KR20220095468A - Steel quality inspection system based on xai and large scale deep learning hpc system - Google Patents
Steel quality inspection system based on xai and large scale deep learning hpc system Download PDFInfo
- Publication number
- KR20220095468A KR20220095468A KR1020200187002A KR20200187002A KR20220095468A KR 20220095468 A KR20220095468 A KR 20220095468A KR 1020200187002 A KR1020200187002 A KR 1020200187002A KR 20200187002 A KR20200187002 A KR 20200187002A KR 20220095468 A KR20220095468 A KR 20220095468A
- Authority
- KR
- South Korea
- Prior art keywords
- artificial intelligence
- deep learning
- quality inspection
- scale deep
- steel quality
- Prior art date
Links
Images
Classifications
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8803—Visual inspection
-
- 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
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8854—Grading and classifying of flaws
- G01N2021/888—Marking defects
-
- G—PHYSICS
- G01—MEASURING; TESTING
- G01N—INVESTIGATING OR ANALYSING MATERIALS BY DETERMINING THEIR CHEMICAL OR PHYSICAL PROPERTIES
- G01N21/00—Investigating or analysing materials by the use of optical means, i.e. using sub-millimetre waves, infrared, visible or ultraviolet light
- G01N21/84—Systems specially adapted for particular applications
- G01N21/88—Investigating the presence of flaws or contamination
- G01N21/8851—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges
- G01N2021/8887—Scan or image signal processing specially adapted therefor, e.g. for scan signal adjustment, for detecting different kinds of defects, for compensating for structures, markings, edges based on image processing techniques
-
- Y—GENERAL TAGGING OF NEW TECHNOLOGICAL DEVELOPMENTS; GENERAL TAGGING OF CROSS-SECTIONAL TECHNOLOGIES SPANNING OVER SEVERAL SECTIONS OF THE IPC; TECHNICAL SUBJECTS COVERED BY FORMER USPC CROSS-REFERENCE ART COLLECTIONS [XRACs] AND DIGESTS
- Y02—TECHNOLOGIES OR APPLICATIONS FOR MITIGATION OR ADAPTATION AGAINST CLIMATE CHANGE
- Y02P—CLIMATE CHANGE MITIGATION TECHNOLOGIES IN THE PRODUCTION OR PROCESSING OF GOODS
- Y02P90/00—Enabling technologies with a potential contribution to greenhouse gas [GHG] emissions mitigation
- Y02P90/30—Computing systems specially adapted for manufacturing
Landscapes
- Engineering & Computer Science (AREA)
- Physics & Mathematics (AREA)
- General Health & Medical Sciences (AREA)
- General Physics & Mathematics (AREA)
- Health & Medical Sciences (AREA)
- Life Sciences & Earth Sciences (AREA)
- Pathology (AREA)
- Analytical Chemistry (AREA)
- Biochemistry (AREA)
- Chemical & Material Sciences (AREA)
- Theoretical Computer Science (AREA)
- Immunology (AREA)
- Data Mining & Analysis (AREA)
- Molecular Biology (AREA)
- Biomedical Technology (AREA)
- Biophysics (AREA)
- Computational Linguistics (AREA)
- Computer Vision & Pattern Recognition (AREA)
- Evolutionary Computation (AREA)
- Artificial Intelligence (AREA)
- Computing Systems (AREA)
- General Engineering & Computer Science (AREA)
- Mathematical Physics (AREA)
- Software Systems (AREA)
- Signal Processing (AREA)
- Image Analysis (AREA)
- Investigating Materials By The Use Of Optical Means Adapted For Particular Applications (AREA)
Abstract
Description
본 발명은 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템에 관한 것이다. The present invention relates to a steel quality inspection system based on explainable artificial intelligence and large-scale deep learning HPC system.
