CN113906471A - Method and apparatus for performing machine learning and testing on multiple images - Google Patents
Method and apparatus for performing machine learning and testing on multiple images Download PDFInfo
- Publication number
- CN113906471A CN113906471A CN202080041294.6A CN202080041294A CN113906471A CN 113906471 A CN113906471 A CN 113906471A CN 202080041294 A CN202080041294 A CN 202080041294A CN 113906471 A CN113906471 A CN 113906471A
- Authority
- CN
- China
- Prior art keywords
- learning
- testing
- images
- criterion
- image
- 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
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
- G06T7/001—Industrial image inspection using an image reference approach
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06N—COMPUTING ARRANGEMENTS BASED ON SPECIFIC COMPUTATIONAL MODELS
- G06N20/00—Machine learning
-
- G—PHYSICS
- G06—COMPUTING; CALCULATING OR COUNTING
- G06T—IMAGE DATA PROCESSING OR GENERATION, IN GENERAL
- G06T7/00—Image analysis
-
- 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
- G06T2207/20—Special algorithmic details
- G06T2207/20081—Training; Learning
Landscapes
- Engineering & Computer Science (AREA)
- Theoretical Computer Science (AREA)
- Software Systems (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Physics & Mathematics (AREA)
- General Physics & Mathematics (AREA)
- Artificial Intelligence (AREA)
- Data Mining & Analysis (AREA)
- Evolutionary Computation (AREA)
- Medical Informatics (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Quality & Reliability (AREA)
- Image Analysis (AREA)
Abstract
A method of performing machine learning and testing on a plurality of images, comprising the steps of: receiving a plurality of images, and performing learning and testing on the plurality of images; the test is performed by reflecting the result of performing the learning.
Description
Technical Field
Embodiments disclosed in the present specification relate to a method and apparatus for performing machine learning and testing on a plurality of images, and more particularly, to a method and apparatus for performing both machine learning and testing on the same image and performing testing by reflecting the result of machine learning.
Topic information and thank you note in 2019
1. Topic number: 1711094483
2. Explanation of thank-seng: the research is carried out by the research results of the global SW professional company development project of the scientific and technological information communication department and the information communication technology promotion center (IITP-2019-0-01423-
Background
Recently, AI technology is applied to the field of machine vision (machine vision), making a great contribution to factory automation. However, in order to apply the AI technique to the field of machine vision, it is necessary to determine whether or not a product has defects and degrees or a criterion for determining the type of an article, and in order to set such a criterion, machine learning (machine learning) should be performed on a plurality of images of an article photographed in advance. For example, machine learning may be performed on a plurality of images taken of good or defective products, or may be performed on a plurality of images taken of a plurality of types of products.
In a machine vision technique that is generally used, in order to perform a test (test) for distinguishing good products from defective products by taking images, learning (learning) is first performed using images for training (training) that are separately prepared. For example, a method of determining whether an article is good or bad is performed by performing learning on a plurality of images of good or bad articles to grasp the characteristics of the images of good or bad articles, and then testing the images of good or bad articles.
However, in the case of the existing manner of separately performing learning and testing as described above, there are problems as follows.
First, due to the characteristics of the production process in which the occurrence rate of defective products is extremely low, it is not only difficult to obtain an image in which defective products are photographed, but also the work of dividing a plurality of images into images of good products and images of defective products is performed before learning is performed, and therefore, time and effort are required.
Secondly, since the test is performed based on the one-time learning execution result, when an environmental change of illumination brightness or position or the like of the post-learning illuminated article occurs, the accuracy of the test result may be lowered.
In connection with this, in korean patent No. 10-1867475, which is a prior art document, there is disclosed a method of clustering the contents of similar models by performing unsupervised learning on static data (static data) using an automatic encoder.
On the other hand, the foregoing background art is the technical information which the inventors have possessed for the derivation of the present invention or which they have learned in the course of deriving the present invention, and is not necessarily the publicly known art disclosed before applying the present invention.
Disclosure of Invention
Technical problem to be solved
Embodiments disclosed in the present specification provide a machine learning and test execution method and apparatus for reflecting a learning result in real time by performing both machine learning and testing on a plurality of images, thereby improving test accuracy.
Means for solving the problems
In order to improve the accuracy of machine learning and testing, both learning and testing are performed on a plurality of images, wherein the testing is performed by reflecting the result of performing the learning.
Effects of the invention
According to any one of the above means for solving the problems, by performing both learning and testing on a plurality of images, wherein the testing is performed by reflecting the learning result, it is possible to reflect the learning result in real time and perform the testing, and therefore, it is possible to expect an effect of improving the testing accuracy by adaptively responding to environmental changes.
