CN116597430A - Article identification method, apparatus, electronic device, and computer-readable medium - Google Patents
Article identification method, apparatus, electronic device, and computer-readable medium Download PDFInfo
- Publication number
- CN116597430A CN116597430A CN202310512053.0A CN202310512053A CN116597430A CN 116597430 A CN116597430 A CN 116597430A CN 202310512053 A CN202310512053 A CN 202310512053A CN 116597430 A CN116597430 A CN 116597430A
- Authority
- CN
- China
- Prior art keywords
- article
- sample
- page
- image
- screenshot
- Prior art date
- Legal status (The legal status is an assumption and is not a legal conclusion. Google has not performed a legal analysis and makes no representation as to the accuracy of the status listed.)
- Pending
Links
- 238000000034 method Methods 0.000 title claims abstract description 46
- 230000004044 response Effects 0.000 claims abstract description 47
- 238000012015 optical character recognition Methods 0.000 claims abstract description 20
- 238000001514 detection method Methods 0.000 claims abstract description 19
- 238000013528 artificial neural network Methods 0.000 claims description 39
- 238000012549 training Methods 0.000 claims description 21
- 238000012545 processing Methods 0.000 claims description 17
- 238000005457 optimization Methods 0.000 claims description 11
- 230000008859 change Effects 0.000 claims description 10
- 238000004590 computer program Methods 0.000 claims description 9
- 238000007639 printing Methods 0.000 claims description 2
- 238000005303 weighing Methods 0.000 abstract description 19
- 230000001960 triggered effect Effects 0.000 abstract description 15
- 230000008569 process Effects 0.000 abstract description 8
- 238000000605 extraction Methods 0.000 description 19
- 230000006870 function Effects 0.000 description 13
- 238000004891 communication Methods 0.000 description 6
- 238000010586 diagram Methods 0.000 description 6
- 230000008901 benefit Effects 0.000 description 4
- 238000003062 neural network model Methods 0.000 description 4
- 230000003287 optical effect Effects 0.000 description 3
- 238000006243 chemical reaction Methods 0.000 description 2
- 238000013527 convolutional neural network Methods 0.000 description 2
- 238000005516 engineering process Methods 0.000 description 2
- 238000011478 gradient descent method Methods 0.000 description 2
- 239000010893 paper waste Substances 0.000 description 2
- 230000000644 propagated effect Effects 0.000 description 2
- ORILYTVJVMAKLC-UHFFFAOYSA-N Adamantane Natural products C1C(C2)CC3CC1CC2C3 ORILYTVJVMAKLC-UHFFFAOYSA-N 0.000 description 1
- 238000012935 Averaging Methods 0.000 description 1
- 230000004913 activation Effects 0.000 description 1
- 230000001174 ascending effect Effects 0.000 description 1
- 230000000694 effects Effects 0.000 description 1
- 239000000835 fiber Substances 0.000 description 1
- 239000004973 liquid crystal related substance Substances 0.000 description 1
- 230000007257 malfunction Effects 0.000 description 1
- 230000001343 mnemonic effect Effects 0.000 description 1
- 239000013307 optical fiber Substances 0.000 description 1
- 238000011176 pooling Methods 0.000 description 1
- 230000009467 reduction Effects 0.000 description 1
- 239000004065 semiconductor Substances 0.000 description 1
- 238000012546 transfer Methods 0.000 description 1
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
- G06V20/60—Type of objects
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V10/00—Arrangements for image or video recognition or understanding
- G06V10/70—Arrangements for image or video recognition or understanding using pattern recognition or machine learning
- G06V10/82—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using neural networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V30/00—Character recognition; Recognising digital ink; Document-oriented image-based pattern recognition
- G06V30/10—Character recognition
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- General Physics & Mathematics (AREA)
- Multimedia (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Health & Medical Sciences (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- Databases & Information Systems (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Image Analysis (AREA)
Abstract
Embodiments of the present disclosure disclose an article identification method, apparatus, electronic device, and computer readable medium. One embodiment of the method comprises the following steps: acquiring a screenshot set of an article information display page; inputting the display area image into a pre-trained optical character recognition model to obtain weight recognition information; in response to the weight identification information including a weight value and the value being greater than a preset value, performing the steps of: acquiring an image of a bearing disc; carrying out object identification on the bearing disc image; acquiring a bearing disc image in response to the weight identification information being empty; inputting the bearing disc image into a pre-trained empty disc detection model; and responding to the empty disc identification result to represent that the article is carried in the article carrying disc, and carrying out article identification on the carrying disc image to obtain article identification information. The embodiment realizes that the object identification can be triggered even if the weight of the object cannot be read when the weighing disc fails in the weighing process. The application scene of article identification is widened.
Description
Technical Field
Embodiments of the present disclosure relate to the field of computer technology, and in particular, to an article identification method, apparatus, electronic device, and computer readable medium.
Background
Item identification is a technique for identifying items. Currently, in identifying an article, the following methods are generally adopted: the article information identification is triggered according to the change of the weight by directly weighing the article, and then the identified article information is printed on the label paper.
However, when the above manner is adopted to identify the article, there are often the following technical problems:
first, through the mode of directly weighing the article, trigger article information discernment according to the change of weight, when weighing the dish and breaking down in the weighing process, the weight of article can not be read, leads to unable triggering article discernment.
Secondly, the object information is identified through directly acquired bearing disc images, the background of the bearing disc images is interfered, and the accuracy of the identified object information is low. For the label paper with wrong printed article information, the article information needs to be recognized again, so that the label paper is wasted.
Disclosure of Invention
The disclosure is in part intended to introduce concepts in a simplified form that are further described below in the detailed description. The disclosure is not intended to identify key features or essential features of the claimed subject matter, nor is it intended to be used to limit the scope of the claimed subject matter.
Some embodiments of the present disclosure propose article identification methods, apparatuses, electronic devices, and computer-readable media to address one or more of the technical problems mentioned in the background section above.
In a first aspect, some embodiments of the present disclosure provide an article identification method comprising: and acquiring a screenshot set of the item information display page within a preset time period. The page screenshot in the page screenshot set comprises a weight display area; generating a weight information display area image as a display area image according to the page screenshot set; inputting the display area image into a pre-trained optical character recognition model to obtain weight recognition information in the display area image; in response to determining that the weight identification information includes a weight identification value and that the weight identification value is greater than a preset value, performing the steps of: and determining that the article identification triggering condition is met. Wherein the article carrying tray corresponds to the article information display page; acquiring an article carrying tray image of the article carrying tray as a first article carrying tray image; carrying out article identification on the first article carrying tray image to obtain article identification information; acquiring an article carrying tray image of the article carrying tray as a second article carrying tray image in response to determining that the weight identification information is empty; and inputting the second object bearing disc image into a pre-trained empty disc detection model to obtain an empty disc identification result. The empty disc identification result represents whether the article bearing disc bears articles or not; responding to the fact that the empty disc identification result represents that articles are carried in the article carrying disc, and determining that the article identification triggering condition is met; and in response to determining that the article identification triggering condition is met, carrying out article identification on the second article carrying tray image to obtain article identification information.
