CN112419241A - Object identification method and device based on artificial intelligence and readable storage medium - Google Patents
Object identification method and device based on artificial intelligence and readable storage medium Download PDFInfo
- Publication number
- CN112419241A CN112419241A CN202011217756.3A CN202011217756A CN112419241A CN 112419241 A CN112419241 A CN 112419241A CN 202011217756 A CN202011217756 A CN 202011217756A CN 112419241 A CN112419241 A CN 112419241A
- Authority
- CN
- China
- Prior art keywords
- image
- identified
- repaired
- model
- inputting
- 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 31
- 238000013473 artificial intelligence Methods 0.000 title claims abstract description 18
- 238000003860 storage Methods 0.000 title claims abstract description 9
- 230000007547 defect Effects 0.000 claims abstract description 47
- 238000003384 imaging method Methods 0.000 claims abstract description 26
- 238000013256 Gubra-Amylin NASH model Methods 0.000 claims description 16
- 230000002950 deficient Effects 0.000 claims description 11
- 238000013135 deep learning Methods 0.000 claims description 8
- 238000012512 characterization method Methods 0.000 claims description 7
- 230000006870 function Effects 0.000 claims description 6
- 238000005286 illumination Methods 0.000 claims description 5
- 238000010586 diagram Methods 0.000 description 5
- 238000009826 distribution Methods 0.000 description 4
- 230000000694 effects Effects 0.000 description 4
- 238000007689 inspection Methods 0.000 description 4
- 238000010422 painting Methods 0.000 description 4
- 238000011156 evaluation Methods 0.000 description 3
- 238000004519 manufacturing process Methods 0.000 description 3
- ORFSSYGWXNGVFB-UHFFFAOYSA-N sodium 4-amino-6-[[4-[4-[(8-amino-1-hydroxy-5,7-disulfonaphthalen-2-yl)diazenyl]-3-methoxyphenyl]-2-methoxyphenyl]diazenyl]-5-hydroxynaphthalene-1,3-disulfonic acid Chemical group COC1=C(C=CC(=C1)C2=CC(=C(C=C2)N=NC3=C(C4=C(C=C3)C(=CC(=C4N)S(=O)(=O)O)S(=O)(=O)O)O)OC)N=NC5=C(C6=C(C=C5)C(=CC(=C6N)S(=O)(=O)O)S(=O)(=O)O)O.[Na+] ORFSSYGWXNGVFB-UHFFFAOYSA-N 0.000 description 2
- 238000006467 substitution reaction Methods 0.000 description 2
- 238000009827 uniform distribution Methods 0.000 description 1
Images
Classifications
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/0008—Industrial image inspection checking presence/absence
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/22—Matching criteria, e.g. proximity measures
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06F—ELECTRIC DIGITAL DATA PROCESSING
- G06F18/00—Pattern recognition
- G06F18/20—Analysing
- G06F18/24—Classification techniques
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/04—Architecture, e.g. interconnection topology
- G06N3/045—Combinations of networks
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N3/00—Computing arrangements based on biological models
- G06N3/02—Neural networks
- G06N3/08—Learning methods
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/0002—Inspection of images, e.g. flaw detection
- G06T7/0004—Industrial image inspection
- G06T7/001—Industrial image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/60—Analysis of geometric attributes
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
- G06T7/90—Determination of colour characteristics
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T2207/00—Indexing scheme for image analysis or image enhancement
- G06T2207/30—Subject of image; Context of image processing
- G06T2207/30108—Industrial image inspection
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Data Mining & Analysis (AREA)
- General Engineering & Computer Science (AREA)
- Life Sciences & Earth Sciences (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Biophysics (AREA)
- General Health & Medical Sciences (AREA)
- Bioinformatics & Cheminformatics (AREA)
- Health & Medical Sciences (AREA)
- Biomedical Technology (AREA)
- Bioinformatics & Computational Biology (AREA)
- Computational Linguistics (AREA)
- Evolutionary Biology (AREA)
- Molecular Biology (AREA)
- Computing Systems (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Quality & Reliability (AREA)
- Geometry (AREA)
- Image Analysis (AREA)
Abstract
The invention discloses an object identification method, device and readable storage medium based on artificial intelligence, comprising the following steps: acquiring a first image, wherein the first image comprises an object to be identified; inputting the first image into a pre-trained model to generate a second image, wherein the second image comprises a repaired object, the imaging conditions of the repaired object in the second image are the same as or similar to those of the object to be identified in the first image, and the repaired object represents that the defect on the object to be identified is repaired; and determining whether the object to be identified has defects or not based on the first image and the second image. Therefore, the scheme generates the second images with the same or similar imaging conditions for each first image, and when the scheme is used, the second images are used as the comparison template to judge whether the object to be identified in the first images has defects or not, so that the accuracy of whether the object to be identified is intact or not can be improved, and the yield is improved.