In addition, since it is not necessary to separately prepare images for learning, it is possible to reduce the time and effort required to surely take images of defective products or sort images.
Effects that can be obtained by the disclosed embodiments are not limited to the above-mentioned effects, and other effects not mentioned can be clearly understood by those skilled in the art from the following description.
Drawings
FIG. 1 is a diagram illustrating a system for performing machine learning and testing on a plurality of images, according to an embodiment.
Fig. 2 is a block diagram for explaining the structure of the computing apparatus shown in fig. 1.
Fig. 3 to 8 are flowcharts for explaining a method for performing machine learning and testing on a plurality of images of the embodiment.
Detailed Description
As a technical means for achieving the above technical problem, according to an embodiment, a method of performing machine learning and testing on a plurality of images may include the steps of: receiving a plurality of images, and performing learning and testing on the plurality of images; the test is performed by reflecting the result of performing the learning.
According to another embodiment, there is provided a computer program for executing a method of performing machine learning and testing on a plurality of images in a computing device, the method of performing machine learning and testing on a plurality of images may comprise the steps of: receiving a plurality of images, and performing learning and testing on the plurality of images; the test is performed by reflecting the result of performing the learning.
According to still another embodiment, there is provided a computer-readable recording medium having recorded thereon a program for executing the method of performing machine learning and testing on a plurality of images in a computing apparatus, the method of performing machine learning and testing on a plurality of images may include the steps of: receiving a plurality of images, and performing learning and testing on the plurality of images; the test is performed by reflecting the result of performing the learning.
According to yet another embodiment, an apparatus for performing machine learning and testing, comprises: an input output section that receives operations and data inputs related to machine learning and testing and displays a data processing result, a communication section that communicates with an external device to transmit and receive data, a storage section that stores a program for executing the machine learning and testing, and a control section that executes the program to execute the machine learning and testing on a plurality of images; the control section performs learning and testing on the plurality of images received through the communication section; the control portion executes the test by reflecting a result of executing the learning.
Hereinafter, various embodiments are described in detail with reference to the accompanying drawings. The embodiments described below may also be modified and implemented in various different forms. In order to more clearly describe the features of the embodiments, detailed descriptions of matters known to those of ordinary skill in the art to which the following embodiments belong will be omitted. Moreover, in the drawings, portions irrelevant to the description of the embodiments are omitted, and like reference numerals are given to like portions throughout the specification.
Throughout the specification, when it is described that one component is "connected" to another component, it includes not only the case of "directly connected" but also the case of "connected with another component interposed therebetween". Also, when it is described that one component "includes" another component, it is meant that other components may be included, not excluded, unless otherwise specified.
FIG. 1 is a diagram illustrating a system for performing machine learning and testing on a plurality of images, according to an embodiment. Referring to fig. 1, a system for performing machine learning and testing on a plurality of images according to an embodiment may include a photographing part 10 and a computing device 100.
The system shown in fig. 1 is a machine vision system for determining whether an article 1 is good or bad, and when an image of the article 1 is captured by the imaging unit 10 and transmitted to the computing device 100, the computing device 100 determines whether the captured article 1 is good or bad by analyzing the received image. For this reason, the computing apparatus 100 can grasp the features of the normal image (image capturing good products) and the abnormal image (image capturing defective products) by learning (hereinafter, the meaning of "machine learning") the received images.
The imaging unit 10 is a structure for imaging the appearance of the article 1, and may be implemented by a camera including an image sensor or the like. Although the photographing part 10 is illustrated as a structure separated from the computing device 100 in fig. 1, the photographing part 10 may be a structure included in the computing device 100, unlike this. In addition, the photographing part 10 may further include illumination for illuminating the article 1 at the time of photographing to ensure a clearer image.
The photographing section 10 photographs the articles 1 moving by a conveyor belt or the like one by one, and may transmit the photographed image to the computing device 100.
The computing apparatus 100 performs machine learning and testing on the plurality of images received from the photographing section 10. The computing apparatus 100 analyzes the image of the captured article 1 to determine whether the article 1 included in the image is a good article or a defective article.
In particular, the computing apparatus 100 performs both learning and testing on a plurality of images, wherein the testing is performed by reflecting the result of performing the learning, so that the learning result can be reflected in the testing in real time. That is, the computing apparatus 100 can perform both learning and testing on the same image without separating the image for learning and the image for testing from each other. Therefore, since it is not necessary to separately prepare an image for learning, an effect of improving user convenience can be expected. In addition, the image on which the test is performed is learned while the test is performed, and the result of the learning is reflected when the test is performed on the next image, so that it is possible to expect an effect of improving the test accuracy by adaptively responding even if the surrounding environment such as the brightness or position of the illumination is changed.
