CN117115569A - Automatic object image identification and classification method and system based on machine learning - Google Patents
Automatic object image identification and classification method and system based on machine learning Download PDFInfo
- Publication number
- CN117115569A CN117115569A CN202311378761.6A CN202311378761A CN117115569A CN 117115569 A CN117115569 A CN 117115569A CN 202311378761 A CN202311378761 A CN 202311378761A CN 117115569 A CN117115569 A CN 117115569A
- Authority
- CN
- China
- Prior art keywords
- object image
- image
- article
- preset
- identifiable
- 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.)
- Granted
Links
- 238000000034 method Methods 0.000 title claims abstract description 43
- 238000010801 machine learning Methods 0.000 title claims abstract description 35
- 238000012545 processing Methods 0.000 claims abstract description 22
- 238000012549 training Methods 0.000 claims abstract description 15
- 238000003384 imaging method Methods 0.000 claims description 14
- 238000013507 mapping Methods 0.000 claims description 12
- 238000012216 screening Methods 0.000 claims description 10
- 238000001914 filtration Methods 0.000 claims description 8
- 238000010276 construction Methods 0.000 claims description 5
- 238000012163 sequencing technique Methods 0.000 claims description 5
- 238000006243 chemical reaction Methods 0.000 claims description 4
- 230000006978 adaptation Effects 0.000 claims description 3
- 238000002955 isolation Methods 0.000 claims description 3
- 230000000694 effects Effects 0.000 abstract description 11
- 239000000463 material Substances 0.000 abstract description 9
- 230000009286 beneficial effect Effects 0.000 description 11
- 230000008859 change Effects 0.000 description 2
- 239000003550 marker Substances 0.000 description 2
- 238000012986 modification Methods 0.000 description 2
- 230000004048 modification Effects 0.000 description 2
- 230000002159 abnormal effect Effects 0.000 description 1
- 230000004075 alteration Effects 0.000 description 1
- 238000004364 calculation method Methods 0.000 description 1
- 238000011161 development Methods 0.000 description 1
- 238000010586 diagram Methods 0.000 description 1
- 238000005516 engineering process Methods 0.000 description 1
- 230000000737 periodic effect Effects 0.000 description 1
- 230000008569 process Effects 0.000 description 1
- 230000003245 working effect Effects 0.000 description 1
Classifications
-
- 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
- G06V10/764—Arrangements for image or video recognition or understanding using pattern recognition or machine learning using classification, e.g. of video 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
- G06V10/40—Extraction of image or video features
- G06V10/44—Local feature extraction by analysis of parts of the pattern, e.g. by detecting edges, contours, loops, corners, strokes or intersections; Connectivity analysis, e.g. of connected components
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06V—IMAGE OR VIDEO RECOGNITION OR UNDERSTANDING
- G06V20/00—Scenes; Scene-specific elements
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Multimedia (AREA)
- General Physics & Mathematics (AREA)
- Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- General Health & Medical Sciences (AREA)
- Medical Informatics (AREA)
- Software Systems (AREA)
- Health & Medical Sciences (AREA)
- Databases & Information Systems (AREA)
- Computing Systems (AREA)
- Artificial Intelligence (AREA)
- Image Analysis (AREA)
Abstract
The invention provides a method and a system for automatically identifying and classifying object images based on machine learning, wherein the method comprises the following steps: acquiring an identifiable article image, performing learning training on the identifiable article image, extracting article characteristics of the identifiable article, and constructing an object image class gallery based on the article characteristics; image scanning is carried out on the object to be identified based on a cash register to obtain a target object image, the target object image is processed, and a preset object image matched with the target object image in an object image category gallery is determined based on a processing result; extracting attribute information of a preset object image, determining a class label, an article name and an article price of a target object image based on the attribute information, and displaying the class label, the article name and the article price of the article to be identified based on a preset display screen. The accuracy and the efficiency of identifying the articles to be identified are guaranteed, the classification accuracy of the articles to be identified is also guaranteed, and the identification and classification effects of the articles are also guaranteed while a large amount of manpower and material resources are saved.
Description
Technical Field
The invention relates to the technical field of data identification and image processing, in particular to an automatic object image identification and classification method and system based on machine learning.
Background
At present, along with the continuous development of science and technology, equipment used in life of people is more and more intelligent, and intelligent equipment provides convenience for people and improves working efficiency and working effect, wherein intelligent collection object image identification is one of intelligent equipment;
the automatic identification of the object image means that when a user settles the settlement of the object, the cash register can automatically photograph the current object, and then the photographed image is identified and analyzed, so that the category of the object, the corresponding object name, the object price and the like are obtained;
however, in the market at present, the cashier can acquire corresponding article information after manually scanning the bar code on the article through the scanning gun, once the types of the article are too many and the number of the purchasers is too many, the settlement progress of the user on the article can be seriously affected, and because the settlement is manually participated, the conditions of missing calculation, article classification error or repeated settlement are easily generated, a large amount of manpower and material resources are wasted, and the identification effect on the article is greatly reduced;
therefore, in order to overcome the above-mentioned drawbacks, the present invention provides a method and a system for automatically identifying and classifying object images based on machine learning.
Disclosure of Invention
The invention provides an automatic object image identification and classification method and system based on machine learning, which are used for realizing comprehensive and accurate construction of an object image type image library by learning and training identifiable object images, providing accurate reference basis for automatic object image identification, scanning objects to be identified through a cash register, analyzing the scanned target object images, realizing accurate and effective judgment of class labels, object names and object prices of the target object images through preset object images in the object image type image library, and finally displaying the obtained class labels, object names and object prices, thereby ensuring the accuracy and efficiency of object identification to be identified, ensuring the classification accuracy of objects to be identified, facilitating effective understanding of users on purchased commodities, saving a large amount of manpower and material resources and simultaneously ensuring the identification and classification effects of the objects.
The invention provides an object image automatic identification and classification method based on machine learning, which comprises the following steps:
step 1: acquiring an identifiable article image, performing learning training on the identifiable article image, extracting article characteristics of the identifiable article, and constructing an object image class gallery based on the article characteristics;
Step 2: image scanning is carried out on the object to be identified based on a cash register to obtain a target object image, the target object image is processed, and a preset object image matched with the target object image in an object image category gallery is determined based on a processing result;
step 3: extracting attribute information of a preset object image, determining a class label, an article name and an article price of a target object image based on the attribute information, and displaying the class label, the article name and the article price of the article to be identified based on a preset display screen.