In a second aspect, some embodiments of the present disclosure provide an article identification device, the device comprising: the first acquisition unit is configured to acquire a screenshot set of an article information display page within a preset time period; a first input unit configured to input the display area image into a pre-trained optical character recognition model to obtain weight recognition information in the display area image; a first determining unit configured to determine that an article identification triggering condition is satisfied, acquire an article carrying tray image of the article carrying tray as a first article carrying tray image, and perform article identification on the first article carrying tray image to obtain article identification information; a second acquisition unit configured to acquire an article carrying tray image of the article carrying tray as a second article carrying tray image in response to determining that the weight identification information is empty; the second input unit is configured to input the second article bearing disc image into a pre-trained empty disc detection model to obtain an empty disc identification result; a second determining unit configured to determine that an article identification triggering condition is satisfied in response to determining that the empty tray identification result characterizes that the article is carried in the article carrying tray; and the identification unit is used for carrying out article identification on the second article carrying tray image to obtain article identification information in response to determining that the article identification triggering condition is met.
In a third aspect, some embodiments of the present disclosure provide an electronic device comprising: one or more processors; a storage device having one or more programs stored thereon, which when executed by one or more processors causes the one or more processors to implement the method described in any of the implementations of the first aspect above.
In a fourth aspect, some embodiments of the present disclosure provide a computer-readable medium having a computer program stored thereon. Wherein the program, when executed by a processor, implements the method described in any of the implementations of the first aspect above.
The above embodiments of the present disclosure have the following advantages: by the article identification method of some embodiments of the present disclosure, when a weighing pan fails during weighing, article identification may be triggered even if the weight of the article is not read. The application scene of article identification is widened. Specifically, the reason why the article information identification cannot be triggered to identify the article is that: by directly weighing the articles, the article information identification is triggered according to the change of the weight, and when a weighing disc fails in the weighing process, the weight of the articles cannot be read, so that the article identification cannot be triggered. Based on this, in the article identification method of some embodiments of the present disclosure, a screenshot set of an article information display page in a preset period of time is first obtained. The page screenshot in the page screenshot set comprises a weight display area. Thus, various screenshots of the item information display page in the preset time period can be obtained. And secondly, generating a weight information display area image as a display area image according to the screenshot set. Thus, an image in which weight information is displayed can be obtained. And then, inputting the display area image into a pre-trained optical character recognition model to obtain weight recognition information in the display area image. Thus, the weight-related information displayed in the display area image can be recognized by the optical character recognition model. Then, in response to determining that the weight identification information includes a weight identification value and the weight identification value is greater than a preset value, performing the steps of: first, it is determined that an item identification triggering condition is satisfied. Wherein the article carrying tray corresponds to the article information display page. And a second step of acquiring an article carrying tray image of the article carrying tray as a first article carrying tray image. And thirdly, carrying out article identification on the first article carrying tray image to obtain article identification information. Therefore, when the identified weight identification information comprises the weight identification value which is larger than the preset value, the article identification can be triggered, and the article identification information can be obtained. And then, in response to determining that the weight identification information is empty, acquiring an article carrying tray image of the article carrying tray as a second article carrying tray image. Thus, when the weight identification information is empty, an image of the article carrying tray can be acquired. And then inputting the second object bearing disc image into a pre-trained empty disc detection model to obtain an empty disc identification result. The empty tray identification result represents whether the article bearing tray bears articles or not. Thereby, whether or not the article is carried in the article carrying tray can be identified by the empty tray detection model. And then, responding to the fact that the empty disc identification result represents that the article is carried in the article carrying disc, and determining that the article identification triggering condition is met. Thereby, the article identification may be triggered upon determining that an article is carried in the article carrying tray. Then, in response to determining that the item identification triggering condition is satisfied. And carrying out article identification on the second article carrying tray image to obtain article identification information. Therefore, the acquired image of the article carrying tray can be subjected to article identification, and article identification information can be obtained. Also, when the weight identification information is empty, whether the article is loaded in the article carrying tray can be identified by the image of the article carrying tray, so that the article identification can be continued when it is determined that the article is loaded in the article carrying tray. Therefore, when the weighing tray fails in the weighing process, the article identification can be triggered even if the weight of the article cannot be read. Thereby widening the application scene of article identification.
Drawings
The above and other features, advantages, and aspects of embodiments of the present disclosure will become more apparent by reference to the following detailed description when taken in conjunction with the accompanying drawings. The same or similar reference numbers will be used throughout the drawings to refer to the same or like elements. It should be understood that the figures are schematic and that elements and components are not necessarily drawn to scale.
FIG. 1 is a flow chart of some embodiments of an item identification method according to the present disclosure;
FIG. 2 is a schematic structural view of some embodiments of an article identification device according to the present disclosure;
fig. 3 is a schematic structural diagram of an electronic device suitable for use in implementing some embodiments of the present disclosure.
Detailed Description
Embodiments of the present disclosure will be described in more detail below with reference to the accompanying drawings. While certain embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be embodied in various forms and should not be construed as limited to the embodiments set forth herein. Rather, these embodiments are provided so that this disclosure will be thorough and complete. It should be understood that the drawings and embodiments of the present disclosure are for illustration purposes only and are not intended to limit the scope of the present disclosure.
It should be noted that, for convenience of description, only the portions related to the present invention are shown in the drawings. Embodiments of the present disclosure and features of embodiments may be combined with each other without conflict.
It should be noted that the terms "first," "second," and the like in this disclosure are merely used to distinguish between different devices, modules, or units and are not used to define an order or interdependence of functions performed by the devices, modules, or units.
It should be noted that references to "one", "a plurality" and "a plurality" in this disclosure are intended to be illustrative rather than limiting, and those of ordinary skill in the art will appreciate that "one or more" is intended to be understood as "one or more" unless the context clearly indicates otherwise.
The names of messages or information interacted between the various devices in the embodiments of the present disclosure are for illustrative purposes only and are not intended to limit the scope of such messages or information.
The present disclosure will be described in detail below with reference to the accompanying drawings in conjunction with embodiments.
Fig. 1 illustrates a flow 100 of some embodiments of an item identification method according to the present disclosure. The article identification method comprises the following steps:
Step 101, acquiring a screenshot set of an item information display page in a preset time period.
In some embodiments, an executing body (e.g., computing device) of the item identification method may obtain a set of page shots. The page screen shot may be an image including a weight display area. The article information display page may be a page for displaying article information of an article. The page shots in the page shot set may be arranged in ascending order according to the interception time of intercepting the page image. In practice, the execution body may intercept each screenshot of the item information display page within a preset period of time as a screenshot set.
Step 102, generating a weight display area image as a display area image according to the screenshot set.
In some embodiments, the execution subject may generate the weight display area image as a display area image according to the screenshot set.
In some optional implementations of some embodiments, the executing entity may generate the weight display area image as the display area image according to the set of page shots by:
The first step, gray processing is carried out on each page screenshot in the page screenshot set to obtain a page gray screenshot set. The page gray screen shot in the page gray screen shot set can be an image with only one gray value for each pixel point in the page gray image. The gray values may be represented by integers ranging from 0 to 255, 0 representing black, 255 representing white, and the middle value representing different gray levels.