Hereinafter, a specific method of the computing apparatus 100 performing learning and testing on a plurality of images will be described with reference to the drawings.
Fig. 2 is a block diagram for explaining the structure of the computing apparatus 100 shown in fig. 1. Referring to fig. 2, the computing device 100 according to an embodiment may include an input output part 110, a communication part 120, a control part 130, and a storage part 140.
The input/output unit 110 is configured to receive an input from a user for executing settings and the like related to machine learning and testing, and output a test result and the like. According to an embodiment, the input/output section 110 receives a setting input from a user through an input device such as a keyboard and a mouse, and may display the result of performing learning and testing on a screen.
The communication section 120 is a structure for performing communication with an external device to transmit and receive data, and may support various manners of wired or wireless communication. For example, the communication unit 120 may sequentially receive images of the photographed article 1 from the photographing unit 10. The communication section 120 may be implemented as a communication chipset supporting various communication protocols.
The control section 130 is a structure including at least one processor such as a CPU, and controls the overall operation of the computing apparatus 100. In particular, the control section 130 may perform machine learning and testing on a plurality of images by executing a program stored in the storage section 140. Hereinafter, a specific operation of the control section 130 to perform machine learning and testing on a plurality of images will be described in detail.
Various types of programs and data may be stored in the storage section 140. In particular, the storage unit 140 may store a program for executing machine learning and testing on a plurality of images, or may store a plurality of images to be subjects of learning and testing.
According to an embodiment, the control section 130 performs learning and testing on the plurality of images received through the communication section 120, wherein the testing can be performed by reflecting the result of performing the learning. At this time, the communication unit 120 receives the images of the photographed article 1 from the photographing unit 10 in sequence, and the control unit 130 may perform learning and testing one by one in the order of receiving the images.
When the control section 130 performs learning and testing on a plurality of images one by one, it is possible to perform testing on the next image by reflecting the result of the previous learning. For example, when learning and testing are performed on any one image (first image), the control section 130 performs testing on the next image (second image) by reflecting the result of learning on the image (first image). In addition, the control section 130 may perform learning on the next image (second image), and may reflect the result thereof when performing a test on the still next image (third image).
In order to perform a test on an image, the control unit 130 needs to know in advance a standard for the test (hereinafter, referred to as a "test standard"), for example, a feature of a normal image (an image of a good product). Such test criteria may be stored in the storage section 140 in advance, or the control section 130 directly grasps and stores in the storage section 140 in the process of performing learning on the image.
Hereinafter, an example in which the control unit 130 directly grasps the test standard in the process of learning the image will be described.
The control unit 130 starts to sequentially perform learning on a plurality of images one by one. First, since the test standard is not prepared, the control section 130 performs learning only on the image. If learning is sequentially performed on images one by one and then a predetermined certain criterion (hereinafter, referred to as "preparation criterion") is satisfied, the control section 130 determines that preparation for setting a test criterion is completed, and performs both learning and testing on images in the subsequent order, wherein the testing is performed by reflecting the result of learning performed until the preparation criterion is satisfied and the result of learning on images in the subsequent order.
In detail, when it is determined that the preparation criterion is satisfied, the control section 130 sets a test criterion based on a result of learning performed until the preparation criterion is satisfied, and performs a test on images in the subsequent order. At this time, the control unit 130 executes learning while executing a test on the subsequent image, and continues updating the test standard based on the result. That is, when the control unit 130 executes a test on an image in the subsequent order, the test is executed by reflecting the learning result on another image executed before the execution of the test.
The method for the control unit 130 to determine whether the preparation criterion is satisfied is as follows.
The control section 130 may determine whether or not the preparation criterion is satisfied based on the time for which learning has been performed so far or the number of images for which learning has been performed so far. This is because the test criteria such as the characteristics of the normal image can be grasped only if learning is performed for at least a certain time or more or learning is performed for at least a certain number of images or more.
For example, the control section 130 confirms the time until the learning is performed, and may determine that the preparation criterion is satisfied when the confirmed time exceeds a predetermined value. Alternatively, for example, the control section 130 confirms the number of images for which learning has been performed so far, and may determine that the preparation criterion is satisfied when the confirmed number exceeds a predetermined value.
In addition to this, the control section 130 may determine whether or not preparation for test standard setting is completed (whether or not the preparation standard is satisfied) in various ways.
When the preparation criteria are satisfied, the control section 130 performs both learning and testing on the images in the subsequent order as follows.
When it is determined that the preparation criterion is satisfied, the control section 130 grasps the feature of the normal image and sets it as a test criterion based on the learning result performed until the preparation criterion is satisfied. Here, the test standard is exemplified as the feature of the normal image, but unlike this, the feature of the abnormal image or the like may be the test standard.
