A Benchmark Dataset and Deep Learning-Based Image Reconstruction for Electrical Capacitance Tomography

Zheng, Jiangbin; Li, Jinku; Li, Yang; Peng, Lihui

doi:10.3390/s18113701

Cited by 34 publications

(24 citation statements)

References 45 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…Example of classifier model which is built on limited training images was reported by [26]. Nevertheless, this dataset will be a valuable framework in developing tomographic sensing interpretation using ML [27] [28].…”

Section: Machine Learning Methods Using Mutual Induction Datamentioning

confidence: 99%

Interior Void Classification in Liquid Metal using Multi-Frequency Magnetic Induction Tomography with a Machine Learning Approach

Muttakin¹,

Soleimani²

2021

Preprint

View full text Add to dashboard Cite

Identification of gas bubble, void detection and porosity estimation are important factors in many liquid metal processes. In steel casting, the importance of flow condition and phase distribution in crucial parts, such as submerged entry nozzle (SEN) and mould raises the needs to observe the phenomena. Cross-section of flow shapes can be visualised using the magnetic induction tomography (MIT) technique. However, the inversion procedure in the image reconstruction has either limited resolution or complex computation degrading its real-time capability. Additionally, in some cases, the actual image may not be essential whereas the void fraction or porosity needs to be estimated. This work proposes an interior void classifier based on multi-frequency mutual induction measurements with eutectic alloy GaInSn as a cold liquid metal model contained in a 3D printed plastic miniature of an SEN. The sensors consist of eight coils arranged in a circle encapsulating the column, providing combinatorial detection on conductive surface and depth. The datasets are induced voltage collections of several non-metallic inclusions (NMI) patterns in liquid metal static test and used to train a machine learning model. The model architectures are a fully connected neural network (FCNN) for 1D; and a convolutional neural network (CNN) for 2D data. The classifier using 1D data has been trained to approximately 86% accuracy on this dataset. CNN classification using multi-dimensional data with more classes produces 96% of test accuracy. Refined with representative flow scenarios, the trained model could be deployed for an intelligent online control system of the liquid metal process.

show abstract

Section: Machine Learning Methods Using Mutual Induction Datamentioning

confidence: 99%

Interior Void Classification in Liquid Metal using Multi-Frequency Magnetic Induction Tomography with a Machine Learning Approach

Muttakin¹,

Soleimani²

2021

Preprint

View full text Add to dashboard Cite

show abstract

“…With the development of data-driven methods designed for robotics applications [33], the importance of synthetic data has been highlighted. Recent works [8,9,[34][35][36] combine real and synthetic data to generate 3D object datasets, which render 3D object models on real backgrounds in order to produce synthesized images. YCB-Video [9] dataset is the mostly used 3D object datasets for 6D object pose estimation.…”

Section: Related Workmentioning

confidence: 99%

RobotP: A Benchmark Dataset for 6D Object Pose Estimation

Yuan

Hoogenkamp

Veltkamp

2021

Sensors

View full text Add to dashboard Cite

Deep learning has achieved great success on robotic vision tasks. However, when compared with other vision-based tasks, it is difficult to collect a representative and sufficiently large training set for six-dimensional (6D) object pose estimation, due to the inherent difficulty of data collection. In this paper, we propose the RobotP dataset consisting of commonly used objects for benchmarking in 6D object pose estimation. To create the dataset, we apply a 3D reconstruction pipeline to produce high-quality depth images, ground truth poses, and 3D models for well-selected objects. Subsequently, based on the generated data, we produce object segmentation masks and two-dimensional (2D) bounding boxes automatically. To further enrich the data, we synthesize a large number of photo-realistic color-and-depth image pairs with ground truth 6D poses. Our dataset is freely distributed to research groups by the Shape Retrieval Challenge benchmark on 6D pose estimation. Based on our benchmark, different learning-based approaches are trained and tested by the unified dataset. The evaluation results indicate that there is considerable room for improvement in 6D object pose estimation, particularly for objects with dark colors, and photo-realistic images are helpful in increasing the performance of pose estimation algorithms.

show abstract

“…In 2017, Klosowski et al [ 25 ] proposed two reconstruction methods based on fully-connected neural network (FCNN) and CNN, respectively, which proved that CNN can be directly used between the input and the output. In 2018, Zheng et al [ 26 , 27 ] developed an auto-encoder method to achieve complicated reconstruction in electrical capacitance tomography. Feng et al [ 28 ] investigated the feasibility of a back-propagation neural network (BPNN) to reestablish the distribution of optical properties in a diffuse optical tomography (DOT) problem.…”

Section: Introductionmentioning

confidence: 99%

Application of a Generative Adversarial Network in Image Reconstruction of Magnetic Induction Tomography

Yang

Liu

Wang

et al. 2021

Sensors

View full text Add to dashboard Cite

Image reconstruction of Magnetic induction tomography (MIT) is an ill-posed problem. The non-linear characteristics lead many difficulties to its solution. In this paper, a method based on a Generative Adversarial Network (GAN) is presented to tackle these barriers. Firstly, the principle of MIT is analyzed. Then the process for finding the global optimum of conductivity distribution is described as a training process, and the GAN model is proposed. Finally, the image was reconstructed by a part of the model (the generator). All datasets are obtained from an eight-channel MIT model by COMSOL Multiphysics software. The voltage measurement samples are used as input to the trained network, and its output is an estimate for image reconstruction of the internal conductivity distribution. The results based on the proposed model and the traditional algorithms were compared, which have shown that average root mean squared error of reconstruction results obtained by the proposed method is 0.090, and the average correlation coefficient with original images is 0.940, better than corresponding indicators of BPNN and Tikhonov regularization algorithms. Accordingly, the GAN algorithm was able to fit the non-linear relationship between input and output, and visual images also show that it solved the usual problems of artifact in traditional algorithm and hot pixels in L2 regularization, which is of great significance for other ill-posed or non-linear problems.

show abstract

A Benchmark Dataset and Deep Learning-Based Image Reconstruction for Electrical Capacitance Tomography

Cited by 34 publications

References 45 publications

Interior Void Classification in Liquid Metal using Multi-Frequency Magnetic Induction Tomography with a Machine Learning Approach

Interior Void Classification in Liquid Metal using Multi-Frequency Magnetic Induction Tomography with a Machine Learning Approach

RobotP: A Benchmark Dataset for 6D Object Pose Estimation

Application of a Generative Adversarial Network in Image Reconstruction of Magnetic Induction Tomography

Contact Info

Product

Resources

About