종래 기술에 따르면, 강재에 포함된 균열 및 결함을 검출하기 위해 해당분야에 고도의 전문지식을 보유하고 오랜 기간 품질관리 경험이 축적된 전문가가 육안으로 강재 표면을 관찰하며 결함부위를 검출하였으나, 대형 중장비에 사용되는 강판의 경우 면적이 넓어 검사시간이 매우 오래 걸리고 검사자의 컨디션과 피로도에 따라 다른 결과가 도출되며 제안된 시간 내 많은 물량을 검사할 경우 검출 정확도가 저하되는 문제점이 있다.According to the prior art, in order to detect cracks and defects contained in steel, an expert who has a high degree of expertise in the field and has accumulated quality control experience for a long time visually observed the steel surface and detected the defect, but In the case of steel plates used for heavy equipment, the inspection time is very long because the area is large, and different results are derived depending on the condition and fatigue of the inspector, and there is a problem in that the detection accuracy is lowered when a large quantity is inspected within the suggested time.
또한 최근 스마트팩토리를 중심으로 본격적으로 도입되기 시작된 머신비전기술을 활용한 품질검사도 육안검사에 비해 검출성능이 향상되었으나 조명 등 작업환경의 변화에 매우 취약하고 일정수준 이상 정확도를 향상시킬 수 없는 등 한계를 보이고 있다. In addition, quality inspection using machine vision technology, which has recently been introduced in earnest around smart factories, has improved detection performance compared to visual inspection, but it is very vulnerable to changes in the work environment such as lighting and cannot improve accuracy beyond a certain level. showing limitations.
또한, 알고리즘의 결함검출에 대한 오류 발생시 알고리즘의 결과를 추론한 과정을 알 수 없는 블랙박스 문제로 오류에 대한 정확한 원인규명을 할 수가 없어 알고리즘의 성능향상에 한계가 있다.In addition, when an error occurs in the detection of a defect in the algorithm, the exact cause of the error cannot be identified due to a black box problem in which the process of inferring the result of the algorithm cannot be known, so there is a limit to the performance improvement of the algorithm.
본 발명은 전술한 문제점을 해결하기 위해 제안된 것으로, 결함의 종류에 상관없이 획일적인 기존 알고리즘의 검출정확도 저하 문제점을 해결하고, 2D 영상을 입력으로 받는 딥러닝 알고리즘, 3D 포인트 클라우드를 입력으로 받는 Point CNN, 설명능력이 있는 모듈화된 XAI 알고리즘 기반으로 강재의 균열 및 결함의 종류에 따라 최적화된 알고리즘을 적용하여 최적의 조건에서 결함을 검출하고 결함의 정확한 위치, 크기, 형상의 표현이 가능한 시스템을 제공하는데 그 목적이 있다.The present invention has been proposed to solve the above problems, solves the problem of lowering the detection accuracy of the existing uniform algorithm regardless of the type of defect, and receives a deep learning algorithm that receives a 2D image as an input, and a 3D point cloud as an input Based on the Point CNN, a modularized XAI algorithm with explanatory ability, an algorithm optimized according to the type of cracks and defects in steel is applied to detect defects under optimal conditions and to develop a system that can express the exact location, size, and shape of defects. Its purpose is to provide
본 발명에 따른 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템은 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 프로그램이 저장된 메모리 및 상기 프로그램을 실행시키는 프로세서를 포함하되, 상기 프로세서는 강재의 균열 및 결함 정보를 영상을 통해 획득하고, 이를 인공지능 기반으로 검출하고 정량화한다. The explainable artificial intelligence and large-scale deep learning HPC system-based steel quality inspection system according to the present invention includes a memory in which an explainable artificial intelligence and large-scale deep learning HPC system-based steel quality inspection program is stored and a processor for executing the program, The processor acquires information about cracks and defects of steel through images, and detects and quantifies them based on artificial intelligence.
상기 프로세서는 이동식 스테레오 이미징 기반으로 영상을 획득한다. The processor acquires an image based on mobile stereo imaging.