Description
Technical Field
The invention relates to the technical field of object identification, in particular to an object identification method and device based on artificial intelligence and a readable storage medium.
Background
In the entity manufacturing industry, a product image is often used to judge whether a product has a defect, and the specific method is to compare the product image with a product template image and judge whether the product has a defect according to a comparison result.
However, in practical applications, the photographed product images are often different for the same batch of products, and if the product images with different appearances are compared with the same product template image, the defect-free product is easily judged to be defective by mistake, and the error rate is high.
Disclosure of Invention
The embodiment of the invention provides an object identification method and device based on artificial intelligence and a readable storage medium, which are used for improving the accuracy of whether an object to be identified has defects.
The invention provides an object identification method based on artificial intelligence, which comprises the following steps: acquiring a first image, wherein the first image comprises an object to be identified; inputting the first image into a pre-trained model to generate a second image, wherein the second image comprises a repaired object, the imaging conditions of the repaired object in the second image are the same as or similar to those of the object to be identified in the first image, and the repaired object represents that the defect on the object to be identified is repaired; and determining whether the object to be identified has defects or not based on the first image and the second image.
In one embodiment, the model is a GAN model, and the model training process is: inputting a plurality of image samples into the GAN model for training, wherein the plurality of image samples are all defect-free samples.
In one possible embodiment, the GAN model applies a loss function based on an absolute error algorithm.
In an embodiment, the determining whether the object to be authenticated has a defect based on the first image and the second image includes: and inputting the first image and the second image into a matching network based on deep learning for comparison, and outputting a comparison result for representing whether the object to be identified has defects or not.
In an embodiment, the inputting the first image and the second image into a matching network based on deep learning for comparison includes: identifying and acquiring an object to be identified in the first image and a repaired object in the second image; and inputting the identified object to be identified and the repaired object into the matching network for comparison.
In one embodiment, the matching network is a twin network.
In one embodiment, the first and second images include images, video, and 3D models.
In one embodiment, the elements of the imaging condition include a shooting distance, a shooting angle, and a shooting illumination of the image and a position of an object in the image.
Another aspect of the present invention provides an object authentication apparatus based on artificial intelligence, the apparatus comprising: the data acquisition module is used for acquiring a first image, wherein the first image comprises an object to be identified; the data generation module is used for inputting the first image into a pre-trained model and generating a second image, wherein the second image comprises a repaired object, the imaging conditions of the repaired object in the second image and the characterization of the object to be identified in the first image are the same or similar, and the repaired object characterizes that the defect on the object to be identified is repaired; and the data identification module is used for determining whether the object to be identified has defects or not based on the first image and the second image.
Another aspect of the invention provides a computer-readable storage medium comprising a set of computer-executable instructions that, when executed, perform any of the artificial intelligence based object authentication methods described above.
In the embodiment of the invention, the second image with the same or similar imaging conditions is generated for each first image, and when the method is used, the second image is used as the comparison template to judge whether the object to be identified in the first image has defects, so that the accuracy of whether the object to be identified is intact can be improved, and the yield is improved.