The control section 130 performs a test on any one image (first image) based on the grasped feature of the normal image. That is, the control section 130 determines whether the first image is a normal image or an abnormal image. When the first image is determined to be an abnormal image, the control unit 130 may display a prompt indicating that a defective product is detected via the input/output unit 110.
The control section 130 performs not only the test but also learning on the first image. The control section 130 updates the feature of the normal image based on the result of performing learning on the first image. Therefore, when a test is performed on an image next to the first image (second image), the control section 130 may determine whether or not the second image is a normal image based on the feature of the normal image reflecting the learning result of the first image. As described above, the control section 130 can continue to update the test standard by performing the test on each image while performing the learning.
On the other hand, it has been described above that the control section 130 updates the test standard (the feature of the normal image) each time learning is performed on one image, and if so frequently updated, the processing speed or resource utilization may be inefficient. Therefore, the test criterion may be updated only in a case where a predetermined certain criterion (hereinafter, referred to as "update criterion") is satisfied, instead of updating the test criterion for each image. Hereinafter, such an embodiment will be explained.
When it is determined that the preparation criteria are satisfied, the control section 130 may grasp the feature of the normal image based on the result of learning performed previously. Here, the test standard is exemplified as the feature of the normal image, but unlike this, the feature of the abnormal image or the like may be the test standard.
The control section 130 performs a test on any one image (first image) based on the grasped feature of the normal image. That is, the control section 130 determines whether the first image is a normal image or an abnormal image. When the first image is determined to be an abnormal image, the control unit 130 may display a prompt indicating that a defective product is detected via the input/output unit 110.
The control section 130 performs not only the test but also learning on the first image. When the learning of the first image is completed, the control section 130 determines whether or not a predetermined update criterion is satisfied. In more detail, the control section 130 confirms the time at which the learning is performed after the time at which the feature grasping of the normal image is completed or the time at which the updating of the feature of the normal image is completed, and may judge that the update criterion is satisfied when the confirmed time exceeds a predetermined value. Alternatively, the control section 130 confirms the number of images for which learning is performed after the time at which feature grasping of the normal image is completed or the time at which updating of the features of the normal image is completed, and may determine that the update criterion is satisfied when the confirmed number exceeds a predetermined value.
As described above, the control unit 130 can reflect the learning result and update the test criterion only when a certain criterion (update criterion) is satisfied, thereby improving efficiency while reflecting the learning result in real time.
Hereinafter, a method of performing machine learning and testing on a plurality of images according to an embodiment will be described with reference to fig. 3 to 8. The machine learning and test execution method according to the embodiment shown in fig. 3 to 8 includes the steps of chronological processing in the computing device 100 shown in fig. 1 and 2. Accordingly, even if omitted below, the contents described above in relation to the computing device 100 shown in fig. 1 and 2 may be applied to the machine learning and test execution method according to the embodiments shown in fig. 3 to 8.
FIG. 3 is a flow diagram illustrating a method for performing machine learning and testing on a plurality of images, according to an embodiment.
Referring to FIG. 3, in step 301, computing device 100 receives a plurality of images. At this time, the computing apparatus 100 may receive a plurality of images at a time, and may also sequentially receive images as targets in order while performing learning and testing.
In step 302, the computing device 100 performs both learning and testing on the plurality of images, wherein the testing is performed by reflecting the learning results. In other words, the computing apparatus 100 performs learning and testing on a plurality of images in sequence, reflecting the result of learning on the previous image to perform testing on the next image.
Fig. 4 is a flowchart for explaining detailed steps included in step 302 of fig. 3. Referring to fig. 4, in step 401, the computing apparatus 100 performs learning on the received plurality of images one by one.
In step 402, the computing device 100 determines whether there is a next image to learn, terminates the process if there is no next image, and executes step 403 if there is a next image.
In step 403, the computing device 100 determines whether a predetermined criterion (preparation criterion) is satisfied, and if the preparation criterion is not satisfied, returns to perform step 401, and if the preparation criterion is satisfied, performs step 404. Here, the "preparation criterion" refers to a criterion for judging whether or not preparation of the criterion (test criterion) for setting the test as described above is completed. That is, when the preparation criterion is satisfied, the computing apparatus 100 judges that the preparation for setting the test criterion is completed. The computing apparatus 100 determines whether or not the preparation criterion is satisfied based on the time at which learning has been performed so far or the number of images on which learning has been performed so far, and a detailed process will be described with reference to fig. 7 and 8 below.
In step 404, the computing device 100 performs both learning and testing on the remaining images, wherein the testing is performed reflecting the learning results of the other images.