상기 프로세서는 설명가능 인공지능 기반으로 상기 균열 및 결함을 검출한다. The processor detects the cracks and defects based on explainable artificial intelligence.
상기 프로세서는 상기 균열 및 결함 정보를 시각화하고, 도면화한다. The processor visualizes and plots the crack and defect information.
상기 프로세서는 검출을 위한 인공지능 알고리즘 트레이닝을 대규모 딥러닝 HPC 시스템 기반으로 수행한다. The processor performs artificial intelligence algorithm training for detection based on a large-scale deep learning HPC system.
상기 프로세서는 벤치마크 데이터셋으로 객체에 대한 Heatmap 및 Captioning 성능을 향상시킨 후 트랜스퍼 러닝을 통해 가중치행렬을 이전한다. The processor transfers a weight matrix through transfer learning after improving heatmap and captioning performance for an object with a benchmark dataset.
상기 프로세서는 가설기자재에 특화된 인공지능 필터로 오염된 가설기자재에 포함된 균열 및 결함 검출을 수행한다. The processor performs the detection of cracks and defects included in the temporary equipment contaminated with an artificial intelligence filter specialized for the temporary equipment.
본 발명에 따르면, 2D 영상을 입력으로 받는 딥러닝 알고리즘, 3D 포인트 클라우드를 입력으로 받는 Point CNN, 설명능력이 있는 모듈화된 XAI 알고리즘 기반으로 강재의 균열 및 결함의 종류에 따라 최적화된 알고리즘을 적용하여 최적의 조건에서 결함을 검출하고 결함의 정확한 위치, 크기, 형상의 표현이 가능하다. According to the present invention, an algorithm optimized according to the type of cracks and defects in steel is applied based on a deep learning algorithm that receives a 2D image as an input, a Point CNN that receives a 3D point cloud as an input, and a modularized XAI algorithm with explanatory ability. It is possible to detect defects under optimal conditions and to express the exact location, size, and shape of the defect.
결함의 크기와 관계없이 2D 영상 기반 알고리즘으로 검출이 가능한 불순물, 화학적 부산물, 긁힘, 녹발생과 결함의 크기에 따라 2D 또는 3D 기반으로 검출 가능한 눌림과 패임, 그리고 3D 기반 알고리즘으로 검출이 가능한 강재의 변형, 파손, 마모를 구분하여 최적의 알고리즘을 적용하여 검출성능을 향상시키는 것이 가능한 효과가 있다. Impurities, chemical by-products, scratches, rust, and dents that can be detected with a 2D image-based algorithm regardless of the size of the defect, and presses and dents that can be detected on a 2D or 3D basis depending on the size of the defect, and of steel that can be detected with a 3D-based algorithm It is possible to improve the detection performance by applying the optimal algorithm by classifying deformation, breakage, and wear.
또한 XAI 알고리즘의 설명기능을 이용해 결함검출 결과를 Heatmap과 Captioning으로 시각적으로 출력하여 알고리즘의 추론과정과 분류결과의 일관성을 확인할 수 있고 사용자가 추론과정의 오류를 인지하여 알고리즘의 성능향상이 가능한 효과가 있다.In addition, by using the explanatory function of the XAI algorithm, the defect detection result can be visually output as heatmap and captioning to check the consistency of the algorithm's reasoning process and classification result. have.
본 발명의 효과는 이상에서 언급한 것들에 한정되지 않으며, 언급되지 아니한 다른 효과들은 아래의 기재로부터 당업자에게 명확하게 이해될 수 있을 것이다.Effects of the present invention are not limited to those mentioned above, and other effects not mentioned will be clearly understood by those skilled in the art from the following description.
도 1은 본 발명의 실시예에 따른 강재표면에 발생한 결함을 검출하는 설명가능 인공지능 알고리즘 구성도를 도시한다.
도 2는 본 발명의 실시예에 따른 강재표면에 발생한 결함의 종류를 도시한다.
도 3은 본 발명의 실시예에 따른강재에 발생한 입체적인 결함의 종류를 도시한다.