And secondly, dividing the page gray screenshot set according to a sliding window mode to obtain a page gray screenshot group set. The page gray screen capture group in the page gray screen capture group set comprises a first page gray screen capture and a second page gray screen capture. For example, the first page gray level screenshot may be an image in the page gray level screenshot group, which has a screenshot time earlier than another page gray level screenshot. The second page gray screen shot is an image with the screen shot time later than that of the other page gray screen shot in the gray screen shot group. As an example, the set of page grayscale shots may be (shots a, b, c). The time interval of each screenshot is 1, the size of a time window is set to be 2, the sliding time length is 1, the initial window position is 0, and after division, a page gray screenshot group set ((screenshot a, screenshot b), (screenshot b, screenshot c)) is obtained.
Third, for each page gray screen shot group in the page gray screen shot group set, executing the following steps:
and a first sub-step of determining the pixel value of each pixel point in the first page gray level screenshot included in the page gray level screenshot group as a first pixel value set.
And a second sub-step of determining the pixel value of each pixel point in the second page gray screenshot included in the page gray screenshot group as a second pixel value set.
And a third sub-step of determining a pixel difference average value according to the first pixel value set and the second pixel value set. In practice, the difference between each first pixel value in the first pixel value set and each second pixel value in the second pixel value set is taken as an absolute value, so as to obtain a pixel difference value set. And averaging the pixel difference values in the pixel difference value set to obtain a pixel difference average value. For example, the first set of pixel values may be (150, 160, 170, 180), the second set of pixel values may be (50, 60, 70, 80), the set of pixel differences may be (100, 100, 100, 100), and the average value of pixel differences may be 100.
And a fourth step of determining each of the determined pixel difference averages as a set of pixel difference averages.
And fifthly, selecting the pixel difference average value meeting the preset numerical condition from the pixel difference average value set as a target pixel difference average value. The preset numerical condition may be a maximum pixel difference average value in the pixel difference average value set.
And sixthly, determining the page gray screenshot group corresponding to the average value of the target pixel differences as a target page gray screenshot group.
And seventh, generating a difference image according to the first page gray screen shot and the second page gray screen shot included in the target page gray screen shot group. In practice, the executing body may perform differential processing on the first page gray level screenshot and the second page gray level screenshot included in the target page gray level screenshot group, so as to generate a differential graph.
And eighth, determining the pixel value of each pixel point in the difference graph as a pixel value set.
And a ninth step of determining the position information of each pixel point corresponding to the pixel value set as a change area pixel point position information set in response to determining that each pixel value in the pixel value set is larger than a preset threshold value. For example, the preset threshold may be 50.
And tenth, determining a weight display area in the second page gray screen shot corresponding to the target page gray screen shot group according to the pixel point position information set of the change area. The change region pixel point position information in the change region pixel point position information set may be a coordinate value in a two-dimensional coordinate system. In practice, a two-dimensional coordinate system can be established in the second page gray screen shot, and polygon area extraction processing is carried out on the second page gray screen shot according to the coordinates of each pixel point, so that a weight display area is obtained.
Eleventh, generating a weight display area image based on the determined weight display area. In practice, the execution subject may generate the weight display area image by a template matching image processing technique based on the determined weight display area.
In some optional implementations of some embodiments, the executing body may perform gray processing on each screenshot in the set of page shots to obtain a set of page gray shots by:
the first step, for each page screenshot in the page screenshot set, executing the following steps:
and a first sub-step, determining each pixel point in the page screenshot as a pixel point set.
And a second sub-step of executing the following pixel processing step for each pixel in the pixel set.
And a first substep, determining the numerical values of the red, green and blue color channels of the pixel point as a color value set.
And a second sub-step of generating a color average value as a gray value according to each color value in the color value set. The gray value may be a color value corresponding to a channel.
And a third sub-step of updating the pixel value of the pixel point to the gray value.
And a third sub-step, determining the page screenshot of each pixel point processed by the pixel point processing step as the page gray screenshot.
And secondly, determining each determined page gray screenshot as a page gray screenshot set.
In some optional implementations of some embodiments, the executing entity may generate the difference map according to a first page gray level screenshot and a second page gray level screenshot included in the target page gray level screenshot group by:
and determining the pixel value of each pixel point in the second page gray screen shot as a second page gray screen shot pixel value set.
The second step, for each pixel point in the first page gray level screenshot, executing the following steps:
the first substep, the pixel value of the pixel point is determined as the first page gray level screenshot pixel value.
And a second sub-step of generating a target pixel value according to the pixel value corresponding to the first page gray screenshot pixel value in the first page gray screenshot pixel value and the second page gray screenshot pixel value set. In practice, first, the executing body may determine, as the initial target pixel value, a difference between the first page gray-scale screenshot pixel value and a second page gray-scale screenshot pixel value in the corresponding second page gray-scale screenshot pixel value set. The execution body may take the initial target pixel value as an absolute value of the target pixel value. As an example, the first page gray screen shot pixel value may be 60, and the second page gray screen shot pixel value in the corresponding second page gray screen shot pixel value set may be 40, then the initial target pixel value may be 20, and the target pixel value may be 20.
And a third sub-step of updating the pixel value of the pixel point to the target pixel value to obtain an updated pixel point. As an example, the pixel value of the updated pixel point may be 20.
And thirdly, generating a difference map according to each updated pixel point. In practice, the updated pixel points can be processed through digital images to generate a difference map.
And step 103, inputting the display area image into a pre-trained optical character recognition model to obtain weight recognition information in the display area image.
In some embodiments, the execution subject may input the display area image into a pre-trained optical character recognition model to obtain the weight recognition information in the display area image. The optical character recognition model may be a neural network model using a display area image as input data and weight recognition information as output data. For example, the neural network model may be an SVM model, a CNN model.
Alternatively, the optical character recognition model may be trained by:
first, a sample set is obtained. The samples in the sample set comprise a sample weight display area screenshot and sample target weight value information corresponding to the sample weight display area screenshot.
Second, the following training steps are performed based on the sample set:
and a first sub-step of inputting a screenshot of a sample weight display area of at least one sample in the sample set to an initial neural network to obtain sample identification weight numerical value information corresponding to each sample in the at least one sample.
And a second sub-step of comparing the sample identification weight value information corresponding to each sample in the at least one sample with the corresponding sample target weight value information to obtain a loss average value. In practice, the execution subject may determine a gap between the sample identification weight value information corresponding to each sample in the at least one sample and the corresponding sample target weight value information by comparing the cross entropy loss function.
And a third sub-step of optimizing parameters of the initial neural network based on the loss average value and a preset optimizer. The preset optimizer may be an Adam optimizer.
And a fourth sub-step, responding to the current training times as preset training times, and taking the initial neural network with optimized parameters as an optical character recognition model with completed training.
And a fifth sub-step of taking the initial neural network with optimized parameters as the initial neural network in response to the current training times being smaller than the preset training times, and forming a sample set by using unused samples, and executing the training steps again. As an example, the network parameters of the initial neural network described above may be adjusted using a back propagation algorithm (Back Propagation Algorithm, BP algorithm) and a gradient descent method (e.g., a small batch gradient descent algorithm).
Step 104, in response to determining that the weight identification information includes a weight identification value and the weight identification value is greater than a preset value, performing the steps of:
step 1041, determining that an item identification triggering condition is satisfied. Wherein, article loading tray corresponds article information display page.