When it is determined that the preparation criterion is satisfied, the computing apparatus 100 performs both learning and testing on the images of the subsequent order, wherein the testing is performed by reflecting the learning result performed until the preparation criterion is satisfied and the learning result on the images of the subsequent order. In detail, when it is determined that the preparation criterion is satisfied, the computing apparatus 100 sets a test criterion based on a result of learning performed until the preparation criterion is satisfied, and performs a test on images in the subsequent order. At this time, the computing apparatus 100 performs learning while performing a test on images in the subsequent order, and continues to update the test criterion based on the result thereof. That is, when the computing apparatus 100 performs a test on images in the subsequent order, the test is performed by reflecting the learning result of other images performed immediately before the test is performed.
A specific procedure in which the computing apparatus 100 performs the test by reflecting the learning result of the other image in step 404 will be described below with reference to fig. 5 and 6.
Fig. 5 is a flowchart for explaining detailed steps included in step 404 of fig. 4. Referring to fig. 5, in step 501, the computing apparatus 100 grasps the feature of the normal image based on the previous learning execution result.
In step 502, the computing device 100 detects an abnormal image by performing a test on the image based on features of the normal image. The computing apparatus 100 can determine whether each image is a normal image or an abnormal image by comparing the features of the normal image grasped at step 501 with each image.
In step 503, the computing device 100 performs learning on the image on which the test was performed in step 502. In step 504, the computing apparatus 100 updates the feature of the normal image based on the result of the learning performed in step 503.
In step 505, the computing device 100 determines whether there is a next image, returns to performing step 502 if there is a next image, and terminates the process if there is no next image.
On the other hand, as described above, in the case of updating the features of the normal image, the computing apparatus 100 is not updated every time learning of one image is completed, but is updated only in the case where a predetermined certain condition (update condition) is satisfied, thereby improving efficiency while reflecting the learning result in real time. Such an embodiment will be described below with reference to fig. 6.
Fig. 6 is a flowchart for explaining detailed steps included in step 404 of fig. 4. Referring to fig. 6, in step 601, the computing apparatus 100 grasps the feature of the normal image based on the previous learning execution result.
In step 602, the computing device 100 detects an abnormal image by performing a test on the image based on features of the normal image. The computing apparatus 100 can determine whether each image is a normal image or an abnormal image by comparing the features of the normal image grasped at step 601 with each image.
In step 603, the computing device 100 performs learning on the image on which the test was performed in step 602. In step 604, the computing device 100 determines whether there is a next image, and if the determination result is that there is no next image, the process is terminated, and conversely, if there is a next image, step 605 is executed.
In step 605, the computing apparatus 100 determines whether a predetermined criterion (update criterion) is satisfied. In detail, the computing apparatus 100 confirms the time at which learning is performed after the time at which feature grasping of the normal image is completed or the time at which updating of the features of the normal image is completed, and when the confirmed time exceeds a predetermined value, it may be determined that the update criterion is satisfied. Alternatively, the computing apparatus 100 confirms the number of images for which learning is performed after the time at which feature grasping of the normal image is completed or the time at which updating of the features of the normal image is completed, and when the confirmed number exceeds a predetermined value, it may be determined that the update criterion is satisfied.
As a result of the determination in step 605, if it is determined that the predetermined criterion is not satisfied, the computing apparatus 100 returns to perform step 602 again, and conversely, if it is determined that the predetermined criterion is satisfied, the computing apparatus 100 performs step 606.
In step 606, the computing device 100 updates the features of the normal image based on the result of learning on the image after the features of the normal image were updated last time. After performing step 606, computing device 100 returns to perform step 602 again.
A specific method by which the computing apparatus 100 determines whether or not the preparation criterion is satisfied at step 403 in fig. 4 will be described below with reference to fig. 7 and 8.
Referring to fig. 7, in step 701, the computing apparatus 100 confirms the time until learning is performed. In step 702, the computing arrangement 100 determines whether the confirmed time exceeds a predetermined value, and if so, executes step 404 of fig. 4, and if not, executes step 401 of fig. 4 again.
Referring to fig. 8, in step 801, the computing apparatus 100 confirms the number of images on which learning has been performed so far. In step 802, the computing device 100 determines whether the number of confirmations exceeds a predetermined value, and if so, executes step 404 of fig. 4, and if not, executes step 401 of fig. 4 again.
As described above, the computing apparatus 100 can grasp the standard for testing by performing learning on an image until the preparation standard is satisfied, and perform testing based on the test standard grasped from the learning result of the previous image while performing learning on the subsequent image.
In the case according to the embodiment explained above, both learning and testing are performed on a plurality of images, wherein the testing is performed by reflecting the learning result, so that it is possible to reflect the learning result in real time and perform the testing, and therefore, the effect of improving the testing accuracy by adaptively responding to environmental changes can be expected.