Preferably, in step 1, an object image automatic identification and classification method based on machine learning, the method includes:
acquiring an identifiable object set and a preset imaging requirement, and adapting imaging parameters and shooting angles of a preset camera based on the preset imaging requirement;
performing multi-angle multi-frequency shooting on each identifiable object in the identifiable object set according to a preset camera based on an adaptation result to obtain an object image set of each identifiable object, and classifying object images in the object image set based on shooting angles;
obtaining a sub-article image set corresponding to each shooting angle based on the classification result, and performing de-duplication and screening on sub-article images in the sub-article image set based on a preset image screening index to obtain a standard sub-article image corresponding to each shooting angle;
And summarizing the standard sub-object images corresponding to all shooting angles of the same identifiable object, and obtaining the identifiable object image corresponding to the identifiable object based on the summarizing result.
Preferably, in step 1, learning training is performed on identifiable object images, and object features of the identifiable objects are extracted, including:
acquiring an acquired identifiable object image, traversing pixel points of the identifiable object image, and acquiring position information and pixel information of effective pixel points in the identifiable object image;
obtaining edge characteristics of the identifiable object image based on the position information and the pixel information of the effective pixel points, and dividing the identifiable object image based on the edge characteristics to obtain a sub-object image block;
extracting texture direction characteristics and color characteristics in each sub-article image block, and obtaining local characteristics of each sub-article image block based on the texture direction characteristics and the color characteristics;
filtering the obtained local features based on the edge features, obtaining target local features corresponding to the sub-article image blocks based on the filtering result, and splicing the target local features of different sub-article image blocks based on the relative position relationship among the sub-article image blocks to obtain the article features of the identifiable articles.
Preferably, in step 1, an object image classification gallery is constructed based on object features, which includes:
acquiring the obtained object characteristics, determining a class label corresponding to the identifiable object based on the object characteristics, and classifying the identifiable object image based on the class label to obtain an identifiable object class atlas;
extracting configuration information of a preset gallery and byte information of an identifiable object category atlas, dividing the preset gallery based on the configuration information and the byte information to obtain N independent storage spaces, and isolating the N independent storage spaces;
and respectively storing the identifiable object class atlas into corresponding independent storage spaces based on the isolation result, generating an image storage catalog based on the storage result and the class label, displaying the display identification position of the image storage catalog in a preset gallery, and obtaining the object image class gallery based on the display result.
Preferably, an automatic object image recognition and classification method based on machine learning obtains an object image class gallery based on a display result, including:
traversing the identifiable objects based on a preset time interval, obtaining a current identifiable object set based on a traversing result, and analyzing the current identifiable object set to obtain object type information corresponding to the current identifiable object set;
Comparing the object type information with an object image type gallery, determining the object to be updated which is not contained in the object image type gallery based on a comparison result, and extracting features of an object image corresponding to the object to be updated to obtain features of the object to be updated;
and determining a corresponding class atlas to be updated in the object image class atlas based on the characteristics of the object to be updated, adding the object image corresponding to the object to be updated in the class atlas to be updated, and finishing updating the object image class atlas.
Preferably, in step 2, image scanning is performed on an object to be identified based on a cash register to obtain a target object image, the target object image is processed, and a preset object image matched with the target object image in an object image class gallery is determined based on a processing result, which comprises the following steps:
acquiring an object to be identified, and carrying out multi-dimensional scanning on the object to be identified based on a cash register to obtain an initial object image corresponding to the object to be identified in each dimension;
extracting appearance characteristic mark points of an object to be identified, determining the relative position relation of the appearance characteristic mark points in a plane based on an initial object image, performing space conversion on the appearance characteristic mark points based on the scanning angle and the relative position relation of a cash register to obtain the relative position relation of the appearance characteristic mark points in a three-dimensional space, and matching the relative position relation in the three-dimensional space with an actual position relation;
When the relative position relation in the three-dimensional space is inconsistent with the actual position relation, judging that the identifiable object in the obtained target object image has deformation, and correcting each pixel point in the initial object image based on a matching result and the appearance characteristic mark points to obtain the target object image;
mapping each preset object image in the target object image and object image category gallery to a preset two-dimensional coordinate system, respectively obtaining image size information of the target object image and each preset object image based on a mapping result, and performing block processing on the target object image and each preset object image based on the image size information;
respectively extracting characteristic points of a first image block in the segmented target object image and a second image block at the same position in each preset object image, and obtaining characteristic point pairs of the first image block and the second image block based on the characteristic points;
respectively extracting image depth features corresponding to the feature points, carrying out binary coding on the image depth features based on the multidimensional feature description factors, and determining the similarity between the feature point pairs based on the Hamming distance of the binary coding;
determining corresponding target weights at target positions of a target object image and preset object images respectively based on the first image block and the second image block, and carrying out weighted average on the similarity of different characteristic point pairs based on the target weights to respectively obtain the similarity of the target object image and each preset object image;
And obtaining a final preset object image matched with the target object image in the object image category gallery based on the similarity.
Preferably, an automatic object image recognition and classification method based on machine learning obtains a final preset object image matched with a target object image in an object image class gallery based on similarity, including:
obtaining the similarity between the obtained target object image and each preset object image, and sequencing the similarity based on the descending order of values;
and determining a preset object image corresponding to the first similarity based on the sorting result, and judging the preset object image corresponding to the first similarity as a final preset object image.
Preferably, in step 3, attribute information of a preset object image is extracted, and a class label of a target object image is determined based on the attribute information, which includes:
acquiring an obtained preset object image, extracting appearance characteristics of articles in the preset object image, and generating an information access index based on the appearance characteristics;
searching the object catalogs in the server based on the information access index, obtaining a target object catalogs corresponding to the preset object images based on the search result, and calling corresponding attribute information from a corresponding database based on the target object catalogs;
Analyzing the attribute information to obtain an article category corresponding to the preset object image, obtaining a category label, an article name and an article price corresponding to the preset object image based on the article category, and marking the target object image based on the category label.
Preferably, in step 3, a method for automatically identifying and classifying object images based on machine learning displays a category label, an object name and an object price of an object to be identified based on a preset display screen, including:
acquiring a class label, an article name and an article price of the obtained target object image, acquiring configuration parameters of a preset display screen, and determining a reading format of data to be displayed of the preset display screen and a typeable template of the data to be displayed based on the configuration parameters;
performing data transcoding on the category label, the article name and the article price based on the reading format to obtain displayable data, and performing pre-typesetting on the displayable data based on a typeable template;
determining display parameters of the category labels, the article names and the article prices under different typesetting templates based on the pre-typesetting results, checking the display parameters, and judging the typesetting templates with the display parameters meeting expected requirements as target typesetting templates, wherein the target typesetting templates are one;
And mapping the category label, the item name and the item price to corresponding target positions in the target typesetting template, and displaying the category label, the item name and the item price of the item to be identified based on the mapping result.