Drawings
The above and other objects, features and advantages of exemplary embodiments of the present invention will become readily apparent from the following detailed description read in conjunction with the accompanying drawings. Several embodiments of the invention are illustrated by way of example, and not by way of limitation, in the figures of the accompanying drawings and in which:
in the drawings, the same or corresponding reference numerals indicate the same or corresponding parts.
FIG. 1 is a schematic diagram of an implementation flow of an object identification method based on artificial intelligence according to an embodiment of the present invention;
FIG. 2 is a schematic view of an embodiment of the present invention in which the object to be authenticated is a nut;
FIG. 3 is a schematic diagram of an embodiment of the present invention in which an object to be authenticated is a computer motherboard;
FIG. 4 is a schematic view of an embodiment of the present invention in which the object to be identified is a kitten;
FIG. 5 is a schematic diagram of a first image and a repaired second image according to an embodiment of the invention;
FIG. 6 is a schematic illustration of the repair of a non-defective object and a defective object in an embodiment of the invention;
fig. 7 is a schematic structural diagram of an object identification apparatus based on artificial intelligence according to an embodiment of the present invention.
Detailed Description
In order to make the objects, features and advantages of the present invention more obvious and understandable, the technical solutions in the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings in the embodiments of the present invention, and it is apparent that the described embodiments are only a part of the embodiments of the present invention, and not all of the embodiments. All other embodiments, which can be derived by a person skilled in the art from the embodiments given herein without making any creative effort, shall fall within the protection scope of the present invention.
Fig. 1 is a schematic flow chart illustrating an implementation of an object identification method based on artificial intelligence according to an embodiment of the present invention.
As shown in fig. 1, an aspect of the present invention provides an artificial intelligence-based object authentication method, including:
and 103, determining whether the object to be identified has defects or not based on the first image and the second image.
In this embodiment, in step 101, the first image and the second image may be images, videos or 3D models. The object to be authenticated may be a food item (such as a nut as shown in fig. 2), a product part (such as a computer motherboard part as shown in fig. 3), or an animal or plant (such as a kitten as shown in fig. 4).
In step 102, the pre-trained model may be a tv (total variation) model, a GAN model, or an ImageNet model, and is mainly used to repair the input first image and generate a second image including the repaired object. The repairing effect can be seen with reference to fig. 5, the nut image in the first image is the object to be identified, it can be seen that a black defect exists on the nut image, after the pre-trained model repairing process, the black defect on the nut image in the second image has been repaired, and a complete nut image is formed; if the nut image in the first image is a defect-free image, the second image has almost the same appearance as the first image after the first image is subjected to the model repairing process which is trained in advance.
Continuing to refer to fig. 5, the imaging conditions of the repaired object in the second image and the object to be identified in the first image are the same or similar, wherein the imaging conditions are the same, which means that the positions of the object to be identified and the repaired object in the images are the same, the shape and the size are the same, the illumination is the same, the angle is the same, and the like; the imaging conditions are similar, which means that the object to be identified and the repaired object have only slight differences in overall image comparison, wherein the slight differences include overall shape size differences, color differences and the like. In addition, regarding the background images of the first image and the second image, it should be noted that the background images of the first image and the second image may be identical, for example, the background images of the first image and the second image are the same skeins; the background images of the two images may also be non-uniform, such as the background image in the first image is pure black, and the background image in the second image is pure blue or even no background image.
In step 103, the manner of determining whether the object to be authenticated has defects may be: identifying and extracting objects in the first image and the second image, comparing the pixel values of the object to be identified and the repaired object, or directly comparing the pixel values of the first image and the second image, and determining whether the object to be identified has defects according to the difference value of the comprehensive pixel values; the first image and the second image can also be input into a network model (such as a twin network) which is trained in advance and used for image matching to carry out image similarity comparison, and whether the object to be identified has defects or not can be determined according to the output result of the model.