In addition, since it is not necessary to separately prepare images for learning, it is possible to reduce the time and effort required to surely take images of defective products or sort images.
The term "section" used in the above embodiments refers to a hardware component such as software or a Field Programmable Gate Array (FPGA) or an Application Specific Integrated Circuit (ASIC), and the "section" may perform a certain role. However, the section is not limited to software or hardware. The "parts" may be configured to reside on the addressable storage medium and may be configured to run one or more processors. Therefore, the "section" includes, as an example: components such as software components, object-oriented software components, class components, and task components; proceeding; a function; an attribute; carrying out a procedure; a sub-routine; a program specific code segment; a driver; firmware; a microcode; a circuit; data; a database; a data structure; table; an array; and a variable.
The functions provided in the components and the "parts" may be combined into a smaller number of constituent elements and "parts" or separated from additional components and "parts".
Furthermore, the components and "—" may also be implemented as one or more central processors in a running device or secure multimedia card.
The machine learning and test execution methods according to the embodiments described by fig. 3 to 8 may also be implemented in the form of a computer-readable medium storing instructions and data executable by a computer. At this time, the instructions and data may be stored in the form of program code, which, when executed by a processor, may generate predetermined program modules to perform predetermined operations. Also, computer readable media can be any available media that can be accessed by the computer and includes both volatile and nonvolatile media, removable and non-removable media. Also, computer readable media may be computer recording media including volatile and nonvolatile, removable and non-removable media implemented in any method or technology for storage of information such as computer readable instructions, data structures, program modules or other data. For example, the computer recording medium may be a magnetic storage medium such as a Hard Disk Drive (HDD) and a Solid State Disk (SSD); optical recording media such as Compact Discs (CDs), Digital Video Discs (DVDs), and blu-ray discs; or a memory included in a server accessible through a network.
Furthermore, the machine learning and test execution method according to the embodiments described by fig. 3 to 8 may also be implemented as a computer program (or computer program product) comprising instructions executable by a computer. The computer program includes programmable machine instructions that are processed by the processor and may be implemented in a High-level Programming Language (High-level Programming Language), an Object-oriented Programming Language (Object-oriented Programming Language), an assembly Language, or a machine Language. Further, the computer program may be recorded on a tangible computer-readable recording medium (e.g., a memory, a hard disk, a magnetic/optical medium, or a Solid-State Drive (SSD), etc.).
Thus, the machine learning and test execution method according to the embodiments described by fig. 3 to 8 may be implemented by executing a computer program as described above by a computing device. The computing device may also include at least a portion of a processor, memory, a storage device, a high-speed interface connected to the memory and the high-speed expansion port, and a low-speed interface connected to the low-speed bus and the storage device. Each of these components is interconnected using various buses, and may be mounted on a common motherboard or in any other suitable manner.
Among other things, the processor may process instructions in the computing device, which may include instructions stored in the memory or storage device to display graphical information for providing a Graphical User Interface (GUI) on an external input or output device, such as a display connected to a high speed Interface. For example, as another embodiment, multiple processors and/or multiple buses may be used, as appropriate, along with multiple memories and memory modalities. Also, the processor may be implemented as a chipset of chips that include multiple separate analog and/or digital processors.
Also, the memory stores information in the computing device. As an example, the memory may be configured as a volatile memory unit or a collection thereof. As another example, the memory may be configured as a non-volatile memory unit or collection thereof. Also, the memory may be another form of computer-readable medium, such as a magnetic or optical disk.
Furthermore, the storage device may provide a large amount of storage space for the computing device. The Storage device may be a computer-readable medium or a component containing such a medium, and may also include, for example, a device in a Storage Area Network (SAN) or other component, and may be a floppy disk device, a hard disk device, an optical disk device, or a tape device, a flash memory, or other similar semiconductor Storage device or device array.
The above-described embodiments are for illustration purposes, and those skilled in the art will appreciate that the above-described embodiments can be easily modified into other specific forms without changing the technical idea or essential features thereof. It is therefore to be understood that the above embodiments are illustrative and not restrictive in all respects. For example, each component described as singular can be implemented in distributed form, and components described as distributed can also be implemented in combined form.
The scope intended to be protected by the present specification is indicated by the appended claims, rather than the foregoing detailed description, and is to be understood as including all changes and modifications that are derived from the meaning and scope of the claims, and equivalents thereof.
Claims (14)
1. A method of performing machine learning and testing on a plurality of images,
the method comprises the following steps:
receiving a plurality of images, an
Performing learning and testing on the plurality of images;
the test is performed by reflecting the result of performing the learning.