The invention provides an object image automatic identification and classification system based on machine learning, which comprises:
the gallery construction module is used for acquiring the identifiable object images, learning and training the identifiable object images, extracting object characteristics of the identifiable objects, and constructing an object image category gallery based on the object characteristics;
the image processing module is used for carrying out image scanning on the object to be identified based on the cash register to obtain a target object image, processing the target object image and determining a preset object image matched with the target object image in the object image category gallery based on a processing result;
the article classifying and displaying module is used for extracting attribute information of the preset object image, determining a class label, an article name and an article price of the target object image based on the attribute information, and displaying the class label, the article name and the article price of the article to be identified based on the preset display screen.
Compared with the prior art, the invention has the following beneficial effects:
1. The method has the advantages that through learning and training the identifiable object images, the object image type image library is comprehensively and accurately constructed, an accurate reference basis is provided for automatic identification of the object images, secondly, the object images to be identified are scanned through the cash register, the scanned object images are analyzed, the accurate and effective judgment of the type labels, the object names and the object prices of the object images, which are attributed to the object images, is realized through the preset object images in the object image type image library, and finally, the obtained type labels, the object names and the object prices are displayed, so that the accuracy and the efficiency of identification of the object to be identified are ensured, the classification accuracy of the object to be identified is also ensured, the user can conveniently and effectively learn the purchased object images, and the identification and classification effects of the object are also ensured while a large amount of manpower and material resources are saved.
2. The method comprises the steps of carrying out multidimensional scanning on the object to be identified through a cash register, accurately and effectively acquiring an initial object image of the object to be identified, analyzing the initial object image, correcting the initial object image when the object to be identified in the initial object image is deformed, guaranteeing the accuracy and reliability of a finally obtained target object image, dividing preset object images in a target object image and object image class library, carrying out similarity comparison according to the division result, accurately and effectively determining the preset object image similar to the target object image in the object image class library through similarity, guaranteeing the accuracy and reliability of the obtained preset object image, improving the accuracy of identifying and classifying the object to be identified, saving a large amount of manpower and material resources, and guaranteeing the identifying and classifying effects of the object.
Additional features and advantages of the invention will be set forth in the description which follows, and in part will be obvious from the description, or may be learned by practice of the invention. The objectives and other advantages of the invention will be realized and attained by the structure particularly pointed out in the written description and claims thereof as well as the appended drawings.
The technical scheme of the invention is further described in detail through the drawings and the embodiments.
Drawings
The accompanying drawings are included to provide a further understanding of the invention and are incorporated in and constitute a part of this specification, illustrate the invention and together with the embodiments of the invention, serve to explain the invention. In the drawings:
FIG. 1 is a flow chart of an automatic object image identification and classification method based on machine learning according to an embodiment of the invention;
FIG. 2 is a flowchart of step 1 in a method for automatically identifying and classifying object images based on machine learning according to an embodiment of the present invention;
fig. 3 is a block diagram of an automatic object image recognition and classification system based on machine learning according to an embodiment of the present invention.
Detailed Description
The preferred embodiments of the present invention will be described below with reference to the accompanying drawings, it being understood that the preferred embodiments described herein are for illustration and explanation of the present invention only, and are not intended to limit the present invention.
Example 1:
the embodiment provides an object image automatic identification and classification method based on machine learning, as shown in fig. 1, including:
step 1: acquiring an identifiable article image, performing learning training on the identifiable article image, extracting article characteristics of the identifiable article, and constructing an object image class gallery based on the article characteristics;
step 2: image scanning is carried out on the object to be identified based on a cash register to obtain a target object image, the target object image is processed, and a preset object image matched with the target object image in an object image category gallery is determined based on a processing result;
step 3: extracting attribute information of a preset object image, determining a class label, an article name and an article price of a target object image based on the attribute information, and displaying the class label, the article name and the article price of the article to be identified based on a preset display screen.
In this embodiment, the identifiable object image refers to all training sample images, that is, images corresponding to existing objects, so as to facilitate construction of a corresponding object image class gallery.
In this embodiment, the item features refer to the appearance of the identifiable item in the identifiable item image, the corresponding shape and size, and so on.
In this embodiment, the object image class gallery refers to a gallery constructed according to the object characteristics of different identifiable objects and corresponding identifiable object images, and the identifiable object images of different classes in the object image class gallery are a group, and the identifiable object images of different classes are separated, so as to provide an effective reference basis when determining the class of the object to be identified.
In this embodiment, the cash register is set in advance, is a main body for identifying the object to be identified, and is internally provided with a camera.
In this embodiment, the target object image refers to an image that is obtained by scanning an image of an object to be identified through a cash register and is capable of recording the appearance condition of the object to be identified.
In this embodiment, processing the target object image refers to correcting the image of the identifiable object when the identifiable object in the target object image is deformed, so as to ensure the accuracy and reliability of the final obtained target object image.
In this embodiment, the preset object image refers to an identifiable object image that matches a target object image in the object image class gallery, and the preset object image is one of the identifiable object images.
In this embodiment, the attribute information refers to a category of a preset object image, an item name, a corresponding price, and the like.
In this embodiment, the category label refers to a marking symbol capable of marking different kinds of object images.
In this embodiment, the preset display screen is set in advance, and is used for displaying the obtained category label, the obtained item name and the corresponding item price, and is a component in the cash register.
The beneficial effects of the technical scheme are as follows: the method has the advantages that through learning and training the identifiable object images, the object image type image library is comprehensively and accurately constructed, an accurate reference basis is provided for automatic identification of the object images, secondly, the object images to be identified are scanned through the cash register, the scanned object images are analyzed, the accurate and effective judgment of the type labels, the object names and the object prices of the object images, which are attributed to the object images, is realized through the preset object images in the object image type image library, and finally, the obtained type labels, the object names and the object prices are displayed, so that the accuracy and the efficiency of identification of the object to be identified are ensured, the classification accuracy of the object to be identified is also ensured, the user can conveniently and effectively learn the purchased object images, and the identification and classification effects of the object are also ensured while a large amount of manpower and material resources are saved.
Example 2:
on the basis of embodiment 1, the present embodiment provides an automatic object image recognition and classification method based on machine learning, as shown in fig. 2, in step 1, the method includes:
step 101: acquiring an identifiable object set and a preset imaging requirement, and adapting imaging parameters and shooting angles of a preset camera based on the preset imaging requirement;
step 102: performing multi-angle multi-frequency shooting on each identifiable object in the identifiable object set according to a preset camera based on an adaptation result to obtain an object image set of each identifiable object, and classifying object images in the object image set based on shooting angles;
step 103: obtaining a sub-article image set corresponding to each shooting angle based on the classification result, and performing de-duplication and screening on sub-article images in the sub-article image set based on a preset image screening index to obtain a standard sub-article image corresponding to each shooting angle;
step 104: and summarizing the standard sub-object images corresponding to all shooting angles of the same identifiable object, and obtaining the identifiable object image corresponding to the identifiable object based on the summarizing result.
In this embodiment, the preset imaging requirements are known in advance, including information such as resolution of imaging and imaging size.
In this embodiment, the default camera is a component in the cash register, known in advance.