도 4는 본 발명의 실시예에 따른 강재표면에 발생한 결함을 검출 및 정량화하는 인공지능 알고리즘 순서도를 도시한다.
도 5는 본 발명의 실시예에 따른 강재표면에 발생한 결함(화학부산물)의 특징을 설명하는 Captioning 출력결과를 도시한다.
도 6은 본 발명의 실시예에 따른 강재표면에 발생한 결함(긁힘)의 특징을 설명하는 Captioning 출력결과를 도시한다.
도 7은 본 발명의 실시예에 따른 강재표면에 발생한 결함의 특징을 가중치별로 표현하는 히트맵 출력결과를 도시한다.
도 8은 본 발명의 실시예에 따른 히트맵 성능을 높이기 위한 벤치마크 데이터셋 트레이닝을 도시한다.
도 9는 본 발명의 실시예에 따른Captioning 성능을 높이기 위한 벤치마크 데이터셋 트레이닝을 도시한다.
도 10은 본 발명의 실시예에 따른 스테레오이미징 기술 기반으로 결함을 입체화하는 프로세스를 도시한다.
도 11은 본 발명의 실시예에 따른 대규모 딥러닝 HPC 시스템을 도시한다. 1 is a diagram illustrating an explainable artificial intelligence algorithm for detecting defects occurring on a steel surface according to an embodiment of the present invention.
Figure 2 shows the types of defects that occur on the surface of the steel according to the embodiment of the present invention.
Figure 3 shows the types of three-dimensional defects occurred in the steel according to the embodiment of the present invention.
4 is a flowchart of an artificial intelligence algorithm for detecting and quantifying defects occurring on the steel surface according to an embodiment of the present invention.
5 shows the captioning output results for explaining the characteristics of defects (chemical by-products) occurring on the surface of the steel according to the embodiment of the present invention.
6 shows the captioning output results for explaining the characteristics of defects (scratches) occurring on the surface of the steel according to the embodiment of the present invention.
7 shows a heat map output result expressing the characteristics of defects occurring on the surface of a steel material according to an embodiment of the present invention by weight.
8 shows a benchmark dataset training for improving heat map performance according to an embodiment of the present invention.
9 shows a benchmark dataset training for improving captioning performance according to an embodiment of the present invention.
10 shows a process for stereoscopicizing a defect based on a stereo imaging technique according to an embodiment of the present invention.
11 shows a large-scale deep learning HPC system according to an embodiment of the present invention.
본 발명의 전술한 목적 및 그 이외의 목적과 이점 및 특징, 그리고 그것들을 달성하는 방법은 첨부되는 도면과 함께 상세하게 후술되어 있는 실시예들을 참조하면 명확해질 것이다. The above and other objects, advantages and features of the present invention, and a method for achieving them will become apparent with reference to the embodiments described below in detail in conjunction with the accompanying drawings.
그러나 본 발명은 이하에서 개시되는 실시예들에 한정되는 것이 아니라 서로 다른 다양한 형태로 구현될 수 있으며, 단지 이하의 실시예들은 본 발명이 속하는 기술분야에서 통상의 지식을 가진 자에게 발명의 목적, 구성 및 효과를 용이하게 알려주기 위해 제공되는 것일 뿐으로서, 본 발명의 권리범위는 청구항의 기재에 의해 정의된다. However, the present invention is not limited to the embodiments disclosed below, but may be implemented in various different forms, and only the following examples are provided to those of ordinary skill in the art to which the present invention pertains. It is only provided to easily inform the composition and effect, and the scope of the present invention is defined by the description of the claims.
