Predictive model-based quality inspection using Machine Learning and Edge Cloud Computing

Schmitt, Jacqueline; Bönig, Jochen; Borggräfe, Thorbjörn; Beitinger, Gunter; Deuse, Jochen

doi:10.1016/j.aei.2020.101101

Cited by 128 publications

(41 citation statements)

References 61 publications

Supporting

Mentioning

Contrasting

Unclassified

Order By: Relevance

“…We live in an era of machine learning [ 43 ]. However, looking for a processing pipeline for a particular problem solicited from the industry, we often lack a larger dataset.…”

Section: Discussionmentioning

confidence: 99%

Shape-Based Alignment of the Scanned Objects Concerning Their Asymmetric Aspects

Lúčny

Dillinger

Kacurova

et al. 2021

Sensors

View full text Add to dashboard Cite

We introduce an integrated method for processing depth maps measured by a laser profile sensor. It serves for the recognition and alignment of an object given by a single example. Firstly, we look for potential object contours, mainly using the Retinex filter. Then, we select the actual object boundary via shape comparison based on Triangle Area Representation (TAR). We overcome the limitations of the TAR method by extension of its shape descriptor. That is helpful mainly for objects with symmetric shapes but other asymmetric aspects like squares with asymmetric holes. Finally, we use point-to-point pairing, provided by the extended TAR method, to calculate the 3D rigid affine transform that aligns the scanned object to the given example position. For the transform calculation, we design an algorithm that overcomes the Kabsch point-to-point algorithm’s accuracy and accommodates it for a precise contour-to-contour alignment. In this way, we have implemented a pipeline with features convenient for industrial use, namely production inspection.

show abstract

“…We live in an era of machine learning [ 43 ]. However, looking for a processing pipeline for a particular problem solicited from the industry, we often lack a larger dataset.…”

Section: Discussionmentioning

confidence: 99%

Shape-Based Alignment of the Scanned Objects Concerning Their Asymmetric Aspects

Lúčny

Dillinger

Kacurova

et al. 2021

Sensors

View full text Add to dashboard Cite

show abstract

“…Very High Industrial AI can be used for inline monitoring or control due to low inference times, thus managing to cope with the constraints imposed by the takt time. Examples include inline sorting, quality control [43] and assisting workers in manual assembly tasks [20].…”

Section: ) Process Optimizationmentioning

confidence: 99%

“…Thus, effective methods to enable the automated and early detection of potential defects during production using real-time data are highly desirable to manufacturers. Emerging applications include automated visual inspection using deep learning methods [23], [27], defect prediction to mitigate multistage propagation (aligned with the zero-defect manufacturing paradigm) [40] and online quality prediction [43].…”

Section: ) Quality Controlmentioning

confidence: 99%

Industrial Artificial Intelligence in Industry 4.0 - Systematic Review, Challenges and Outlook

et al. 2020

View full text Add to dashboard Cite

“…Currently, through the use of classical images it is possible to determine erosion and land use chang [ 38 , 39 ]. Additionally, for ANN metrology, its possible to determine imprecise temporal-spatial parameters on images [ 40 , 41 ]. For this reason, one improving spatial solution on adverse resolution conditions are the implementation of Recurrent Neural Networks (RNN), Deep Reinforcement Learning (DRL), and Convolutional Neural Network (CNN) [ 42 , 43 , 44 ].…”

Section: Introductionmentioning

confidence: 99%

Improving Spatial Resolution of Multispectral Rock Outcrop Images Using RGB Data and Artificial Neural Networks

Marques

Souza

Müller

et al. 2020

Sensors

View full text Add to dashboard Cite

Spectral information provided by multispectral and hyperspectral sensors has a great impact on remote sensing studies, easing the identification of carbonate outcrops that contribute to a better understanding of petroleum reservoirs. Sensors aboard satellites like Landsat series, which have data freely available usually lack the spatial resolution that suborbital sensors have. Many techniques have been developed to improve spatial resolution through data fusion. However, most of them have serious limitations regarding application and scale. Recently Super-Resolution (SR) convolution neural networks have been tested with encouraging results. However, they require large datasets, more time and computational power for training. To overcome these limitations, this work aims to increase the spatial resolution of multispectral bands from the Landsat satellite database using a modified artificial neural network that uses pixel kernels of a single spatial high-resolution RGB image from Google Earth as input. The methodology was validated with a common dataset of indoor images as well as a specific area of Landsat 8. Different downsized scale inputs were used for training where the validation used the ground truth of the original size images, obtaining comparable results to the recent works. With the method validated, we generated high spatial resolution spectral bands based on RGB images from Google Earth on a carbonated outcrop area, which were then properly classified according to the soil spectral responses making use of the advantage of a higher spatial resolution dataset.

show abstract

Predictive model-based quality inspection using Machine Learning and Edge Cloud Computing

Cited by 128 publications

References 61 publications

Shape-Based Alignment of the Scanned Objects Concerning Their Asymmetric Aspects

Shape-Based Alignment of the Scanned Objects Concerning Their Asymmetric Aspects

Industrial Artificial Intelligence in Industry 4.0 - Systematic Review, Challenges and Outlook

Improving Spatial Resolution of Multispectral Rock Outcrop Images Using RGB Data and Artificial Neural Networks

Contact Info

Product

Resources

About