In this embodiment, the imaging parameters refer to configuration parameters corresponding to the preset camera when performing image photographing on the identifiable object, including an optional resolution value and the like.
In this embodiment, multi-angle multi-frequency photographing refers to image photographing of an identifiable object through a plurality of different directions, and the number of photographing times in each direction is not unique.
In this embodiment, the article image set refers to images corresponding to all directions obtained after image capturing of each identifiable article.
In this embodiment, the sub-item image set refers to all shot images corresponding to each shooting angle of the identifiable item, where the sub-item image is a single image in the sub-item image set.
In this embodiment, the preset image screening index is set in advance, and is used for characterizing a criterion for screening the object image of the sub-object, for example, the resolution needs to be higher than a specific value, and the shot identifiable object needs to be in the middle part of the image.
In this embodiment, the standard sub-item image refers to a final sub-item image obtained by de-duplication and screening of sub-item images in the sub-item image set.
The beneficial effects of the technical scheme are as follows: after the preset camera is adapted according to preset imaging requirements, accurate and effective image acquisition of identifiable objects is achieved, the acquired object images are subjected to de-duplication and screening, the standardization of the finally obtained identifiable objects is ensured, and the accuracy and reliability of the constructed object image category gallery are ensured.
Example 3:
on the basis of embodiment 1, the embodiment provides an object image automatic identification and classification method based on machine learning, in step 1, learning training is performed on identifiable object images, and object features of the identifiable objects are extracted, including:
acquiring an acquired identifiable object image, traversing pixel points of the identifiable object image, and acquiring position information and pixel information of effective pixel points in the identifiable object image;
obtaining edge characteristics of the identifiable object image based on the position information and the pixel information of the effective pixel points, and dividing the identifiable object image based on the edge characteristics to obtain a sub-object image block;
Extracting texture direction characteristics and color characteristics in each sub-article image block, and obtaining local characteristics of each sub-article image block based on the texture direction characteristics and the color characteristics;
filtering the obtained local features based on the edge features, obtaining target local features corresponding to the sub-article image blocks based on the filtering result, and splicing the target local features of different sub-article image blocks based on the relative position relationship among the sub-article image blocks to obtain the article features of the identifiable articles.
In this embodiment, the pixel information refers to a specific pixel value of an effective pixel point in the identifiable item image, where the effective pixel point refers to a pixel point in the identifiable item image that can represent a condition of the identifiable item.
In this embodiment, an edge feature refers to a collection of pixels that can identify discontinuities in the distribution of a characteristic (e.g., pixel gray, texture, etc.) in an image of an item, with a step change or ridge change in the characteristic around the image.
In this embodiment, the sub-item image block refers to a plurality of different image areas obtained by dividing the identifiable item image according to edge features.
In this embodiment, the texture direction feature refers to the texture distribution condition in the image block, and the color feature refers to the specific color matching condition corresponding to different pixel points in the image block.
In this embodiment, the local features are image parameters for characterizing the identifiable item conditions characterized by the different sub-item image blocks, i.e. the image features that the current image block is able to provide in the entire identifiable item image.
In this embodiment, the target local feature refers to a final local image obtained by filtering the obtained local feature through an edge feature, where filtering the obtained local feature refers to removing an invalid local feature.
The beneficial effects of the technical scheme are as follows: the obtained identifiable object images are divided, and the divided different sub-object image blocks are analyzed, so that the local features of the different sub-object image blocks are effectively determined, and finally, the local features of the different sub-object image blocks are summarized, so that the object features of the identifiable object are accurately and effectively determined, the identifiable object images are conveniently classified according to the object features, and the accuracy and the reliability of the constructed object image class gallery are ensured.
Example 4:
on the basis of embodiment 1, the embodiment provides an automatic object image identification and classification method based on machine learning, in step 1, an object image class gallery is constructed based on object features, which comprises the following steps:
Acquiring the obtained object characteristics, determining a class label corresponding to the identifiable object based on the object characteristics, and classifying the identifiable object image based on the class label to obtain an identifiable object class atlas;
extracting configuration information of a preset gallery and byte information of an identifiable object category atlas, dividing the preset gallery based on the configuration information and the byte information to obtain N independent storage spaces, and isolating the N independent storage spaces;
and respectively storing the identifiable object class atlas into corresponding independent storage spaces based on the isolation result, generating an image storage catalog based on the storage result and the class label, displaying the display identification position of the image storage catalog in a preset gallery, and obtaining the object image class gallery based on the display result.
In this embodiment, the identifiable item category atlas refers to each group of image sets obtained by classifying identifiable item images according to category labels.
In this embodiment, the preset gallery is set in advance, and the configuration information of the preset gallery is parameters capable of representing the capacity condition of the preset gallery.
In this embodiment, the byte information refers to the size case of the identifiable item category atlas.
In this embodiment, the independent storage space refers to dividing the storage space of the preset gallery into a plurality of independent spaces, so as to ensure that the identifiable item class atlas of different classes is stored in order.
In this embodiment, the image storage catalog is used to record the storage positions of the identifiable item category atlases of different categories in the preset gallery, so as to facilitate effective retrieval of the identifiable item images of different categories.
In this embodiment, the display identification bit refers to a position capable of displaying and marking the storage condition in the preset gallery, that is, a position capable of displaying the image storage directory.
The beneficial effects of the technical scheme are as follows: the method has the advantages that the class labels of the identifiable objects are accurately and effectively determined according to the obtained object characteristics, the identifiable object images are classified according to the class labels, the preset gallery is divided according to the configuration information of the preset gallery and the byte information of the identifiable object class atlas, the independent storage spaces corresponding to the different classes of the identifiable object class atlas after division are stored, and therefore the object image class gallery is accurately and effectively constructed, reference basis is provided for identifying and classifying the objects to be identified, and accuracy of identification and classification is guaranteed.
Example 5:
on the basis of embodiment 4, the embodiment provides an automatic object image identification and classification method based on machine learning, and an object image class gallery is obtained based on a display result, which comprises the following steps:
traversing the identifiable objects based on a preset time interval, obtaining a current identifiable object set based on a traversing result, and analyzing the current identifiable object set to obtain object type information corresponding to the current identifiable object set;
comparing the object type information with an object image type gallery, determining the object to be updated which is not contained in the object image type gallery based on a comparison result, and extracting features of an object image corresponding to the object to be updated to obtain features of the object to be updated;
and determining a corresponding class atlas to be updated in the object image class atlas based on the characteristics of the object to be updated, adding the object image corresponding to the object to be updated in the class atlas to be updated, and finishing updating the object image class atlas.
In this embodiment, the preset time interval is set in advance, and is used for characterizing a period of traversing the identifiable object, so as to find the updated identifiable object in time, thereby being convenient for accurately and effectively updating the object image class library.