2. The method of performing machine learning and testing on a plurality of images of claim 1,
the step of performing the test comprises the steps of:
learning the received images one by one while determining whether a predetermined first criterion is satisfied, an
When the judgment result is that the first standard is met, learning and testing are carried out on the rest images;
the test is performed by reflecting the learning result performed until the first criterion is satisfied and the learning result for the remaining images.
3. The method of performing machine learning and testing on a plurality of images of claim 2,
a step of performing a test by reflecting the result of the learning that has been performed, comprising the steps of:
grasping the feature of a normal image based on a result of learning performed until the first criterion is satisfied,
judging whether the first image is a normal image based on the grasped features of the normal image,
performing learning on the first image, an
Updating the feature of the normal image based on a result of performing learning on the first image.
4. The method of performing machine learning and testing on a plurality of images of claim 2,
a step of performing a test by reflecting the result of the learning that has been performed, comprising the steps of:
grasping the feature of a normal image based on a result of learning performed until the first criterion is satisfied,
judging whether the first image is a normal image or not based on the grasped features of the normal image;
performing learning on the first image and performing learning on the first image,
determining whether a predetermined second criterion is fulfilled, an
When the determination result is that the second criterion is satisfied, the feature of the normal image is updated based on the result of performing learning on the first image, and when the determination result is that the second criterion is not satisfied, testing and learning are performed on the next image based on the feature of the normal image.
5. The method of performing machine learning and testing on a plurality of images of claim 2,
the step of judging whether the first criterion is satisfied comprises the following steps:
confirming the time at which learning has been performed so far, and
determining that the first criterion is satisfied when the confirmed time exceeds a predetermined time.
6. The method of performing machine learning and testing on a plurality of images of claim 2,
the step of judging whether the first criterion is satisfied comprises the following steps:
confirming the number of images on which learning has been performed so far, and
determining that the first criterion is satisfied when the confirmed number exceeds a predetermined number.
7. A computer-readable recording medium recording a program for executing the method of performing machine learning and testing on a plurality of images according to claim 1.
8. A computer program executed by an apparatus for performing machine learning and testing and stored in a medium to perform the method of performing machine learning and testing on a plurality of images according to claim 1.
9. An apparatus for performing machine learning and testing,
the method comprises the following steps:
an input/output section that receives operations and data inputs related to machine learning and testing and displays data processing results,
a communication part communicating with an external device to transmit and receive data,
a storage part storing a program for executing machine learning and testing, an
A control section that performs machine learning and testing on a plurality of images by executing the program;
the control section performs learning and testing on the plurality of images received through the communication section;
the control portion executes the test by reflecting a result of executing the learning.
10. The apparatus for performing machine learning and testing according to claim 9,
in the course of the execution of the test,
the control section that determines whether or not a predetermined first criterion is satisfied while learning the received images one by one, and performs learning and testing on the remaining images when the determination result is that the first criterion is satisfied,
the control section executes a test by reflecting a learning result executed until the first criterion is satisfied and a learning result for the remaining image.
11. The apparatus for performing machine learning and testing according to claim 10,
when the test is performed by reflecting the result of the learning that has been performed,
the control unit grasps a feature of a normal image based on a result of learning performed until the first criterion is satisfied, determines whether or not a first image is a normal image based on the grasped feature of the normal image, performs learning on the first image, and updates the feature of the normal image based on a result of performing learning on the first image.
12. The apparatus for performing machine learning and testing according to claim 10,
when the test is performed by reflecting the result of the learning that has been performed,
the control unit grasps a feature of a normal image based on a learning result executed until the first criterion is satisfied, and determines whether or not the first image is a normal image based on the grasped feature of the normal image; the learning is performed on the first image, and then it is determined whether or not a predetermined second criterion is satisfied, and when the determination result is that the second criterion is satisfied, the feature of the normal image is updated based on the result of performing the learning on the first image, and when the determination result is that the second criterion is not satisfied, the test and the learning are performed on the next image based on the feature of the normal image.
13. The apparatus for performing machine learning and testing according to claim 10,
in determining whether the first criterion is met,
the control portion confirms a time for which learning has been performed so far, and determines that the first criterion is satisfied when the confirmed time exceeds a predetermined time.
14. The apparatus for performing machine learning and testing according to claim 10,
in determining whether the first criterion is met,
the control section confirms the number of images for which learning has been performed so far, and determines that the first criterion is satisfied when the confirmed number exceeds a predetermined number.