Therefore, the scheme generates the second images with the same or similar imaging conditions for each first image, and when the scheme is used, the second images are used as the comparison template to judge whether the object to be identified in the first images has defects or not, so that the accuracy of whether the object to be identified is intact or not can be improved, and the yield is improved.
The scheme can be applied to the entity manufacturing industry, for example, if the first image is a computer mainboard, if the object to be identified is a memory bank slot and a fracture defect exists in the memory bank slot, a second image including the complete memory bank slot is generated through a model, because the imaging conditions of the first image and the second image are the same or similar, the appearance of the generated second image is approximately the same as that of the first image, only the pixel difference of the fracture part of the memory bank slot exists, at the moment, when the second image is used as a template and is compared with the first image, the memory bank slot on the current mainboard is easy to judge to have the defect and needs to be repaired. If the memory bank slot in the first image is not defective, the memory bank slot in the generated second image is also non-defective, and the first image and the second image are identical or similar in appearance, then the memory bank slot on the current mainboard is determined to be non-defective.
The scheme can also be applied to art or 3D evaluation teaching, the first image is a painting work or a 3D modeling work, the object to be identified is painting content or a modeling model, the first image generates a repaired second image through a pre-trained model, and the first image and the second image are compared to judge whether the original work is complete, so that an evaluator can comprehensively score according to the judgment result.
The scheme can also be applied to the appearance inspection of vehicle annual inspection, at the moment, the first image is a vehicle to be detected, the object to be identified is the whole vehicle or a part of the vehicle, the first image generates a repaired second image through a pre-trained model, the first image and the second image are compared, whether the current vehicle to be detected is complete or not can be judged, and therefore an evaluator can comprehensively detect the vehicle with reference to the judgment result.
In one embodiment, the model is a GAN model, and the model training process is:
and inputting a plurality of image samples into the GAN model for training, wherein the plurality of image samples are all defect-free samples.
In this embodiment, the GAN model is respectively composed of a generative model (generative model) and a discriminant model (discriminant model), where the generative model is used to capture the distribution of sample data, and a sample similar to real training data is generated by using noise obeying a certain distribution (uniform distribution, gaussian distribution, etc.), and the pursuit effect is better as the real sample is; the discriminant model is a two-classifier that estimates the probability that a sample is from training data (rather than from the generated data), and outputs a high probability if the sample is from the real training data, and a low probability otherwise.
The model training process of the scheme is as follows: training a large number of defect-free image samples as sample data to reconstruct the model, so that the model learns the distribution of normal data, and finally the GAN model achieves the following effects: reference is made to fig. 6, which shows a defect-free nut image as input to the GAN model, resulting in an image that is nearly identical to itself; and taking the defective nut image as an input of a GAN model to obtain an image of the repaired defect.
In one possible embodiment, the GAN model applies a loss function based on an absolute error algorithm.
In this embodiment, the existing GAN model mainly uses a loss function based on an absolute error algorithm (referred to as L1) and a loss function based on a mean square error algorithm (referred to as L2), and the loss function based on the absolute error algorithm is selected in the scheme, so that the image is closer to reality.
In one embodiment, determining whether the object to be authenticated has a defect based on the first image and the second image comprises:
and inputting the first image and the second image into a matching network based on deep learning for comparison, and outputting a comparison result for representing whether the object to be identified has defects or not.
In this embodiment, the matching network may be a twin network, a MatchNet model, or the like, preferably, the twin network is used for comparing similarity of images of the two images, and if the result output by the matching network indicates that the similarity of the two images is high, it is determined that the object to be identified in the first image is defect-free, whereas if the result output by the matching network indicates that the similarity of the two images is low, it is determined that the object to be identified in the first image is defect-free.
In one embodiment, the inputting the first image and the second image into a matching network based on deep learning for comparison includes:
identifying and acquiring an object to be identified in the first image and a repaired object in the second image;
and inputting the identified object to be identified and the repaired object into a matching network for comparison.