In this embodiment, the current identifiable item set refers to all of the identifiable items currently.
In this embodiment, the item to be updated refers to the kind of the item not included in the object class library that has been constructed.
In this embodiment, the characteristics of the to-be-updated article refer to the appearance condition of the article corresponding to the to-be-updated article, the characteristics of the corresponding specific article, and the like.
In this embodiment, the class atlas to be updated refers to an identifiable object class atlas in the object class library, which is matched with the feature of the object to be updated, and is one of the object class libraries.
The beneficial effects of the technical scheme are as follows: through periodic traversal of the identifiable objects, the identifiable objects to be updated are locked in time when new identifiable objects appear, corresponding identifiable object image sets in the object image type library are locked according to object characteristics of the identifiable objects to be updated, the object images corresponding to the items to be updated are added in the object image type atlas, the object image type library is updated effectively in real time, accuracy and effectiveness of the object image type library are guaranteed, convenience is provided for object identification, and accuracy of object identification and classification is also guaranteed.
Example 6:
on the basis of embodiment 1, the embodiment provides an automatic object image identification and classification method based on machine learning, in step 2, image scanning is performed on an object to be identified based on a cash register to obtain a target object image, the target object image is processed, and a preset object image matched with the target object image in an object image class gallery is determined based on a processing result, and the method comprises the following steps:
acquiring an object to be identified, and carrying out multi-dimensional scanning on the object to be identified based on a cash register to obtain an initial object image corresponding to the object to be identified in each dimension;
extracting appearance characteristic mark points of an object to be identified, determining the relative position relation of the appearance characteristic mark points in a plane based on an initial object image, performing space conversion on the appearance characteristic mark points based on the scanning angle and the relative position relation of a cash register to obtain the relative position relation of the appearance characteristic mark points in a three-dimensional space, and matching the relative position relation in the three-dimensional space with an actual position relation;
when the relative position relation in the three-dimensional space is inconsistent with the actual position relation, judging that the identifiable object in the obtained target object image has deformation, and correcting each pixel point in the initial object image based on a matching result and the appearance characteristic mark points to obtain the target object image;
Mapping each preset object image in the target object image and object image category gallery to a preset two-dimensional coordinate system, respectively obtaining image size information of the target object image and each preset object image based on a mapping result, and performing block processing on the target object image and each preset object image based on the image size information;
respectively extracting characteristic points of a first image block in the segmented target object image and a second image block at the same position in each preset object image, and obtaining characteristic point pairs of the first image block and the second image block based on the characteristic points;
respectively extracting image depth features corresponding to the feature points, carrying out binary coding on the image depth features based on the multidimensional feature description factors, and determining the similarity between the feature point pairs based on the Hamming distance of the binary coding;
determining corresponding target weights at target positions of a target object image and preset object images respectively based on the first image block and the second image block, and carrying out weighted average on the similarity of different characteristic point pairs based on the target weights to respectively obtain the similarity of the target object image and each preset object image;
and obtaining a final preset object image matched with the target object image in the object image category gallery based on the similarity.
In this embodiment, the multi-dimensional scanning refers to scanning the object to be identified through different angles or directions, so as to ensure the comprehensiveness of the obtained object image of the object to be identified.
In this embodiment, the initial object image refers to an image obtained after the object image to be identified is scanned, and the object to be identified recorded in the image may have abnormal conditions such as deformation, and the initial object image needs to be processed.
In this embodiment, the appearance feature mark points refer to position points capable of representing the appearance condition of the object to be identified, so that the similarity between the target object image and the preset object image can be determined conveniently according to the appearance feature mark points.
In this embodiment, the relative positional relationship is used to characterize the relative position of the appearance feature marker points in the initial object image, so as to facilitate determining whether the object to be identified recorded in the initial object image is deformed.
In this embodiment, the spatial conversion refers to converting the relative positional relationship of the appearance feature marker points in the plane into a three-dimensional stereoscopic relationship according to the relative positional relationship.
In this embodiment, the preset two-dimensional coordinate system is set in advance, so as to calibrate the image size of the target object image and the preset object image, thereby facilitating the comparison of the positions of the target object image and the preset object image.
In this embodiment, the first image block refers to a plurality of image area blocks obtained by performing a block division process on the target object image.
In this embodiment, the second image block refers to a plurality of image areas in the obtained preset object images after the preset object images are subjected to the block processing.
In this embodiment, the feature points refer to position points capable of characterizing key image information in the first image block and the second image block.
In this embodiment, the pair of feature points refers to a pair of feature points obtained by associating feature points of a first image block and a second image block at the same position.
In this embodiment, the image depth feature refers to a parameter capable of characterizing the specific conditions such as color and gray scale in the image feature point.
In this embodiment, the multidimensional feature describing factor is set in advance, and is used to convert the image depth feature into a corresponding binary code, where the binary code is a binary value.
In this embodiment, the hamming distance is a parameter used to characterize the degree of similarity between binary codes, and the smaller the distance, the more identical the cluster centers of the two, i.e., the more similar the two.
In this embodiment, the target position is a position, such as a center position or a corner position, corresponding to the first image block and the second image block in the corresponding target object image and the preset object image, respectively.
In this embodiment, the target weight is used to characterize the influence degree of different image blocks on the overall image effect, for example, the influence degree of the image block in the center on the overall image is large, that is, the larger the corresponding target weight value is.
The beneficial effects of the technical scheme are as follows: the method comprises the steps of carrying out multidimensional scanning on the object to be identified through a cash register, accurately and effectively acquiring an initial object image of the object to be identified, analyzing the initial object image, correcting the initial object image when the object to be identified in the initial object image is deformed, guaranteeing the accuracy and reliability of a finally obtained target object image, dividing preset object images in a target object image and object image class library, carrying out similarity comparison according to the division result, accurately and effectively determining the preset object image similar to the target object image in the object image class library through similarity, guaranteeing the accuracy and reliability of the obtained preset object image, improving the accuracy of identifying and classifying the object to be identified, saving a large amount of manpower and material resources, and guaranteeing the identifying and classifying effects of the object.
Example 7:
on the basis of embodiment 6, the present embodiment provides a machine learning-based object image automatic identification and classification method, which obtains a final preset object image matched with a target object image in an object image class gallery based on similarity, including:
Obtaining the similarity between the obtained target object image and each preset object image, and sequencing the similarity based on the descending order of values;
and determining a preset object image corresponding to the first similarity based on the sorting result, and judging the preset object image corresponding to the first similarity as a final preset object image.
In this embodiment, the first-order similarity refers to the maximum similarity value determined according to the descending order of values.
The beneficial effects of the technical scheme are as follows: by sequencing the obtained similarity, the final preset object image is accurately and effectively determined according to the sequencing result, and the accuracy and reliability of the obtained preset object image are ensured.