In some embodiments, the executing entity may determine that the item identification triggering condition is satisfied. Wherein, the article identification triggering condition may be that the article carrying tray is not empty.
Step 1042, an article carrier tray image of the article carrier tray is acquired as a first article carrier tray image.
In some embodiments, the executing body may acquire the article carrying tray image of the article carrying tray as the first article carrying tray image. In practice, the executing body may acquire the image of the article carrying tray acquired by the camera as the first image of the article carrying tray.
Step 1043, performing article identification on the first article carrying tray image to obtain article identification information.
In some embodiments, the executing body may perform article identification on the first article carrying tray image to obtain article identification information. The article identification information may include article code information, article name, and article value.
In step 105, in response to determining that the weight identification information is empty, an article carrying tray image of the article carrying tray is acquired as a second article carrying tray image.
In some embodiments, the executing body may acquire the article carrier tray image of the article carrier tray as the second article carrier tray image in response to determining that the weight identification information is empty. In practice, the executing body may acquire the image of the article carrying tray acquired by the camera as the second image of the article carrying tray.
And 106, inputting the image of the second object bearing disc into a pre-trained empty disc detection model to obtain an empty disc identification result.
In some embodiments, the executing body may input the second article-carrying tray image into a pre-trained empty tray detection model, to obtain an empty tray recognition result. The empty tray identification result can represent whether the article bearing tray bears articles or not.
Alternatively, the empty disc detection model may be trained by:
first, a first sample set is obtained. The first sample in the first sample set includes a sample article carrying tray image, a real background removing article carrying tray image corresponding to the sample article carrying tray image, and sample target label information representing whether the carrying tray corresponding to the sample article carrying tray image is empty.
Second, the following empty detection model training steps are performed based on the first sample set:
and a first sub-step of inputting sample article carrying tray images of at least one first sample in the first sample set to a first initial neural network to obtain sample prediction background article carrying tray images corresponding to each second sample in the at least one first sample.
And a third sub-step of comparing the sample prediction background-removed object carrying tray image corresponding to each first sample in the at least one first sample with the corresponding real background-removed object carrying tray image to obtain a first comparison result. In practice, the execution subject may compare through the cross entropy loss function to determine a gap between the sample predicted de-background article carrying tray image and the corresponding real de-background article carrying tray image.
A fourth sub-step of updating, for each sample article-carrying tray image in the at least one first sample, the sample article-carrying tray image in the first sample set to a sample predictive background article-carrying tray image corresponding to the sample article-carrying tray image.
And a fifth sub-step of determining the updated first sample set as the second sample set.
And a sixth sub-step of inputting a sample background-removed object carrying disc image of at least one second sample in the second sample set to a second initial neural network to obtain sample prediction label information representing whether a carrying disc corresponding to the sample background-removed object carrying disc image of each second sample in the at least one second sample is empty or not.
And a seventh sub-step of comparing the sample prediction tag information corresponding to each second sample in the at least one second sample with the corresponding sample target tag information to obtain a second comparison result. In practice, the execution subject may determine a gap between the sample prediction tag information and the corresponding sample target tag information by comparing the cross entropy loss function.
And an eighth sub-step of determining whether the first initial neural network and the second initial neural network reach a preset optimization target or not according to the first comparison result, the second comparison result and a preset weight. Wherein the optimization objective may be to minimize the loss function and maximize the likelihood function.
And a ninth sub-step, in response to determining that the first initial neural network and the second initial neural network reach the optimization objective, using the first initial neural network and the second initial neural network as the trained optical character recognition model.
And a tenth sub-step of adjusting network parameters of the first and second initial neural networks in response to determining that the first and second initial neural networks do not meet the optimization objective, and forming a first sample set using the unused first samples, and performing the empty detection model training step again using the adjusted first and second initial neural networks. As an example, the network parameters of the first and second initial neural networks described above may be adjusted using a back propagation algorithm (Back Propagation Algorithm, BP algorithm).
And step 107, determining that the article identification triggering condition is met in response to determining that the empty tray identification result represents that the article is loaded in the article loading tray.
In some embodiments, the executing entity may determine that the article identification triggering condition is met in response to determining that the empty tray identification result indicates that the article is carried in the article carrying tray.
And step 108, in response to the condition that the article identification triggering condition is met, carrying out article identification on the second article carrying tray image to obtain article identification information.
In some embodiments, the executing body may perform article identification on the second article-carrying tray image in response to meeting an article identification triggering condition, to obtain article identification information.
In some optional implementations of some embodiments, the executing entity may perform the article identification on the second article-carrying tray image to obtain the article identification information in response to determining that the article identification triggering condition is satisfied.
The method comprises the steps of firstly, obtaining an article picture set of a preset article circulation source as a target picture set. The target picture set comprises target pictures of objects corresponding to the second object carrying disc image.
And secondly, acquiring the article codes of the articles corresponding to the target picture set to obtain an article code set. Wherein the article codes can be article numbers, self-codes and mnemonics.
And thirdly, acquiring article information groups of all articles corresponding to the target picture set to obtain an article information group set. Wherein, the article information in the article information group can comprise an article name and an article value.
And step four, generating an article data set according to the article coding set and the article information set. Wherein the article data in the article data set comprises an article code and an article information group.
And fifthly, inputting the second article carrying tray image into a pre-trained image extraction model to obtain an article image in the second article carrying tray image. The image extraction model may be a neural network model using the second object carrier tray image as input data and using the object image as output data. For example, the neural network model may be a LeNet model, a CNN model.
And sixthly, inputting the object image into a pre-trained code recognition model to obtain an object code as a target object code.
Seventh, item data corresponding to the target item code in the item data set is determined as item identification information.
Alternatively, the execution body may control the associated printing apparatus to print the obtained article information onto the label paper. The associated robotic arm may then be controlled to apply the label paper to the articles carried in the article carrying tray.
Alternatively, the code recognition model may be trained by:
in the first step, a third sample set is obtained. Wherein the third sample in the third sample set includes a sample item image and a sample target item code corresponding to the sample item image.
Second, for the third sample in the third sample set, performing the following coding recognition model training steps:
and a first sub-step of inputting the sample object image in the third sample to an input layer of the code identification network to obtain initial identification information. The code identification network comprises the input layer, a first feature extraction network layer, a second feature extraction network layer, a third feature extraction network layer, a feature splicing prediction layer and an output layer. The input layer may be used for inputting data. The first feature extraction network layer described above may be used for feature extraction. The second feature extraction network layer may be used to convert linear features into nonlinear features. The third feature extraction network layer may be used for feature dimension reduction. The feature stitching prediction layer described above may be used to combine features to determine a final predicted feature vector. The output layer may be configured to convert the prediction feature vector into a prediction result and output the prediction result.
And a second sub-step of inputting the initial identification information into the first feature extraction network layer to obtain initial first feature identification information. Wherein, the first feature extraction network layer may be a convolution layer. The first feature identification information may be an initial feature map of the article.
And a third sub-step of inputting the initial first feature identification information into the second feature extraction network layer to obtain second feature identification information. Wherein, the second feature extraction network layer may be an activation layer. The second feature identification information may be a nonlinear feature map.