한편, 본 명세서에서 사용된 용어는 실시예들을 설명하기 위한 것이며 본 발명을 제한하고자 하는 것은 아니다. 본 명세서에서, 단수형은 문구에서 특별히 언급하지 않는 한 복수형도 포함한다. 명세서에서 사용되는 "포함한다(comprises)" 및/또는 "포함하는(comprising)"은 언급된 구성소자, 단계, 동작 및/또는 소자가 하나 이상의 다른 구성소자, 단계, 동작 및/또는 소자의 존재 또는 추가됨을 배제하지 않는다.On the other hand, the terms used herein are for the purpose of describing the embodiments and are not intended to limit the present invention. In this specification, the singular also includes the plural unless otherwise specified in the phrase. As used herein, “comprises” and/or “comprising” means that a referenced element, step, operation and/or element is the presence of one or more other elements, steps, operations and/or elements. or added.
본 발명의 실시예에 따르면, 기존 딥러닝 알고리즘의 문제점을 해결하기 위하여 제안된 것으로, XAI, 포인트클라우드, HPC 시스템 기반으로 강재의 균열 및 결함 검출을 수행함으로써, 2D 기반 딥 뉴럴 네트워크의 단점인 성능향상의 한계, 추론이유 설명불가, 결함의 입체화 불가, 긴 연산시간 문제를 해결하는 것이 가능하다. According to an embodiment of the present invention, it is proposed to solve the problems of the existing deep learning algorithm, and by performing crack and defect detection of steel based on XAI, point cloud, and HPC systems, the performance, which is a disadvantage of 2D-based deep neural networks It is possible to solve the problems of limitation of improvement, reason of reasoning inexplicable, impossibility of three-dimensionalization of defects, and long operation time.
도 1은 본 발명의 실시예에 따른 강재표면에 발생한 결함을 검출하는 설명가능 인공지능 알고리즘 구성도를 도시하고, 도 2는 본 발명의 실시예에 따른 강재표면에 발생한 결함의 종류를 도시하고, 도 3은 본 발명의 실시예에 따른강재에 발생한 입체적인 결함의 종류를 도시하고, 도 4는 본 발명의 실시예에 따른 강재표면에 발생한 결함을 검출 및 정량화하는 인공지능 알고리즘 순서도를 도시하고, 도 5는 본 발명의 실시예에 따른 강재표면에 발생한 결함(화학부산물)의 특징을 설명하는 Captioning 출력결과를 도시하고, 도 6은 본 발명의 실시예에 따른 강재표면에 발생한 결함(긁힘)의 특징을 설명하는 Captioning 출력결과를 도시하고, 도 7은 본 발명의 실시예에 따른 강재표면에 발생한 결함의 특징을 가중치별로 표현하는 히트맵 출력결과를 도시하고, 도 8은 본 발명의 실시예에 따른 히트맵 성능을 높이기 위한 벤치마크 데이터셋 트레이닝을 도시하고, 도 9는 본 발명의 실시예에 따른Captioning 성능을 높이기 위한 벤치마크 데이터셋 트레이닝을 도시하고, 도 10은 본 발명의 실시예에 따른 스테레오이미징 기술 기반으로 결함을 입체화하는 프로세스를 도시하며, 도 11은 본 발명의 실시예에 따른 대규모 딥러닝 HPC 시스템을 도시한다. 1 shows an explanatory artificial intelligence algorithm configuration diagram for detecting defects occurring on a steel surface according to an embodiment of the present invention, and FIG. 2 shows the types of defects occurring on a steel surface according to an embodiment of the present invention, Figure 3 shows the type of three-dimensional defects occurred in the steel material according to an embodiment of the present invention, Figure 4 shows a flowchart of an artificial intelligence algorithm for detecting and quantifying defects occurring on the surface of the steel material according to an embodiment of the present invention, Fig. 5 shows a captioning output result explaining the characteristics of defects (chemical by-products) occurring on the steel surface according to an embodiment of the present invention, and FIG. 6 is a feature of defects (scratches) occurring on the steel surface according to an embodiment of the present invention shows the output results of Captioning for explaining, FIG. 7 shows the heat map output results expressing the characteristics of defects occurring on the surface of steel according to an embodiment of the present invention by weight, and FIG. 8 is an embodiment of the present invention Figure 9 shows a benchmark dataset training to increase the heat map performance, Figure 9 shows a benchmark dataset training to increase the captioning performance according to an embodiment of the present invention, Figure 10 is a stereo according to an embodiment of the present invention Fig. 11 shows a large-scale deep-learning HPC system according to an embodiment of the present invention, showing a process for three-dimensionalizing defects based on imaging technology.