In this embodiment, during image comparison, the mature object model (such as the MobileNet model) dedicated to the present solution may be used to identify the object to be identified in the first image and the repaired object in the second image, and then the repaired object and the object to be identified are input into the matching network for comparison.
In one possible embodiment, the matching network is a twin network.
In this embodiment, the input values of the twin network are the first image and the second image, and the output value is a numerical value representing the similarity between the first image and the second image, and if the numerical value is higher than a first preset threshold, it indicates that the similarity between the first image and the second image is higher, that is, the object to be identified in the first image has no defect; if the value is lower than a second preset threshold value, the similarity between the first image and the second image is low, namely the object to be identified in the first image has a defect. The first preset threshold and the second preset threshold can be the same threshold, and can be set according to the required precision requirement in actual use.
In one embodiment, the elements of the imaging conditions include the shooting distance, shooting angle and shooting illumination of the image and the position of the object in the image.
In this embodiment, the shooting distance of the image is the distance between the shooting device and the object to be identified, the shooting angle is the shooting direction of the shooting device, and if the imaging conditions are the same, the shape, size, color and position of the object to be identified in the first image are the same as those of the restored object in the second image.
Fig. 7 is a schematic structural diagram of an object identification apparatus based on artificial intelligence according to an embodiment of the present invention.
As shown in fig. 7, another aspect of the present invention provides an artificial intelligence based object authentication apparatus, including:
a data acquiring module 201, configured to acquire a first image, where the first image includes an object to be identified;
the data generation module 202 is configured to input the first image into a pre-trained model, and generate a second image, where the second image includes a repaired object, an imaging condition of the repaired object in the second image is the same as or similar to an imaging condition of a characterization of an object to be identified in the first image, and a defect on the characterization of the repaired object on the characterization of the object to be identified is repaired;
and the data identification module 203 is used for determining whether the object to be identified has defects or not based on the first image and the second image.
In this embodiment, in the data acquisition module 201, the first image and the second image may be images, videos, or 3D models. The object to be authenticated may be a food item (such as a nut as shown in fig. 2), a product part (such as a computer motherboard part as shown in fig. 3), or an animal or plant (such as a kitten as shown in fig. 4).
In the data generating module 202, the pre-trained model may be a tv (total variation) model, a GAN model, or an ImageNet model, and is mainly used to repair the input first image and generate a second image including the repaired object. The repairing effect can be seen with reference to fig. 5, the nut image in the first image is the object to be identified, it can be seen that a black defect exists on the nut image, after the pre-trained model repairing process, the black defect on the nut image in the second image has been repaired, and a complete nut image is formed; if the nut image in the first image is a defect-free image, the second image has almost the same appearance as the first image after the first image is subjected to the model repairing process which is trained in advance.
Continuing to refer to fig. 5, the imaging conditions of the repaired object in the second image and the object to be identified in the first image are the same or similar, wherein the imaging conditions are the same, which means that the positions of the object to be identified and the repaired object in the images are the same, the shape and the size are the same, the illumination is the same, the angle is the same, and the like; the imaging conditions are similar, which means that the object to be identified and the repaired object have only slight differences in overall image comparison, wherein the slight differences include overall shape size differences, color differences and the like. In addition, regarding the background images of the first image and the second image, it should be noted that the background images of the first image and the second image may be identical, for example, the background images of the first image and the second image are the same skeins; the background images of the two images may also be non-uniform, such as the background image in the first image is pure black, and the background image in the second image is pure blue or even no background image.
In the data authentication module 203, the manner of determining whether the object to be authenticated has a defect may be: identifying and extracting objects in the first image and the second image, comparing the pixel values of the object to be identified and the repaired object, or directly comparing the pixel values of the first image and the second image, and determining whether the object to be identified has defects according to the difference value of the comprehensive pixel values; the first image and the second image can also be input into a network model (such as a twin network) which is trained in advance and used for image matching to carry out image similarity comparison, and whether the object to be identified has defects or not can be determined according to the output result of the model.