And a fourth sub-step of inputting the second feature identification information into the third feature extraction network layer to obtain third feature identification information. Wherein, the third feature extraction network layer may be a pooling layer. The third feature identification information may be a target feature map that reduces the dimension of the nonlinear feature map.
And a fifth sub-step of performing feature vector conversion processing on the third feature identification information to obtain a feature vector. In practice, the execution body may perform feature vector conversion processing through a flattening operation, converting the third feature identification information into a feature vector.
And a sixth substep, inputting the feature vector into the feature stitching prediction layer to obtain a prediction target feature vector. The feature stitching prediction layer may be a full connection layer.
And a seventh substep, inputting the prediction target feature vector into the output layer to obtain sample prediction coding information.
And an eighth sub-step of comparing the sample predictive coding information with a sample target object code included in the third sample. In practice, the execution subject may determine a gap between the sample predictive coding information and the corresponding sample target object code by comparing the cross entropy loss function.
And a ninth substep, determining whether the code recognition network reaches a preset optimization target according to the comparison result.
And a tenth substep, in response to determining that the code recognition network achieves the optimization objective, using the initial neural network as a trained code recognition model.
And an eleventh sub-step of adjusting network parameters of the initial neural network in response to determining that the code recognition network does not meet the above-described optimization objective, and forming a third sample set using unused third samples, and performing the above-described code recognition model training step again using the adjusted code recognition network as the code recognition network. As an example, the network parameters of the code identification network described above may be adjusted using a back propagation algorithm (Back Propagation Algorithm, BP algorithm) and a Gradient descent method (Gradient descent).
The technical scheme and the related content are taken as an invention point of the embodiment of the disclosure, and the second technical problem mentioned in the background art is solved, namely the object information is identified by directly acquiring the image of the bearing disc, the background of the image of the bearing disc is interfered, and the accuracy of the identified object information is lower. For the label paper with wrong printed article information, the article information needs to be recognized again, so that the label paper is wasted. The factors that lead to label paper waste are often as follows: by carrying out article information identification on the directly acquired bearing disc image, the interference of the bearing disc image background is carried out, and the accuracy of the identified article information is lower. For the label paper with wrong printed article information, the article information needs to be recognized again, so that the label paper is wasted. If the above factors are solved, the efficiency of article identification can be achieved. To achieve this effect, the present disclosure is made by the steps of: firstly, acquiring an article picture set of a preset article circulation source as a target picture set. The target picture set comprises target pictures of objects corresponding to the second object carrying disc image. Thus, the article picture set of the article circulation source can be obtained. And secondly, acquiring the object codes of the objects corresponding to the target picture set to obtain an object code set. Thus, a set of item codes can be obtained. And then, acquiring the article information sets of the articles corresponding to the target picture set to obtain an article information set. Thus, a set of item information groups can be obtained. And then, generating an article data set according to the article coding set and the article information group set. Wherein the article data in the article data set comprises an article code and an article information group. Thus, the article data set can be obtained from the article code set and the article information set. And then inputting the second article carrying tray image into a pre-trained image extraction model to obtain an article image in the second article carrying tray image. Therefore, the object image with the background of the bearing disc removed can be obtained through the pre-trained image extraction model. Then, the article image is inputted into a pre-trained code recognition model, and an article code is obtained as a target article code. Thus, the article code corresponding to the article is predicted from the article image by the pre-trained code recognition model. And finally, determining the article data corresponding to the target article codes in the article data set as article identification information. Thus, the object item code can be searched in the data set to obtain item information comprising the item code. And because the image extraction model is adopted, the interference of the image background of the bearing disc is removed, the accuracy of identifying the article information is improved, the label paper with wrong printed article information is reduced, and the label paper waste is further reduced.
The above embodiments of the present disclosure have the following advantages: by the article identification method of some embodiments of the present disclosure, when a weighing pan fails during weighing, article identification may be triggered even if the weight of the article is not read. The application scene of article identification is widened. Specifically, the reason why the article information identification cannot be triggered to identify the article is that: the article information identification is triggered according to the change of the weight, and the weighing tray may malfunction due to the weighing process. Moreover, the identification of the item information may occur such that the weight of the item is not readable, resulting in an inability to trigger the identification of the item information to identify the item. Based on this, in the article identification method of some embodiments of the present disclosure, first, a screenshot set of an article information display page in a preset period of time is obtained. The page screenshot in the page screenshot set comprises a weight display area. Thus, various screenshots of the item information display page in the preset time period can be obtained. And secondly, generating a weight information display area image as a display area image according to the screenshot set. Thus, an image in which weight information is displayed can be obtained. Then, the display area image is inputted into a pre-trained optical character recognition model, and weight recognition information in the display area image is obtained. Thus, the weight-related information displayed in the display area image can be recognized by the optical character recognition model. Then, in response to determining that the weight identification information includes a weight identification value and the weight identification value is greater than a preset value, performing the steps of: first, it is determined that an item identification triggering condition is satisfied. Wherein the article carrying tray corresponds to the article information display page. And a second step of acquiring an article carrying tray image of the article carrying tray as a first article carrying tray image. And thirdly, carrying out article identification on the first article carrying tray image to obtain article identification information. Therefore, when the identified weight identification information comprises the weight identification value which is larger than the preset value, the article identification can be triggered, and the article identification information can be obtained. Then, in response to determining that the weight identification information is empty, an article carrier tray image of the article carrier tray is acquired as a second article carrier tray image. Thus, when the weight identification information is empty, an image of the article carrying tray can be acquired. And then inputting the second object bearing disc image into a pre-trained empty disc detection model to obtain an empty disc identification result. The empty tray identification result represents whether the article bearing tray bears articles or not. Thereby, whether or not the article is carried in the article carrying tray can be identified by the empty tray detection model. And then, responding to the fact that the empty disc identification result represents that the article is carried in the article carrying disc, and determining that the article identification triggering condition is met. Thereby, the article identification may be triggered upon determining that an article is carried in the article carrying tray. Then, in response to determining that the item identification triggering condition is satisfied. And carrying out article identification on the second article carrying tray image to obtain article identification information. Therefore, the acquired image of the article carrying tray can be subjected to article identification, and article identification information can be obtained. Also, when the weight identification information is empty, whether the article is loaded in the article carrying tray can be identified by the image of the article carrying tray, so that the article identification can be continued when it is determined that the article is loaded in the article carrying tray. Therefore, when the weighing tray fails during the weighing process, the identification of the article can be triggered even if the weight of the article cannot be read. Thereby widening the application scene of article identification.
With further reference to fig. 2, as an implementation of the method shown in the figures, the present disclosure provides some embodiments of a web page generation apparatus, which correspond to those method embodiments shown in fig. 1, and which are particularly applicable in various electronic devices.