Example 8:
on the basis of embodiment 1, the present embodiment provides an automatic object image recognition and classification method based on machine learning, in step 3, attribute information of a preset object image is extracted, and a class label of a target object image is determined based on the attribute information, including:
acquiring an obtained preset object image, extracting appearance characteristics of articles in the preset object image, and generating an information access index based on the appearance characteristics;
searching the object catalogs in the server based on the information access index, obtaining a target object catalogs corresponding to the preset object images based on the search result, and calling corresponding attribute information from a corresponding database based on the target object catalogs;
Analyzing the attribute information to obtain an article category corresponding to the preset object image, obtaining a category label, an article name and an article price corresponding to the preset object image based on the article category, and marking the target object image based on the category label.
In this embodiment, the appearance features refer to the appearance shape and the like of the article in the preset object image.
In this embodiment, the information access index refers to a reference basis that can retrieve a preset object image in the object image class gallery, and according to the information access index, the preset object image required in the object image class gallery can be accurately and effectively retrieved.
In this embodiment, the target object directory refers to specific directory information corresponding to the preset object image in the object directory, and is one of the object directories.
In this embodiment, the database is set in advance, and is used for storing attribute information corresponding to different preset object images.
The beneficial effects of the technical scheme are as follows: the method comprises the steps of analyzing the obtained preset object image, accurately and effectively determining the appearance characteristics of the preset object image, generating an information access index through the appearance characteristics, effectively searching the object catalogue in the server, locking the target object catalogue corresponding to the preset object image, effectively calling the attribute information of the preset object image from the corresponding database according to the target object catalogue, accurately and effectively locking the category label, the object name and the object price corresponding to the preset object image according to the attribute information, locking the category label, the object name and the object price of the object to be identified through the preset object image, improving the accuracy and the effect of identifying and classifying the object to be identified, and facilitating the user to accurately and effectively know the effective information of the shopping object.
Example 9:
on the basis of embodiment 1, the embodiment provides an automatic object image identification and classification method based on machine learning, in step 3, category labels, object names and object prices of objects to be identified are displayed based on a preset display screen, including:
acquiring a class label, an article name and an article price of the obtained target object image, acquiring configuration parameters of a preset display screen, and determining a reading format of data to be displayed of the preset display screen and a typeable template of the data to be displayed based on the configuration parameters;
performing data transcoding on the category label, the article name and the article price based on the reading format to obtain displayable data, and performing pre-typesetting on the displayable data based on a typeable template;
determining display parameters of the category labels, the article names and the article prices under different typesetting templates based on the pre-typesetting results, checking the display parameters, and judging the typesetting templates with the display parameters meeting expected requirements as target typesetting templates, wherein the target typesetting templates are one;
and mapping the category label, the item name and the item price to corresponding target positions in the target typesetting template, and displaying the category label, the item name and the item price of the item to be identified based on the mapping result.
In this embodiment, the configuration parameters refer to a display format, typesetting conditions, and the like of the data to be displayed by the preset display screen.
In this embodiment, the reading format refers to a format that is required by the display screen to be displayed, i.e. a format in which data is effectively read.
In this embodiment, the typesetting template refers to a typesetting mode in which the preset display screen can display the data to be displayed, for example, the typesetting mode can display the data in a vertical column or display the data in a horizontal direction.
In this embodiment, the data transcoding refers to converting the formats of the current category label, the item name and the item price according to the reading format, where the displayable data is the data that can be directly displayed on the preset display screen obtained after converting the formats of the category label, the item name and the item price.
In this embodiment, pre-typesetting refers to the simulated typesetting of displayable data through typeable templates, in order to screen out templates suitable for displaying category labels, item names, and item prices.
In this embodiment, the display parameter is used to characterize the display condition of the presentable data under different typeable templates, for example, whether the overlapping area is stored and whether the presentable data can be completely displayed.
In this embodiment, the expected requirement is set in advance, and is used to characterize the display requirement of the preset display screen on the displayable data, and the display requirement can be adjusted according to the actual situation.
In this embodiment, the target typesetting template refers to a typesetting template suitable for displaying the current category label, the item name, and the item price.
In this embodiment, the target location refers to a filling location corresponding to the category label, the item name, and the item price in the target layout template.
The beneficial effects of the technical scheme are as follows: the obtained category label, the item name and the item price of the object to be identified are processed, the category label, the item name and the item price are displayed through the preset display screen, the information of the purchased object is effectively known by a user, the effect of identifying and classifying the object is guaranteed, and a large amount of manpower and material resources are saved.
Example 10:
the embodiment provides an automatic object image recognition and classification system based on machine learning, as shown in fig. 3, including:
the gallery construction module is used for acquiring the identifiable object images, learning and training the identifiable object images, extracting object characteristics of the identifiable objects, and constructing an object image category gallery based on the object characteristics;
The image processing module is used for carrying out image scanning on the object to be identified based on the cash register to obtain a target object image, processing the target object image and determining a preset object image matched with the target object image in the object image category gallery based on a processing result;
the article classifying and displaying module is used for extracting attribute information of the preset object image, determining a class label, an article name and an article price of the target object image based on the attribute information, and displaying the class label, the article name and the article price of the article to be identified based on the preset display screen.
The beneficial effects of the technical scheme are as follows: the method has the advantages that through learning and training the identifiable object images, the object image type image library is comprehensively and accurately constructed, an accurate reference basis is provided for automatic identification of the object images, secondly, the object images to be identified are scanned through the cash register, the scanned object images are analyzed, the accurate and effective judgment of the type labels, the object names and the object prices of the object images, which are attributed to the object images, is realized through the preset object images in the object image type image library, and finally, the obtained type labels, the object names and the object prices are displayed, so that the accuracy and the efficiency of identification of the object to be identified are ensured, the classification accuracy of the object to be identified is also ensured, the user can conveniently and effectively learn the purchased object images, and the identification and classification effects of the object are also ensured while a large amount of manpower and material resources are saved.
It will be apparent to those skilled in the art that various modifications and variations can be made to the present invention without departing from the spirit or scope of the invention. Thus, it is intended that the present invention also include such modifications and alterations insofar as they come within the scope of the appended claims or the equivalents thereof.
Claims (10)
1. An object image automatic identification and classification method based on machine learning is characterized by comprising the following steps:
step 1: acquiring an identifiable article image, performing learning training on the identifiable article image, extracting article characteristics of the identifiable article, and constructing an object image class gallery based on the article characteristics;
step 2: image scanning is carried out on the object to be identified based on a cash register to obtain a target object image, the target object image is processed, and a preset object image matched with the target object image in an object image category gallery is determined based on a processing result;
step 3: extracting attribute information of a preset object image, determining a class label, an article name and an article price of a target object image based on the attribute information, and displaying the class label, the article name and the article price of the article to be identified based on a preset display screen.