Applications Claiming Priority (3)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020190111843A KR102357729B1 (en) | 2019-09-10 | 2019-09-10 | Method and apparatus for performing machine learning and test on plurality of images |
| KR10-2019-0111843 | 2019-09-10 | ||
| PCT/KR2020/000251 WO2021049721A1 (en) | 2019-09-10 | 2020-01-07 | Method and device for performing machine learning and testing on plurality of images |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| CN113906471A true CN113906471A (en) | 2022-01-07 |
Family
ID=74866737
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| CN202080041294.6A Pending CN113906471A (en) | 2019-09-10 | 2020-01-07 | Method and apparatus for performing machine learning and testing on multiple images |
Country Status (3)
| Country | Link |
|---|---|
| KR (1) | KR102357729B1 (en) |
| CN (1) | CN113906471A (en) |
| WO (1) | WO2021049721A1 (en) |
Family Cites Families (8)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| KR100327101B1 (en) * | 1999-08-11 | 2002-03-18 | 김주환 | An automatic teaching method in marking inspection process of a semiconductor packages |
| JP4912028B2 (en) * | 2006-05-01 | 2012-04-04 | 日本電信電話株式会社 | Sequential learning type non-stationary video detection apparatus, sequential learning type non-stationary video detection method, and program implementing the method |
| US10878286B2 (en) * | 2016-02-24 | 2020-12-29 | Nec Corporation | Learning device, learning method, and recording medium |
| KR101842347B1 (en) * | 2016-05-04 | 2018-03-26 | 두산중공업 주식회사 | Plant system, and fault detecting method thereof |
| JP6897042B2 (en) * | 2016-09-27 | 2021-06-30 | 日本電気株式会社 | Image inspection equipment, image inspection method and image inspection program |
| JP6942472B2 (en) * | 2017-01-13 | 2021-09-29 | キヤノン株式会社 | Video recognition device, video recognition method and program |
| JP7074460B2 (en) * | 2017-11-10 | 2022-05-24 | アズビル株式会社 | Image inspection equipment and methods |
| KR101838664B1 (en) * | 2017-11-28 | 2018-03-14 | 서울전선 주식회사 | Apparatus for inspecting surface of cable |
-
2019
- 2019-09-10 KR KR1020190111843A patent/KR102357729B1/en active IP Right Grant
-
2020
- 2020-01-07 CN CN202080041294.6A patent/CN113906471A/en active Pending
- 2020-01-07 WO PCT/KR2020/000251 patent/WO2021049721A1/en active Application Filing
Also Published As
| Publication number | Publication date |
|---|---|
| KR102357729B1 (en) | 2022-02-03 |
| KR20210030576A (en) | 2021-03-18 |
| WO2021049721A1 (en) | 2021-03-18 |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| US11189019B2 (en) | Method for detecting defects, electronic device, and computer readable medium | |
| CN110378966B (en) | Method, device and equipment for calibrating external parameters of vehicle-road coordination phase machine and storage medium | |
| US11521312B2 (en) | Image processing apparatus, image processing method, and storage medium | |
| CN102789405A (en) | Automated testing method and system for mainboard | |
| CN110969600A (en) | Product defect detection method and device, electronic equipment and storage medium | |
| US20230096532A1 (en) | Machine learning system, learning data collection method and storage medium | |
| WO2019126693A1 (en) | Automated analysis of analytical gels and blots | |
| CN112712498A (en) | Vehicle damage assessment method and device executed by mobile terminal, mobile terminal and medium | |
| CN114359160B (en) | Screen detection method and device, electronic equipment and storage medium | |
| JP7204786B2 (en) | Visual search method, device, computer equipment and storage medium | |
| CN109256341B (en) | Method and device for inspecting monocrystalline silicon solar cells | |
| US11430201B2 (en) | Image processing apparatus for character recognition, control method of the same, storage medium and image processing system | |
| CN110716778B (en) | Application compatibility testing method, device and system | |
| CN113632140B (en) | Automatic learning method and system for product inspection | |
| JP7440823B2 (en) | Information processing device, information processing method and program | |
| CN113906471A (en) | Method and apparatus for performing machine learning and testing on multiple images | |
| US20240142935A1 (en) | Method for detecting workpiece based on homogeneous multi-core architecture and edge computing device | |
| CN109087236B (en) | Industrial image detection control method and system for multi-deep learning platform | |
| CN110826504A (en) | Visual identification method and device combined with fine-grained classification and computer equipment | |
| US20230137658A1 (en) | Data processing apparatus and method for controlling data processing apparatus | |
| US9563896B1 (en) | Kinetic tracking in manufacturing to predict and prevent defects | |
| KR20190082336A (en) | Method for detecting pedestrian from video, and apparatus for performing the same | |
| TWI770561B (en) | Product defect detection method, computer device and storage medium | |
| WO2024023948A1 (en) | Analysis device, analysis method, and analysis program | |
| US20240104908A1 (en) | Evaluation method |
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 |