As shown in fig. 2, the web page generating apparatus 200 of some embodiments includes: a first acquisition unit 201, a generation unit 202, a first input unit 203, an execution unit 204, a second acquisition unit 205, a second input unit 206, a determination unit 207, and an identification unit 208. The first obtaining unit 201 is configured to obtain a screenshot set of an item information display page within a preset period of time. The page screenshot in the page screenshot set comprises a weight display area; a generating unit 202 configured to generate a weight display area image as a display area image from the above-described page screenshot set; a first input unit 203 configured to input the display area image into a pre-trained optical character recognition model, and obtain weight recognition information in the display area image; the execution unit 204 is configured to, in response to determining that the weight identification information includes a weight identification value, and that the weight identification value is greater than a preset value, execute the steps of: and determining that the article identification triggering condition is met. Wherein the article carrying tray corresponds to the article information display page; acquiring an article carrying tray image of the article carrying tray as a first article carrying tray image; carrying out article identification on the first article carrying tray image to obtain article identification information; a second acquiring unit 205 configured to acquire an article carrying tray image of the article carrying tray as a second article carrying tray image in response to determining that the weight identification information is empty; a second input unit 206 configured to input the second article. And inputting the bearing disc image into a pre-trained empty disc detection model to obtain an empty disc identification result. The empty disc identification result represents whether the article bearing disc bears articles or not; a determining unit 207 configured to determine that an item identification triggering condition is satisfied in response to determining that the empty tray identification result characterizes an item carried in the item carrying tray; and the identifying unit 208 is configured to identify the article on the second article carrying tray image to obtain article identification information in response to determining that the article identification triggering condition is met.
It will be appreciated that the elements described in the apparatus 200 correspond to the various steps in the method described with reference to fig. 1. Thus, the operations, features and resulting benefits described above for the method are equally applicable to the apparatus 200 and the units contained therein, and are not described in detail herein.
Referring now to fig. 3, a schematic diagram of an electronic device 300 suitable for use in implementing some embodiments of the present disclosure is shown. The electronic device shown in fig. 3 is merely an example and should not impose any limitations on the functionality and scope of use of embodiments of the present disclosure.
As shown in fig. 3, the electronic device 300 may include a processing means (e.g., a central processing unit, a graphics processor, etc.) 301 that may perform various suitable actions and processes in accordance with a program stored in a Read Only Memory (ROM) 302 or a program loaded from a storage means 308 into a Random Access Memory (RAM) 303. In the RAM 303, various programs and data required for the operation of the electronic apparatus 300 are also stored. The processing device 301, the ROM 302, and the RAM 303 are connected to each other via a bus 304. An input/output (I/O) interface 305 is also connected to bus 304.
In general, the following devices may be connected to the I/O interface 305: input devices 306 including, for example, a touch screen, touchpad, keyboard, mouse, camera, microphone, accelerometer, gyroscope, etc.; an output device 307 including, for example, a Liquid Crystal Display (LCD), a speaker, a vibrator, and the like; storage 308 including, for example, magnetic tape, hard disk, etc.; and communication means 309. The communication means 309 may allow the electronic device 300 to communicate with other devices wirelessly or by wire to exchange data. While fig. 3 shows an electronic device 300 having various means, it is to be understood that not all of the illustrated means are required to be implemented or provided. More or fewer devices may be implemented or provided instead. Each block shown in fig. 3 may represent one device or a plurality of devices as needed.
In particular, according to some embodiments of the present disclosure, the processes described above with reference to flowcharts may be implemented as computer software programs. For example, some embodiments of the present disclosure include a computer program product comprising a computer program embodied on a computer readable medium, the computer program comprising program code for performing the method shown in the flow chart. In such embodiments, the computer program may be downloaded and installed from a network via communications device 309, or from storage device 308, or from ROM 302. The computer program, when executed by the processing means 301, performs the functions defined in the methods of some embodiments of the present disclosure.
It should be noted that, the computer readable medium described in some embodiments of the present disclosure may be a computer readable signal medium or a computer readable storage medium, or any combination of the two. The computer readable storage medium can be, for example, but not limited to, an electronic, magnetic, optical, electromagnetic, infrared, or semiconductor system, apparatus, or device, or a combination of any of the foregoing. More specific examples of the computer-readable storage medium may include, but are not limited to: an electrical connection having one or more wires, a portable computer diskette, a hard disk, a Random Access Memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber, a portable compact disc read-only memory (CD-ROM), an optical storage device, a magnetic storage device, or any suitable combination of the foregoing. In some embodiments of the present disclosure, a computer readable storage medium may be any tangible medium that can contain, or store a program for use by or in connection with an instruction execution system, apparatus, or device. In some embodiments of the present disclosure, however, the computer-readable signal medium may comprise a data signal propagated in baseband or as part of a carrier wave, with the computer-readable program code embodied therein. Such a propagated data signal may take any of a variety of forms, including, but not limited to, electro-magnetic, optical, or any suitable combination of the foregoing. A computer readable signal medium may also be any computer readable medium that is not a computer readable storage medium and that can communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device. Program code embodied on a computer readable medium may be transmitted using any appropriate medium, including but not limited to: electrical wires, fiber optic cables, RF (radio frequency), and the like, or any suitable combination of the foregoing.
In some implementations, the clients, servers may communicate using any currently known or future developed network protocol, such as HTTP (HyperText Transfer Protocol ), and may be interconnected with any form or medium of digital data communication (e.g., a communication network). Examples of communication networks include a local area network ("LAN"), a wide area network ("WAN"), the internet (e.g., the internet), and peer-to-peer networks (e.g., ad hoc peer-to-peer networks), as well as any currently known or future developed networks.
The computer readable medium may be embodied in the electronic device described above; or may exist alone without being incorporated into the electronic device. The computer readable medium carries one or more programs which, when executed by the electronic device, cause the electronic device to: acquiring a screenshot set of an item information display page in a preset time period, wherein the screenshot in the screenshot set comprises a weight display area; generating a weight display area image as a display area image according to the page screenshot set; inputting the display area image into a pre-trained optical character recognition model to obtain weight recognition information in the display area image; in response to determining that the weight identification information includes a weight identification value and that the weight identification value is greater than a preset value, performing the steps of: determining that an article identification triggering condition is met, wherein the article carrying tray corresponds to the article information display page; acquiring an article carrying tray image of the article carrying tray as a first article carrying tray image; carrying out article identification on the first article carrying tray image to obtain first article identification information; acquiring an article carrying tray image of the article carrying tray as a second article carrying tray image in response to determining that the weight identification information is empty; inputting the second object bearing disc image into a pre-trained empty disc detection model to obtain an empty disc identification result, wherein the empty disc identification result represents whether an object is borne in the object bearing disc or not; responding to the fact that the empty disc identification result represents that articles are carried in the article carrying disc, and determining that the article identification triggering condition is met; and in response to determining that the article identification triggering condition is met, carrying out article identification on the second article carrying tray image to obtain second article identification information.
Computer program code for carrying out operations for some embodiments of the present disclosure may be written in one or more programming languages, including an object oriented programming language such as Java, smalltalk, C ++ and conventional procedural programming languages, such as the "C" programming language or similar programming languages. The program code may execute entirely on the user's computer, partly on the user's computer, as a stand-alone software package, partly on the user's computer and partly on a remote computer or entirely on the remote computer or server. In the case of a remote computer, the remote computer may be connected to the user's computer through any kind of network, including a Local Area Network (LAN) or a Wide Area Network (WAN), or may be connected to an external computer (for example, through the Internet using an Internet service provider).