강재에 발생하는 입체적인 결함은 강재의 균열, 눌림, 화학적 부산물 등 표면결함과는 달리 결함부위와 배경의 색상차이가 없고, 경계가 뚜렷하지 않아 영상 및 2D 기반 인공지능 알고리즘으로 해당결함을 검출하기가 매우 어렵다. Unlike surface defects such as cracks, crushing, and chemical by-products in steel, three-dimensional defects in steel do not have a color difference between the defect area and the background, and the boundaries are not clear. very difficult
본 발명의 실시예에 따르면, 강재의 변형, 파손, 마모 등 강재의 표면결함과 구분되고 입체적으로 표현할 수 있는 결함에 대해 두 장의 영상을 겹쳐 촬영해 겹침 영역의 특징을 추출하고 이 특징 기반으로 두 장의 영상을 매칭한 후 SfM(Structure from Motion) 알고리즘 기반으로 포인트클라우드로 입체화하는 스테레오 이미징 기법을 적용한다. According to the embodiment of the present invention, two images are superimposed for defects that can be differentiated and expressed three-dimensionally from the surface defects of steel, such as deformation, breakage, and abrasion of steel, to extract the features of the overlapping area, and based on these features, After matching the image of the intestine, a stereo imaging technique that three-dimensionalizes it as a point cloud based on the SfM (Structure from Motion) algorithm is applied.
미세한 변형 등 결함의 정도가 크지 않아 입체화가 어려운 경우 2D 기반 인공지능 알고리즘의 검출결과와 합성히여 정확도를 향상시키는 것이 가능하다. When three-dimensionalization is difficult because the degree of defects such as minute deformation is not large, it is possible to improve the accuracy by synthesizing it with the detection result of a 2D-based artificial intelligence algorithm.
본 발명의 실시예에 따른 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템은 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 프로그램이 저장된 메모리 및 상기 프로그램을 실행시키는 프로세서를 포함하되, 상기 프로세서는 강재의 균열 및 결함 정보를 영상을 통해 획득하고, 이를 인공지능 기반으로 검출하고 정량화한다. The explainable artificial intelligence and large-scale deep learning HPC system-based steel quality inspection system according to an embodiment of the present invention includes a memory in which the explainable artificial intelligence and large-scale deep learning HPC system-based steel quality inspection program is stored and a processor for executing the program However, the processor acquires the crack and defect information of the steel through an image, and detects and quantifies it based on artificial intelligence.
상기 프로세서는 이동식 스테레오 이미징 기반으로 영상을 획득한다. The processor acquires an image based on mobile stereo imaging.
상기 프로세서는 설명가능 인공지능 기반으로 상기 균열 및 결함을 검출한다. The processor detects the cracks and defects based on explainable artificial intelligence.
상기 프로세서는 상기 균열 및 결함 정보를 시각화하고, 도면화한다. The processor visualizes and plots the crack and defect information.
상기 프로세서는 검출을 위한 인공지능 알고리즘 트레이닝을 대규모 딥러닝 HPC 시스템 기반으로 수행한다. The processor performs artificial intelligence algorithm training for detection based on a large-scale deep learning HPC system.
상기 프로세서는 벤치마크 데이터셋으로 객체에 대한 Heatmap 및 Captioning 성능을 향상시킨 후 트랜스퍼 러닝을 통해 가중치행렬을 이전한다. The processor transfers a weight matrix through transfer learning after improving heatmap and captioning performance for an object with a benchmark dataset.