Therefore, the scheme generates the second images with the same or similar imaging conditions for each first image, and when the scheme is used, the second images are used as the comparison template to judge whether the object to be identified in the first images has defects or not, so that the accuracy of whether the object to be identified is intact or not can be improved, and the yield is improved.
The scheme can be applied to the entity manufacturing industry, for example, if the first image is a computer mainboard, if the object to be identified is a memory bank slot and a fracture defect exists in the memory bank slot, a second image including the complete memory bank slot is generated through a model, because the imaging conditions of the first image and the second image are the same or similar, the appearance of the generated second image is approximately the same as that of the first image, only the pixel difference of the fracture part of the memory bank slot exists, at the moment, when the second image is used as a template and is compared with the first image, the memory bank slot on the current mainboard is easy to judge to have the defect and needs to be repaired. If the memory bank slot in the first image is not defective, the memory bank slot in the generated second image is also non-defective, and the first image and the second image are identical or similar in appearance, then the memory bank slot on the current mainboard is determined to be non-defective.
The scheme can also be applied to art or 3D evaluation teaching, the first image is a painting work or a 3D modeling work, the object to be identified is painting content or a modeling model, the first image generates a repaired second image through the model, and the first image and the second image are compared to judge whether the original work is complete or not, so that the evaluation can be carried out by referring to the judgment result.
The scheme can also be applied to the appearance inspection of vehicle annual inspection, at the moment, the first image is a vehicle to be detected, the object to be identified is the whole vehicle or a part of the vehicle, the first image generates a repaired second image through a pre-trained model, the first image and the second image are compared, whether the current vehicle to be detected is complete or not can be judged, and therefore an evaluator can comprehensively detect the vehicle with reference to the judgment result.
In one embodiment, the data authentication module 203 is specifically configured to:
and inputting the first image and the second image into a matching network based on deep learning for comparison, and outputting a comparison result for representing whether the object to be identified has defects or not.
In this embodiment, the matching network may be a twin network, a MatchNet model, or the like, preferably, the twin network is used for comparing similarity of images of the two images, and if the result output by the matching network indicates that the similarity of the two images is high, it is determined that the object to be identified in the first image is defect-free, whereas if the result output by the matching network indicates that the similarity of the two images is low, it is determined that the object to be identified in the first image is defect-free.
In one embodiment, when the first image and the second image are input into the deep learning based matching network for matching, the data identification module 203 is specifically configured to:
identifying and acquiring an object to be identified in the first image and a repaired object in the second image;
and inputting the identified object to be identified and the repaired object into the matching network for comparison.
In this embodiment, during image comparison, the mature object model (such as the MobileNet model) dedicated to the present solution may be used to identify the object to be identified in the first image and the repaired object in the second image, and then the repaired object and the object to be identified are input into the matching network for comparison.
Another aspect of the invention provides a computer-readable storage medium comprising a set of computer-executable instructions that, when executed, perform any of the artificial intelligence based object authentication methods described above.
In one embodiment of the present invention, a computer-readable storage medium comprises a set of computer-executable instructions that, when executed, are configured to obtain a first image, wherein the first image comprises an object to be authenticated; inputting the first image into a pre-trained model to generate a second image, wherein the second image comprises a repaired object, the imaging conditions of the repaired object in the second image are the same as or similar to those of the object to be identified in the first image, and the defect of the object to be identified represented by the repaired object is repaired; and determining whether the object to be identified has defects or not based on the first image and the second image.
Therefore, the scheme generates the second images with the same or similar imaging conditions for each first image, and when the scheme is used, the second images are used as the comparison template to judge whether the object to be identified in the first images has defects.