2. The machine learning-based object image automatic identification and classification method according to claim 1, wherein in step 1, obtaining an identifiable object image comprises:
Acquiring an identifiable object set and a preset imaging requirement, and adapting imaging parameters and shooting angles of a preset camera based on the preset imaging requirement;
performing multi-angle multi-frequency shooting on each identifiable object in the identifiable object set according to a preset camera based on an adaptation result to obtain an object image set of each identifiable object, and classifying object images in the object image set based on shooting angles;
obtaining a sub-article image set corresponding to each shooting angle based on the classification result, and performing de-duplication and screening on sub-article images in the sub-article image set based on a preset image screening index to obtain a standard sub-article image corresponding to each shooting angle;
and summarizing the standard sub-object images corresponding to all shooting angles of the same identifiable object, and obtaining the identifiable object image corresponding to the identifiable object based on the summarizing result.
3. The automatic object image recognition and classification method based on machine learning according to claim 1, wherein in step 1, learning training is performed on the image of the identifiable object, and extracting the object characteristics of the identifiable object comprises:
acquiring an acquired identifiable object image, traversing pixel points of the identifiable object image, and acquiring position information and pixel information of effective pixel points in the identifiable object image;
Obtaining edge characteristics of the identifiable object image based on the position information and the pixel information of the effective pixel points, and dividing the identifiable object image based on the edge characteristics to obtain a sub-object image block;
extracting texture direction characteristics and color characteristics in each sub-article image block, and obtaining local characteristics of each sub-article image block based on the texture direction characteristics and the color characteristics;
filtering the obtained local features based on the edge features, obtaining target local features corresponding to the sub-article image blocks based on the filtering result, and splicing the target local features of different sub-article image blocks based on the relative position relationship among the sub-article image blocks to obtain the article features of the identifiable articles.
4. The automatic object image recognition and classification method based on machine learning according to claim 1, wherein in step 1, an object image class gallery is constructed based on object features, comprising:
acquiring the obtained object characteristics, determining a class label corresponding to the identifiable object based on the object characteristics, and classifying the identifiable object image based on the class label to obtain an identifiable object class atlas;
extracting configuration information of a preset gallery and byte information of an identifiable object category atlas, dividing the preset gallery based on the configuration information and the byte information to obtain N independent storage spaces, and isolating the N independent storage spaces;
And respectively storing the identifiable object class atlas into corresponding independent storage spaces based on the isolation result, generating an image storage catalog based on the storage result and the class label, displaying the display identification position of the image storage catalog in a preset gallery, and obtaining the object image class gallery based on the display result.
5. The automatic recognition and classification method of object images based on machine learning according to claim 4, wherein the obtaining of the object image class gallery based on the display result comprises:
traversing the identifiable objects based on a preset time interval, obtaining a current identifiable object set based on a traversing result, and analyzing the current identifiable object set to obtain object type information corresponding to the current identifiable object set;
comparing the object type information with an object image type gallery, determining the object to be updated which is not contained in the object image type gallery based on a comparison result, and extracting features of an object image corresponding to the object to be updated to obtain features of the object to be updated;
and determining a corresponding class atlas to be updated in the object image class atlas based on the characteristics of the object to be updated, adding the object image corresponding to the object to be updated in the class atlas to be updated, and finishing updating the object image class atlas.
6. The automatic recognition and classification method of object images based on machine learning according to claim 1, wherein in step 2, image scanning is performed on an object to be recognized based on a cash register to obtain a target object image, the target object image is processed, and a preset object image matched with the target object image in an object image class gallery is determined based on a processing result, comprising:
acquiring an object to be identified, and carrying out multi-dimensional scanning on the object to be identified based on a cash register to obtain an initial object image corresponding to the object to be identified in each dimension;
extracting appearance characteristic mark points of an object to be identified, determining the relative position relation of the appearance characteristic mark points in a plane based on an initial object image, performing space conversion on the appearance characteristic mark points based on the scanning angle and the relative position relation of a cash register to obtain the relative position relation of the appearance characteristic mark points in a three-dimensional space, and matching the relative position relation in the three-dimensional space with an actual position relation;
when the relative position relation in the three-dimensional space is inconsistent with the actual position relation, judging that the identifiable object in the obtained target object image has deformation, and correcting each pixel point in the initial object image based on a matching result and the appearance characteristic mark points to obtain the target object image;
Mapping each preset object image in the target object image and object image category gallery to a preset two-dimensional coordinate system, respectively obtaining image size information of the target object image and each preset object image based on a mapping result, and performing block processing on the target object image and each preset object image based on the image size information;
respectively extracting characteristic points of a first image block in the segmented target object image and a second image block at the same position in each preset object image, and obtaining characteristic point pairs of the first image block and the second image block based on the characteristic points;
respectively extracting image depth features corresponding to the feature points, carrying out binary coding on the image depth features based on the multidimensional feature description factors, and determining the similarity between the feature point pairs based on the Hamming distance of the binary coding;
determining corresponding target weights at target positions of a target object image and preset object images respectively based on the first image block and the second image block, and carrying out weighted average on the similarity of different characteristic point pairs based on the target weights to respectively obtain the similarity of the target object image and each preset object image;
and obtaining a final preset object image matched with the target object image in the object image category gallery based on the similarity.
7. The automatic recognition and classification method of object images based on machine learning according to claim 6, wherein obtaining a final preset object image matched with a target object image in an object image class gallery based on similarity comprises:
Obtaining the similarity between the obtained target object image and each preset object image, and sequencing the similarity based on the descending order of values;
and determining a preset object image corresponding to the first similarity based on the sorting result, and judging the preset object image corresponding to the first similarity as a final preset object image.
8. The automatic recognition and classification method of object images based on machine learning according to claim 1, wherein in step 3, extracting attribute information of a preset object image and determining a class label of a target object image based on the attribute information comprises:
acquiring an obtained preset object image, extracting appearance characteristics of articles in the preset object image, and generating an information access index based on the appearance characteristics;
searching the object catalogs in the server based on the information access index, obtaining a target object catalogs corresponding to the preset object images based on the search result, and calling corresponding attribute information from a corresponding database based on the target object catalogs;
analyzing the attribute information to obtain an article category corresponding to the preset object image, obtaining a category label, an article name and an article price corresponding to the preset object image based on the article category, and marking the target object image based on the category label.
9. The automatic object image recognition and classification method based on machine learning according to claim 1, wherein in step 3, the class label, the object name and the object price of the object to be recognized are displayed based on a preset display screen, comprising:
acquiring a class label, an article name and an article price of the obtained target object image, acquiring configuration parameters of a preset display screen, and determining a reading format of data to be displayed of the preset display screen and a typeable template of the data to be displayed based on the configuration parameters;
performing data transcoding on the category label, the article name and the article price based on the reading format to obtain displayable data, and performing pre-typesetting on the displayable data based on a typeable template;
determining display parameters of the category labels, the article names and the article prices under different typesetting templates based on the pre-typesetting results, checking the display parameters, and judging the typesetting templates with the display parameters meeting expected requirements as target typesetting templates, wherein the target typesetting templates are one;
and mapping the category label, the item name and the item price to corresponding target positions in the target typesetting template, and displaying the category label, the item name and the item price of the item to be identified based on the mapping result.