The flowcharts and block diagrams in the figures illustrate the architecture, functionality, and operation of possible implementations of systems, methods and computer program products according to various embodiments of the present disclosure. In this regard, each block in the flowchart or block diagrams may represent a module, segment, or portion of code, which comprises one or more executable instructions for implementing the specified logical function(s). It should also be noted that, in some alternative implementations, the functions noted in the block may occur out of the order noted in the figures. For example, two blocks shown in succession may, in fact, be executed substantially concurrently, or the blocks may sometimes be executed in the reverse order, depending upon the functionality involved. It will also be noted that each block of the block diagrams and/or flowchart illustration, and combinations of blocks in the block diagrams and/or flowchart illustration, can be implemented by special purpose hardware-based systems which perform the specified functions or acts, or combinations of special purpose hardware and computer instructions.
The units described in some embodiments of the present disclosure may be implemented by means of software, or may be implemented by means of hardware. The described units may also be provided in a processor, for example, described as: a processor includes a first acquisition unit, a generation unit, a first input unit, an execution unit, a second acquisition unit, a second input unit, a determination unit, and an identification unit. The names of these units do not constitute a limitation on the unit itself in some cases, and for example, the first acquisition unit may also be described as "a unit that acquires a screenshot set of an item information display page within a preset period of time".
The functions described above herein may be performed, at least in part, by one or more hardware logic components. For example, without limitation, exemplary types of hardware logic components that may be used include: a Field Programmable Gate Array (FPGA), an Application Specific Integrated Circuit (ASIC), an Application Specific Standard Product (ASSP), a system on a chip (SOC), a Complex Programmable Logic Device (CPLD), and the like.
The foregoing description is only of the preferred embodiments of the present disclosure and description of the principles of the technology being employed. It will be appreciated by those skilled in the art that the scope of the invention in the embodiments of the present disclosure is not limited to the specific combination of the above technical features, but encompasses other technical features formed by any combination of the above technical features or their equivalents without departing from the spirit of the invention. Such as the above-described features, are mutually substituted with (but not limited to) the features having similar functions disclosed in the embodiments of the present disclosure.
Claims (10)
1. An article identification method comprising:
acquiring a screenshot set of an item information display page in a preset time period, wherein the screenshot in the screenshot set comprises a weight display area;
generating a weight display area image as a display area image according to the page screenshot set;
inputting the display area image into a pre-trained optical character recognition model to obtain weight recognition information in the display area image;
in response to determining that the weight identification information includes a weight identification value and the weight identification value is greater than a preset value, performing the steps of:
determining that an article identification triggering condition is met, wherein an article carrying disc corresponds to the article information display page;
acquiring an article carrying tray image of the article carrying tray as a first article carrying tray image;
carrying out article identification on the first article carrying tray image to obtain article identification information;
acquiring an article carrying tray image of the article carrying tray as a second article carrying tray image in response to determining that the weight identification information is empty;
inputting the second object bearing disc image into a pre-trained empty disc detection model to obtain an empty disc identification result, wherein the empty disc identification result represents whether an object is borne in the object bearing disc or not;
Determining that an article identification triggering condition is met in response to determining that the empty tray identification result represents that articles are loaded in the article loading tray;
and in response to determining that the article identification triggering condition is met, carrying out article identification on the second article carrying tray image to obtain article identification information.
2. The method of claim 1, wherein the generating a weight display area image as a display area image from the set of page shots comprises:
carrying out gray processing on each page screenshot in the page screenshot set to obtain a page gray screenshot set;
dividing the page gray screen shot set in a sliding window mode to obtain a page gray screen shot group set, wherein the page gray screen shot group in the page gray screen shot group set comprises a first page gray screen shot and a second page gray screen shot;
for each page gray screen shot group in the page gray screen shot group set, executing the following steps:
determining pixel values of all pixel points in a first page gray screenshot included in the page gray screenshot group as a first pixel value set;
determining the pixel value of each pixel point in a second page gray screenshot included in the page gray screenshot group as a second pixel value set;
Determining a pixel difference average value according to the first pixel value set and the second pixel value set;
determining each of the determined pixel difference averages as a set of pixel difference averages;
selecting a pixel difference average value meeting a preset numerical condition from the pixel difference average value set as a target pixel difference average value;
determining a page gray level screenshot group corresponding to the target pixel difference average value as a target page gray level screenshot group;
generating a difference image according to a first page gray screen shot and a second page gray screen shot which are included in the target page gray screen shot group;
determining the pixel value of each pixel point in the difference graph as a pixel value set;
determining the position information of each pixel point corresponding to the pixel value set as a change area pixel point position information set in response to determining that each pixel value in the pixel value set is larger than a preset threshold value;
determining a weight display area in a second page gray screen shot corresponding to the target page gray screen shot group according to the pixel point position information set of the change area;
and generating a weight display area image according to the determined weight display area.
3. The method of claim 2, wherein the gray processing each screenshot in the set of shots to obtain a set of shots comprises:
For each page screenshot in the set of page shots, performing the steps of:
determining each pixel point in the page screenshot as a pixel point set;
for each pixel in the set of pixels, performing the following pixel processing steps:
determining the numerical values of the red, green and blue color channels of the pixel point as a color value set;
generating a color average value as a gray value according to each color value in the color value set;
updating the pixel value of the pixel point to the gray value;
determining the page screenshot of each pixel point processed by the pixel point processing step as page gray screenshot;
and determining each determined page gray screenshot as a page gray screenshot set.
4. The method of claim 2, wherein the generating a difference graph from the first page gray level screenshot and the second page gray level screenshot included in the target page gray level screenshot group includes:
determining the pixel value of each pixel point in the second page gray screen shot as a second page gray screen shot pixel value set;
for each pixel point in the first page gray level screenshot, executing the following steps:
Determining the pixel value of the pixel point as a first page gray screenshot pixel value;
generating a target pixel value according to the pixel value corresponding to the first page gray screenshot pixel value in the first page gray screenshot pixel value and the second page gray screenshot pixel value set;
updating the pixel value of the pixel point to the target pixel value to obtain an updated pixel point;
and generating a difference map according to each updated pixel point.
5. The method of claim 1, wherein the optical character recognition model is trained by:
acquiring a sample set, wherein a sample in the sample set comprises a sample weight display area screenshot and sample target weight numerical value information corresponding to the sample weight display area screenshot;
the following training steps are performed based on the sample set:
inputting a screenshot of a sample weight display area of at least one sample in a sample set to an initial neural network to obtain sample identification weight numerical value information corresponding to each sample in the at least one sample;
comparing sample identification weight value information corresponding to each sample in the at least one sample with corresponding sample target weight value information to obtain a loss average value;
Performing parameter optimization on the initial neural network based on the loss average value and a preset optimizer;
responding to the current training times as preset training times, and taking the initial neural network with optimized parameters as an optical character recognition model with completed training;
and in response to the current training times being smaller than the preset training times, taking the initial neural network with the optimized parameters as the initial neural network, and forming a sample set by using unused samples, and executing the training steps again.