상기 프로세서는 가설기자재에 특화된 인공지능 필터로 오염된 가설기자재에 포함된 균열 및 결함 검출을 수행한다. The processor performs the detection of cracks and defects included in the temporary equipment contaminated with an artificial intelligence filter specialized for the temporary equipment.
Claims (7)
상기 프로그램을 실행시키는 프로세서를 포함하되,
상기 프로세서는 강재의 균열 및 결함 정보를 영상을 통해 획득하고, 이를 인공지능 기반으로 검출하고 정량화하는 것
인 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템.
Memory in which a steel quality inspection program based on explainable artificial intelligence and large-scale deep learning HPC system is stored; and
A processor for executing the program,
The processor acquires the crack and defect information of the steel through an image, and detects and quantifies it based on artificial intelligence
A steel quality inspection system based on explanatory artificial intelligence and large-scale deep learning HPC system.
상기 프로세서는 이동식 스테레오 이미징 기반으로 영상을 획득하는 것
인 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템.
The method of claim 1,
The processor is to acquire an image based on mobile stereo imaging
A steel quality inspection system based on explanatory artificial intelligence and large-scale deep learning HPC system.
상기 프로세서는 설명가능 인공지능 기반으로 상기 균열 및 결함을 검출하는 것
인 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템. The method of claim 1,
The processor detects the cracks and defects based on explainable artificial intelligence.
A steel quality inspection system based on explanatory artificial intelligence and large-scale deep learning HPC system.
상기 프로세서는 상기 균열 및 결함 정보를 시각화하고, 도면화하는 것
인 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템.
The method of claim 1,
The processor visualizes and visualizes the crack and defect information
A steel quality inspection system based on explanatory artificial intelligence and large-scale deep learning HPC system.
상기 프로세서는 검출을 위한 인공지능 알고리즘 트레이닝을 대규모 딥러닝 HPC 시스템 기반으로 수행하는 것
인 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템.
The method of claim 1,
The processor performs artificial intelligence algorithm training for detection based on a large-scale deep learning HPC system
A steel quality inspection system based on explanatory artificial intelligence and large-scale deep learning HPC system.
상기 프로세서는 벤치마크 데이터셋으로 객체에 대한 Heatmap 및 Captioning 성능을 향상시킨 후 트랜스퍼 러닝을 통해 가중치행렬을 이전하는 것
인 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템.
The method of claim 1,
The processor transfers the weight matrix through transfer learning after improving the heatmap and captioning performance for the object with the benchmark dataset
A steel quality inspection system based on explanatory artificial intelligence and large-scale deep learning HPC system.
상기 프로세서는 가설기자재에 특화된 인공지능 필터로 오염된 가설기자재에 포함된 균열 및 결함 검출을 수행하는 것
인 설명가능 인공지능 및 대규모 딥러닝 HPC 시스템 기반 강재 품질검사 시스템. The method of claim 1,
The processor performs the detection of cracks and defects contained in the temporary equipment contaminated with an artificial intelligence filter specialized for the temporary equipment.
A steel quality inspection system based on explanatory artificial intelligence and large-scale deep learning HPC system.