10. An automatic object image recognition and classification system based on machine learning, comprising:
the gallery construction module is used for acquiring the identifiable object images, learning and training the identifiable object images, extracting object characteristics of the identifiable objects, and constructing an object image category gallery based on the object characteristics;
the image processing module is used for carrying out image scanning on the object to be identified based on the cash register to obtain a target object image, processing the target object image and determining a preset object image matched with the target object image in the object image category gallery based on a processing result;
the article classifying and displaying module is used for extracting attribute information of the preset object image, determining a class label, an article name and an article price of the target object image based on the attribute information, and displaying the class label, the article name and the article price of the article to be identified based on the preset display screen.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311378761.6A CN117115569B (en) | 2023-10-24 | 2023-10-24 | Automatic object image identification and classification method and system based on machine learning |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202311378761.6A CN117115569B (en) | 2023-10-24 | 2023-10-24 | Automatic object image identification and classification method and system based on machine learning |
Publications (2)
| Publication Number | Publication Date |
|---|---|
| CN117115569A true CN117115569A (en) | 2023-11-24 |
| CN117115569B CN117115569B (en) | 2024-02-06 |
Family
ID=88809542
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202311378761.6A Active CN117115569B (en) | 2023-10-24 | 2023-10-24 | Automatic object image identification and classification method and system based on machine learning |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN117115569B (en) |
Cited By (1)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117793151A (en) * | 2024-02-28 | 2024-03-29 | 深圳桑达银络科技有限公司 | Distributed network security monitoring system and method based on cloud computing |
Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108648233A (en) * | 2018-03-24 | 2018-10-12 | 北京工业大学 | A kind of target identification based on deep learning and crawl localization method |
| CN110472480A (en) * | 2019-06-28 | 2019-11-19 | 深圳数马电子技术有限公司 | Item identification method, device and computer readable storage medium |
| CN113850167A (en) * | 2021-09-15 | 2021-12-28 | 天津市府易科技股份有限公司 | Commodity identification method and system based on edge calculation and machine deep learning |
| CN113902940A (en) * | 2021-07-26 | 2022-01-07 | 惠州学院 | Neural network-based multi-class article visual identification method and metering equipment |
| KR20220052798A (en) * | 2020-10-21 | 2022-04-28 | 주식회사 팀솔루션 | Method for generating image data for machine learning and apparatus thereof |
| CN114937179A (en) * | 2022-07-27 | 2022-08-23 | 深圳市海清视讯科技有限公司 | Junk image classification method and device, electronic equipment and storage medium |
-
2023
- 2023-10-24 CN CN202311378761.6A patent/CN117115569B/en active Active
Patent Citations (6)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108648233A (en) * | 2018-03-24 | 2018-10-12 | 北京工业大学 | A kind of target identification based on deep learning and crawl localization method |
| CN110472480A (en) * | 2019-06-28 | 2019-11-19 | 深圳数马电子技术有限公司 | Item identification method, device and computer readable storage medium |
| KR20220052798A (en) * | 2020-10-21 | 2022-04-28 | 주식회사 팀솔루션 | Method for generating image data for machine learning and apparatus thereof |
| CN113902940A (en) * | 2021-07-26 | 2022-01-07 | 惠州学院 | Neural network-based multi-class article visual identification method and metering equipment |
| CN113850167A (en) * | 2021-09-15 | 2021-12-28 | 天津市府易科技股份有限公司 | Commodity identification method and system based on edge calculation and machine deep learning |
| CN114937179A (en) * | 2022-07-27 | 2022-08-23 | 深圳市海清视讯科技有限公司 | Junk image classification method and device, electronic equipment and storage medium |
Non-Patent Citations (1)
| Title |
|---|
| 黄志清;贾翔;郭一帆;张菁;: "基于深度学习的端到端乐谱音符识别", 天津大学学报(自然科学与工程技术版), no. 06 * |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN117793151A (en) * | 2024-02-28 | 2024-03-29 | 深圳桑达银络科技有限公司 | Distributed network security monitoring system and method based on cloud computing |
| CN117793151B (en) * | 2024-02-28 | 2024-04-30 | 深圳桑达银络科技有限公司 | Distributed network security monitoring system and method based on cloud computing |
Also Published As
| Publication number | Publication date |
|---|---|
| CN117115569B (en) | 2024-02-06 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US7327888B2 (en) | Geometric hashing method for model-based recognition of an object | |
| CN117115569B (en) | Automatic object image identification and classification method and system based on machine learning | |
| CN111709816A (en) | Service recommendation method, device and equipment based on image recognition and storage medium | |
| CN112070135A (en) | Power equipment image detection method and device, power equipment and storage medium | |
| CN113239227B (en) | Image data structuring method, device, electronic equipment and computer readable medium | |
| CN114511718B (en) | Intelligent management method and system for materials for building construction | |
| EP0923044B1 (en) | Method and means of matching documents based on spatial region layout | |
| JP2006119723A (en) | Device and method for image processing | |
| CN114463567A (en) | Block chain-based intelligent education operation big data plagiarism prevention method and system | |
| CN113627288A (en) | Intelligent information label obtaining method for massive images | |
| CN115169375B (en) | AR and gun ball linkage-based high-level material visualization method | |
| CN116109992A (en) | Goods shelf commodity identification method and system | |
| CN110826551A (en) | Intelligent discrimination method for VIN code rubbing die image of motor vehicle | |
| JP2000082075A (en) | Device and method for retrieving image by straight line and program recording medium thereof | |
| Raisi et al. | Applying content-based image retrieval techniques to provide new services for tourism industry | |
| KR101572330B1 (en) | Apparatus and method for near duplicate video clip detection | |
| CN110737790A (en) | database-based universal image data set management method | |
| CN114758259B (en) | Package detection method and system based on X-ray object image recognition | |
| CN114677541B (en) | Method and system for extracting bonding sample set based on target | |
| CN116308435B (en) | Agricultural product precise identification method and system based on intelligent traceability scale | |
| CN114266887B (en) | Large-scale trademark detection method based on deep learning | |
| CN113627320B (en) | Engineering drawing comparison device and method based on computer vision | |
| Uchiyama et al. | AR representation system for 3D GIS based on camera pose estimation using distribution of intersections | |
| CA2395758C (en) | A geometric hashing method for model-based recognition of an object | |
| Al-Oraiqat et al. | A modified image comparison algorithm using histogram features |
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 | ||
| GR01 | Patent grant | ||
| GR01 | Patent grant |