein, but will conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A tag information recognition method, characterized in that: include: Acquire a target image to be identified; Inputting the target image into a trained label detection model to obtain label information output by the label detection model; Extracting a digital image from the target image according to the tag information, the digital image including an image carrying the digital information in the tag; The digital image is input into a trained label recognition model to obtain digital recognition information output by the label recognition model.

2. The method according to claim 1, characterized in that The label information includes a detection frame of the label and corner point information of the detection frame; The step of extracting a digital image from the target image according to the tag information, wherein the digital image includes an image carrying the digital information in the tag, comprises: Extracting a label image from the target image according to the detection frame and the corner point information; Performing perspective transformation and size adjustment on the label image to obtain a transformed image; Each digital region in the transformed image is extracted and resized to obtain a corresponding digital image.

3. The method according to claim 2, characterized in that After inputting the digital image into the trained label recognition model to obtain the digital recognition information output by the label recognition model, the method further includes: The digital recognition information is detected, and a digital recognition result of the label in the target image is determined according to the detection result.

4. The method according to claim 3, characterized in that: The digital identification information includes the numerical value at each position and the predicted probability value of the numerical value at each position; The detecting the digital identification information and determining the digital identification result of the label in the target image according to the detection result includes: Determining the value of the digits in the digital identification information; Detecting whether the digital digit value complies with a preset digit rule; In response to the digital digit value not complying with the digit rule, a digital recognition result is obtained indicating that the digits in the label are inaccurate.

5. The method according to claim 4, characterized in that After detecting whether the digital digit value meets the preset digit rule, the method further includes: In response to the digital place value complying with the place rule, generating an average value of the predicted probability values ​​of the values ​​at each position; Detecting whether the predicted probability values ​​of the numerical values ​​at each position are all equal to the average value, and whether the average value is greater than a first set value; In response to the predicted probability values ​​of the numerical values ​​at each position being equal to the average value, and the average value being greater than the first set value, the numerical values ​​at each position are combined according to the relationship between the digital areas to obtain a digital recognition result.

6. The method according to claim 5, characterized in that After detecting whether the predicted probability values ​​of the values ​​at each position are all greater than or equal to the average value, and whether the average value is greater than the first set value, the method further includes: In response to the predicted probability values ​​of the numerical values ​​at each position being equal to the average value and the average value being less than or equal to the first set value, or in response to the existence of a predicted probability value not equal to the average value, the predicted probability value less than the average value is selected as the probability value to be detected; Detecting whether all probability values ​​to be detected are greater than or equal to a second set value; In response to all the probability values ​​to be detected being greater than or equal to the second set value, combining the values ​​at each position according to the relationship between the digital regions to obtain a digital recognition result; In response to the existence of a to-be-detected probability value that is less than the second set value, a number recognition result is obtained that indicates that the number in the label is inaccurate.

7. The method according to claim 2, characterized in that After performing perspective transformation and resizing on the label image to obtain a transformed image, the method further includes: Extracting the barcode in the transformed image to obtain an initial barcode image; Processing the initial barcode image to obtain a target barcode image; The target barcode image is recognized to obtain a barcode recognition result of the label in the target image.

8. A label information recognition system, characterized in that: include: A first acquisition module, used to acquire a target image to be identified; A first detection module, used to input the target image into a trained label detection model to obtain label information output by the label detection model; A first extraction module, configured to extract a digital image from the target image according to the tag information, wherein the digital image includes an image carrying the digital information in the tag; The first recognition module is used to input the digital image into a trained label recognition model to obtain digital recognition information output by the label recognition model.

9. An electronic device, characterized in that: include: Memory for storing computer programs; A processor, configured to implement the steps of the tag information identification method according to any one of claims 1 to 7 when executing the computer program.

10. A computer-readable storage medium, characterized in that: The computer-readable storage medium stores a computer program, and when the computer program is executed by a processor, the steps of the tag information recognition method according to any one of claims 1 to 7 are implemented.

Images