Priority Applications (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020200187002A KR20220095468A (en) | 2020-12-30 | 2020-12-30 | Steel quality inspection system based on xai and large scale deep learning hpc system |
Applications Claiming Priority (1)
| Application Number | Priority Date | Filing Date | Title |
|---|---|---|---|
| KR1020200187002A KR20220095468A (en) | 2020-12-30 | 2020-12-30 | Steel quality inspection system based on xai and large scale deep learning hpc system |
Publications (1)
| Publication Number | Publication Date |
|---|---|
| KR20220095468A true KR20220095468A (en) | 2022-07-07 |
Family
ID=82398295
Family Applications (1)
| Application Number | Title | Priority Date | Filing Date |
|---|---|---|---|
| KR1020200187002A KR20220095468A (en) | 2020-12-30 | 2020-12-30 | Steel quality inspection system based on xai and large scale deep learning hpc system |
Country Status (1)
| Country | Link |
|---|---|
| KR (1) | KR20220095468A (en) |
Cited By (2)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116309447A (en) * | 2023-03-17 | 2023-06-23 | 水利部交通运输部国家能源局南京水利科学研究院 | Dam slope crack detection method based on deep learning |
| CN117218123A (en) * | 2023-11-09 | 2023-12-12 | 上海擎刚智能科技有限公司 | Cold-rolled strip steel wire flying equipment fault detection method and system based on point cloud |
-
2020
- 2020-12-30 KR KR1020200187002A patent/KR20220095468A/en unknown
Cited By (4)
| Publication number | Priority date | Publication date | Assignee | Title |
|---|---|---|---|---|
| CN116309447A (en) * | 2023-03-17 | 2023-06-23 | 水利部交通运输部国家能源局南京水利科学研究院 | Dam slope crack detection method based on deep learning |
| CN116309447B (en) * | 2023-03-17 | 2024-01-05 | 水利部交通运输部国家能源局南京水利科学研究院 | Dam slope crack detection method based on deep learning |
| CN117218123A (en) * | 2023-11-09 | 2023-12-12 | 上海擎刚智能科技有限公司 | Cold-rolled strip steel wire flying equipment fault detection method and system based on point cloud |
| CN117218123B (en) * | 2023-11-09 | 2024-02-02 | 上海擎刚智能科技有限公司 | Cold-rolled strip steel wire flying equipment fault detection method and system based on point cloud |
Similar Documents
| Publication | Publication Date | Title |
|---|---|---|
| CN114937055B (en) | Image self-adaptive segmentation method and system based on artificial intelligence | |
| CN107782733B (en) | Image recognition nondestructive detection device and method for metal surface defects | |
| CN102565072B (en) | Stereoscopic vision on-line detection method for surface crack of stretched aluminum alloy plate | |
| CN105067638B (en) | Tire fetal membrane face character defect inspection method based on machine vision | |
| US10783406B1 (en) | Method for detecting structural surface cracks based on image features and bayesian data fusion | |
| Kothari | Detecting welding defects in steel plates using machine learning and computer vision algorithms | |
| CN102735695B (en) | Rapid lens flaw detection method and apparatus thereof | |
| KR20220095468A (en) | Steel quality inspection system based on xai and large scale deep learning hpc system | |
| CN114445366A (en) | Intelligent long-distance pipeline radiographic image defect identification method based on self-attention network | |
| CN113284109B (en) | Pipeline defect identification method, device, terminal equipment and storage medium | |
| Adem et al. | Defect detection of seals in multilayer aseptic packages using deep learning | |
| Fondevik et al. | Image segmentation of corrosion damages in industrial inspections | |
| Maghami et al. | Automated vision-based inspection of drilled CFRP composites using multi-light imaging and deep learning | |
| Zhao et al. | Recognition of flooding and sinking conditions in flotation process using soft measurement of froth surface level and QTA | |
| CN113838043A (en) | Machine vision-based quality analysis method in metal foil manufacturing | |
| CN117173461A (en) | Multi-visual task filling container defect detection method, system and medium | |
| CN116228651A (en) | Cloth defect detection method, system, equipment and medium | |
| Litvintseva et al. | Real-time steel surface defect recognition based on CNN | |
| Hashmi et al. | Computer-vision based visual inspection and crack detection of railroad tracks | |
| Dong et al. | Automated detection of gear tooth flank surface integrity: A cascade detection approach using machine vision | |
| Wang et al. | Identification and detection of surface defects of outer package printed matter based on machine vision | |
| Molefe et al. | Classification of thermite welding defects using local binary patterns and k nearest neighbors | |
| CN113936291A (en) | Aluminum template quality inspection and recovery method based on machine vision | |
| CN114092396A (en) | Method and device for detecting corner collision flaw of packaging box | |
| Haffner et al. | Weld segmentation for diagnostic and evaluation method |