In the description herein, references to the description of the term "one embodiment," "some embodiments," "an example," "a specific example," or "some examples," etc., mean that a particular feature, structure, material, or characteristic described in connection with the embodiment or example is included in at least one embodiment or example of the invention. Furthermore, the particular features, structures, materials, or characteristics described may be combined in any suitable manner in any one or more embodiments or examples. Furthermore, various embodiments or examples and features of different embodiments or examples described in this specification can be combined and combined by one skilled in the art without contradiction.
Furthermore, the terms "first", "second" and "first" are used for descriptive purposes only and are not to be construed as indicating or implying relative importance or implicitly indicating the number of technical features indicated. Thus, a feature defined as "first" or "second" may explicitly or implicitly include at least one such feature. In the description of the present invention, "a plurality" means two or more unless specifically defined otherwise.
The above description is only for the specific embodiments of the present invention, but the scope of the present invention is not limited thereto, and any person skilled in the art can easily conceive of the changes or substitutions within the technical scope of the present invention, and the changes or substitutions should be covered within the scope of the present invention. Therefore, the protection scope of the present invention shall be subject to the protection scope of the claims.
Claims (10)
1. An artificial intelligence based object authentication method, the method comprising:
acquiring a first image, wherein the first image comprises an object to be identified;
inputting the first image into a pre-trained model to generate a second image, wherein the second image comprises a repaired object, the imaging conditions of the repaired object in the second image are the same as or similar to those of the object to be identified in the first image, and the repaired object represents that the defect on the object to be identified is repaired;
and determining whether the object to be identified has defects or not based on the first image and the second image.
2. The method of claim 1, the model being a GAN model, the model training process being:
inputting a plurality of image samples into the GAN model for training, wherein the plurality of image samples are all defect-free samples.
3. The method of claim 2, the GAN model applying a loss function based on an absolute error algorithm.
4. The method of claim 1, wherein said determining whether the object to be authenticated is defective based on the first image and the second image comprises:
and inputting the first image and the second image into a matching network based on deep learning for comparison, and outputting a comparison result for representing whether the object to be identified has defects or not.
5. The method of claim 4, wherein the inputting the first image and the second image into a deep learning based matching network for comparison comprises:
identifying and acquiring an object to be identified in the first image and a repaired object in the second image;
and inputting the identified object to be identified and the repaired object into the matching network for comparison.
6. The method of claim 4 or 5, the matching network being a twin network.
7. The method of claim 1, the first imagery and second imagery including images, video, and 3D models.
8. The method of claim 1, the elements of the imaging conditions including a shooting distance, a shooting angle, and a shooting illumination of the picture and a position of an object in the picture.
9. An artificial intelligence based object authentication apparatus, the apparatus comprising:
the data acquisition module is used for acquiring a first image, wherein the first image comprises an object to be identified;
the data generation module is used for inputting the first image into a pre-trained model and generating a second image, wherein the second image comprises a repaired object, the imaging conditions of the repaired object in the second image and the characterization of the object to be identified in the first image are the same or similar, and the repaired object characterizes that the defect on the object to be identified is repaired;
and the data identification module is used for determining whether the object to be identified has defects or not based on the first image and the second image.