6. The method of claim 1, wherein the empty disc detection model is trained by:
acquiring a first sample set, wherein the first sample in the first sample set comprises a sample article carrying disc image, a real background-removed article carrying disc image corresponding to the sample article carrying disc image and sample target label information representing whether a carrying disc corresponding to the sample article carrying disc image is empty or not;
the following empty detection model training steps are performed based on the first sample set:
inputting a sample article carrying tray image of at least one first sample in a first sample set to a first initial neural network to obtain a sample prediction background article carrying tray image corresponding to each first sample in the at least one first sample;
Comparing a sample prediction background-removed object carrying tray image corresponding to each first sample in the at least one first sample with the corresponding real background-removed object carrying tray image to obtain a first comparison result;
for each sample article-carrying tray image in the at least one first sample, updating the sample article-carrying tray image in the first sample set to a sample predictive background article-carrying tray image corresponding to the sample article-carrying tray image;
determining the updated first sample set as a second sample set;
inputting sample background-removed object carrying disc images of at least one second sample in a second sample set into a second initial neural network to obtain sample prediction label information representing whether carrying discs corresponding to the sample background-removed object carrying disc images of each second sample in the at least one second sample are empty or not;
comparing sample prediction label information corresponding to each second sample in the at least one second sample with corresponding sample target label information to obtain a second comparison result;
determining whether the first initial neural network and the second initial neural network reach a preset optimization target or not according to the first comparison result, the second comparison result and a preset weight;
In response to determining that the first initial neural network and the second initial neural network meet the optimization target, taking the first initial neural network and the second initial neural network as optical character recognition models after training;
and in response to determining that the first initial neural network and the second initial neural network do not meet the optimization target, adjusting network parameters of the first initial neural network and the second initial neural network, forming a first sample set by using unused first samples, and executing the empty disc detection model training step again by using the adjusted first initial neural network and second initial neural network.
7. The method of claim 1, wherein the method further comprises:
controlling an associated printing device to print the obtained article information onto the label paper;
and controlling an associated mechanical arm to paste the label paper on the articles carried in the article carrying tray.
8. An article identification device comprising:
the first acquisition unit is configured to acquire a screenshot set of an article information display page in a preset time period, wherein the screenshot in the screenshot set comprises a weight display area;
a generation unit configured to generate a weight display area image as a display area image from the set of page shots;
A first input unit configured to input the display area image to a pre-trained optical character recognition model, to obtain weight recognition information in the display area image;
an execution unit configured to, in response to determining that the weight identification information includes a weight identification value, and the weight identification value is greater than a preset value, execute the steps of: determining that an article identification triggering condition is met, wherein the article carrying tray corresponds to the article information display page; acquiring an article carrying tray image of the article carrying tray as a first article carrying tray image; carrying out article identification on the first article carrying tray image to obtain article identification information;
a second acquisition unit configured to acquire an article-carrying tray image of the article-carrying tray as a second article-carrying tray image in response to determining that the weight-identifying information is empty;
the second input unit is configured to input the second article carrying tray image into a pre-trained empty tray detection model to obtain an empty tray identification result, wherein the empty tray identification result represents whether articles are carried in the article carrying tray or not;
a determining unit configured to determine that an article identification triggering condition is satisfied in response to determining that the empty tray identification result characterizes that an article is carried in the article carrying tray;
And the identification unit is configured to identify the article from the second article carrying tray image to obtain article identification information in response to determining that the article identification triggering condition is met.
9. An electronic device, comprising:
one or more processors;
a storage device having one or more programs stored thereon;
when executed by the one or more processors, causes the one or more processors to implement the method of any of claims 1 to 7.
10. A computer readable medium having stored thereon a computer program, wherein the program when executed by a processor implements the method of any of claims 1 to 7.
Priority Applications (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310512053.0A CN116597430A (en) | 2023-05-05 | 2023-05-05 | Article identification method, apparatus, electronic device, and computer-readable medium |
Applications Claiming Priority (1)
Application Number | Priority Date | Filing Date | Title |
---|---|---|---|
CN202310512053.0A CN116597430A (en) | 2023-05-05 | 2023-05-05 | Article identification method, apparatus, electronic device, and computer-readable medium |
Publications (1)
Publication Number | Publication Date |
---|---|
CN116597430A true CN116597430A (en) | 2023-08-15 |
Family
ID=87598446
Family Applications (1)
Application Number | Title | Priority Date | Filing Date |
---|---|---|---|
CN202310512053.0A Pending CN116597430A (en) | 2023-05-05 | 2023-05-05 | Article identification method, apparatus, electronic device, and computer-readable medium |
Country Status (1)
Country | Link |
---|---|
CN (1) | CN116597430A (en) |
Cited By (1)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117351004A (en) * | 2023-11-29 | 2024-01-05 | 杭州天眼智联科技有限公司 | Regenerated material identification method, apparatus, electronic device and computer readable medium |
-
2023
- 2023-05-05 CN CN202310512053.0A patent/CN116597430A/en active Pending
Cited By (2)
Publication number | Priority date | Publication date | Assignee | Title |
---|---|---|---|---|
CN117351004A (en) * | 2023-11-29 | 2024-01-05 | 杭州天眼智联科技有限公司 | Regenerated material identification method, apparatus, electronic device and computer readable medium |
CN117351004B (en) * | 2023-11-29 | 2024-02-20 | 杭州天眼智联科技有限公司 | Regenerated material identification method, apparatus, electronic device and computer readable medium |
Similar Documents
Publication | Publication Date | Title |
---|---|---|
CN107578017B (en) | Method and apparatus for generating image | |
CN111476871B (en) | Method and device for generating video | |
CN109344752B (en) | Method and apparatus for processing mouth image | |
CN107609506B (en) | Method and apparatus for generating image | |
CN108230346B (en) | Method and device for segmenting semantic features of image and electronic equipment | |
CN108235116B (en) | Feature propagation method and apparatus, electronic device, and medium | |
US11514263B2 (en) | Method and apparatus for processing image | |
CN109377508B (en) | Image processing method and device | |
US11164004B2 (en) | Keyframe scheduling method and apparatus, electronic device, program and medium | |
CN110288625B (en) | Method and apparatus for processing image | |
CN111080595A (en) | Image processing method, image processing device, electronic equipment and computer readable medium | |
CN110059623B (en) | Method and apparatus for generating information | |
CN111767750A (en) | Image processing method and device | |
CN116597430A (en) | Article identification method, apparatus, electronic device, and computer-readable medium | |
CN112800276A (en) | Video cover determination method, device, medium and equipment | |
US20220319157A1 (en) | Temporal augmentation for training video reasoning system | |
CN108921792B (en) | Method and device for processing pictures | |
CN111292333A (en) | Method and apparatus for segmenting an image | |
CN110633597A (en) | Driving region detection method and device | |
CN116506741A (en) | Method and device for removing flicker of monitoring video, electronic equipment and computer readable medium | |
CN110555799A (en) | Method and apparatus for processing video | |
CN111260756A (en) | Method and apparatus for transmitting information | |
CN113033552B (en) | Text recognition method and device and electronic equipment | |
CN111737575B (en) | Content distribution method, content distribution device, readable medium and electronic equipment | |
CN114120423A (en) | Face image detection method and device, electronic equipment and computer readable medium |
Legal Events
Date | Code | Title | Description |
---|---|---|---|
PB01 | Publication | ||
PB01 | Publication | ||
SE01 | Entry into force of request for substantive examination | ||
SE01 | Entry into force of request for substantive examination |