10. A readable storage medium comprising a set of computer-executable instructions that, when executed, perform the artificial intelligence based object authentication method of any one of claims 1-8.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011217756.3A CN112419241A (en) | 2020-11-04 | 2020-11-04 | Object identification method and device based on artificial intelligence and readable storage medium |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| CN202011217756.3A CN112419241A (en) | 2020-11-04 | 2020-11-04 | Object identification method and device based on artificial intelligence and readable storage medium |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN112419241A true CN112419241A (en) | 2021-02-26 |
Family
ID=74828144
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202011217756.3A Pending CN112419241A (en) | 2020-11-04 | 2020-11-04 | Object identification method and device based on artificial intelligence and readable storage medium |
Country Status (1)
| Country | Link |
|---|---|
| CN (1) | CN112419241A (en) |
Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108509059A (en) * | 2018-03-27 | 2018-09-07 | 联想(北京)有限公司 | A kind of information processing method, electronic equipment and computer storage media |
| CN110619618A (en) * | 2018-06-04 | 2019-12-27 | 杭州海康威视数字技术股份有限公司 | Surface defect detection method and device and electronic equipment |
| KR20200080408A (en) * | 2018-12-19 | 2020-07-07 | 포인드 주식회사 | Artificial Intelligence rapid image generation method using Meta-learning Generative adversarial network |
| CN111724372A (en) * | 2020-06-19 | 2020-09-29 | 深圳新视智科技术有限公司 | Method, terminal and storage medium for detecting cloth defects based on antagonistic neural network |
-
2020
- 2020-11-04 CN CN202011217756.3A patent/CN112419241A/en active Pending
Patent Citations (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN108509059A (en) * | 2018-03-27 | 2018-09-07 | 联想(北京)有限公司 | A kind of information processing method, electronic equipment and computer storage media |
| CN110619618A (en) * | 2018-06-04 | 2019-12-27 | 杭州海康威视数字技术股份有限公司 | Surface defect detection method and device and electronic equipment |
| KR20200080408A (en) * | 2018-12-19 | 2020-07-07 | 포인드 주식회사 | Artificial Intelligence rapid image generation method using Meta-learning Generative adversarial network |
| CN111724372A (en) * | 2020-06-19 | 2020-09-29 | 深圳新视智科技术有限公司 | Method, terminal and storage medium for detecting cloth defects based on antagonistic neural network |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US10964004B2 (en) | Automated optical inspection method using deep learning and apparatus, computer program for performing the method, computer-readable storage medium storing the computer program, and deep learning system thereof | |
| CN111179251B (en) | Defect detection system and method based on twin neural network and by utilizing template comparison | |
| Bergmann et al. | MVTec AD--A comprehensive real-world dataset for unsupervised anomaly detection | |
| CN111507976B (en) | Defect detection method and system based on multi-angle imaging | |
| CN113920107A (en) | Insulator damage detection method based on improved yolov5 algorithm | |
| CN114862776A (en) | Product surface defect detection method and device, computer equipment and medium | |
| US11315229B2 (en) | Method for training defect detector | |
| CN110596120A (en) | Glass boundary defect detection method, device, terminal and storage medium | |
| US20220245402A1 (en) | Ai-based pre-training model determination system, and ai-based vision inspection management system using same for product production lines | |
| CN112766110A (en) | Training method of object defect recognition model, object defect recognition method and device | |
| CN116188475A (en) | Intelligent control method, system and medium for automatic optical detection of appearance defects | |
| US20220076404A1 (en) | Defect management apparatus, method and non-transitory computer readable medium | |
| CN114494780A (en) | Semi-supervised industrial defect detection method and system based on feature comparison | |
| CN115272961A (en) | VR equipment early warning monitoring method and system based on artificial intelligence | |
| CN113554630A (en) | Chip surface defect detection method, system, computer device and storage medium | |
| JP7459697B2 (en) | Anomaly detection system, learning device, anomaly detection program, learning program, anomaly detection method, and learning method | |
| CN112419241A (en) | Object identification method and device based on artificial intelligence and readable storage medium | |
| CN115564734A (en) | Product detection method, electronic equipment and storage medium | |
| CN115620083A (en) | Model training method, face image quality evaluation method, device and medium | |
| KR20230036650A (en) | Defect detection method and system based on image patch | |
| CN103020601B (en) | Hi-line visible detection method and device | |
| CN113034432A (en) | Product defect detection method, system, device and storage medium | |
| Dai et al. | Anomaly detection and segmentation based on defect repaired image resynthesis | |
| JP7345764B2 (en) | Inspection system and inspection program | |
| US20240013369A1 (en) | Image defect detecting system, generation method of image defect detecting system and non-transitory 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 |