1,720,969 research outputs found
Weakly supervised power line detection algorithm using a recursive noisy label update with refined broken line segments
No full textDetection of power lines in aerial images is an important problem to prevent accidents of unmanned aerial vehicles operating at low altitudes in the electrical industry. Recently, pixel-level power line detection using deep learning has been studied but production of the pixel-level annotations for massive dataset is difficult. In this study, we propose a power line detection algorithm using weakly supervised learning method to reduce the labeling cost for dataset generation. The algorithm is divided into two stages. First, an approximately localized mask was generated based on a convolutional neural network which was trained with only patch-level labels. Second, recursive training of segmentation network with refined broken line segments was executed. A refinement algorithm, line segment connecting (LSC) is a power-line-specialized refinement module that connects broken lines by approximating the segments as partially straight. In proposed algorithm, predicted image at each recursive step was updated as a label of the next training and the label was developed by itself with LSC. The comprehensive experimental results of our algorithm showed state-of-art F1-score of 94.3% in weakly supervised learning approaches on public dataset. This result suggests that the proposed algorithm is useful for low labeling cost with high performance in line detection application. © 2020 Elsevier Ltd1
End-to-end recognition of slab identification numbers using a deep convolutional neural network
No full textThis paper proposes a novel algorithm for the end-to-end recognition of slab identification numbers (SINS). In the steel industry, automatic recognition of an individual product information is important for production management. The recognition of SINs in actual factory scenes is a challenging problem due to complicated background and low-quality of characters. Conventional rule-based algorithms were developed to extract information of SINs, but these methods require engineering knowledge and tedious work for parameter tuning. The proposed algorithm employs a data-driven method to overcome these limitations and to handle the challenges for the recognition of SINs. This paper proposes accumulated response map and model-based score function to effectively use the outputs of a deep convolutional neural network. Experiments were thoroughly conducted for industrial data collected from an actual steelworks to verify the effectiveness of the proposed algorithm. Experiment results demonstrate that simultaneous recognition of entire characters in a SIN by optimizing the model-based score function is more effective for the robust performance compared to separated recognition of individual characters. (C) 2017 Elsevier B.V. All rights reserved.112sciescopu
Automated Brittle Fracture Rate Estimator for Steel Property Evaluation Using Deep Learning After Drop-Weight Tear Test
Full text linkThis study proposes an automated brittle fracture rate (BFR) estimator using deep learning. As the demand for line-pipes increases in various industries, the need for BFR estimation through dropweight tear test (DWTT) increases to evaluate steel's property. Conventional BFR or ductile fracture rate (DFR) estimation methods require an expensive 3D scanner. Alternatively, a rule-based approach is used with a single charge-coupled device (CCD) camera. However, it is sensitive to the hyper-parameter. To solve these problems, we propose an approach based on deep learning that has recently been successful in the fields of computer vision and image processing. The method proposed in this study is the first to use deep learning approach for BFR estimation. The proposed method consists of a VGG-based U-Net (VU-Net) which is inspired by U-Net and fully convolutional network (FCN). VU-Net includes a deep encoder and a decoder. The encoder is adopted from VGG19 and transferred with a pre-trained model with ImageNet. In addition, the structure of the decoder is the same as that of the encoder, and the decoder uses the feature maps of the encoder through concatenation operation to compensate for the reduced spatial information. To analyze the proposed VU-Net, we experimented with different depths of networks and various transfer learning approaches. In terms of accuracy used in real industrial application, we compared the proposed VU-Net with U-Net and FCN to evaluate the performance. The experiments showed that VU-Net was the accuracy of approximately 94.9 %, and was better than the other two, which had the accuracies of about 91.8 % and 93.7 %, respectively.11Ysciescopu
Wearable Device-Based System to Monitor a Driver's Stress, Fatigue, and Drowsiness
No full textThis paper proposes a wearable device-based system to monitor the abnormal conditions of a driver, including stress, fatigue, and drowsiness. The system measures the motional and physiological information of the driver using the developed wearable device on the wrist. Preprocessing is used to distinguish the valid signal parts of the measured signals, because various noises can occur in wearable sensors. Features are extracted from the signal parts, and an optimal feature set is determined by an analysis of variance and a sequential floating forward selection algorithm. To classify the driver's state, a support vector machine-based classification method is used to obtain high generalization performance considering interdriver variance. Experiments were conducted on an indoor driving simulator, with 28 subjects, to gather data for each state. The classification accuracy was 98.43% for fivefold cross validation on the data. In a subject-independent test, the accuracy was 68.31% for the four states and 84.46% for the three states consisting of normal, stressed, and fatigued or drowsy states. Using the proposed system, the abnormal conditions of the driver can be detected and distinguished. This advantage contributes to safer and more comfortable driving. Furthermore, the utilization of the wearable device makes the system easy to use.11sciescopu
Automated defect inspection system for metal surfaces based on deep learning and data augmentation
No full textRecent efforts to create a smart factory have inspired research that analyzes process data collected from Internet of Things (IOT) sensors, to predict product quality in real time. This requires an automatic defect inspection system that quantifies product quality data by detecting and classifying defects in real time. In this study, we propose a vision-based defect inspection system to inspect metal surface defects. In recent years, deep convolutional neural networks (DCNNs) have been used in many manufacturing industries and have demonstrated the excellent performance as a defect classification method. A sufficient amount of training data must be acquired, to ensure high performance using a DCNN. However, owing to the nature of the metal manufacturing industry, it is difficult to obtain enough data because some defects occur rarely. Owing to this imbalanced data problem, the generalization performance of the DCNN-based classification algorithm is lowered. In this study, we propose a new convolutional variational autoencoder (CVAE) and deep CNN-based defect classification algorithm to solve this problem. The CVAE-based data generation technology generates sufficient defect data to train the classification model. A conditional CVAE (CCVAE) is proposed to generate images for each defect type in a single CVAE model. We also propose a classifier based on a DCNN with high generalization performance using data generated from the CCVAE. In order to verify the performance of the proposed method, we performed experiments using defect images obtained from an actual metal production line. The results showed that the proposed method exhibited an excellent performance. © 2020 The Society of Manufacturing Engineers1
Transfer learning of a deep convolutional neural network for localizing handwritten slab identification numbers
No full text1
Detection of Internal Short Circuit in Lithium Ion Battery Using Model-Based Switching Model Method
Full text linkEarly detection of an internal short circuit (ISCr) in a Li-ion battery can prevent it from undergoing thermal runaway, and thereby ensure battery safety. In this paper, a model-based switching model method (SMM) is proposed to detect the ISCr in the Li-ion battery. The SMM updates the model of the Li-ion battery with ISCr to improve the accuracy of ISCr resistance R-ISCf estimates. The open circuit voltage (OCV) and the state of charge (SOC) are estimated by applying the equivalent circuit model, and by using the recursive least squares algorithm and the relation between OCV and SOC. As a fault index, the R-ISCf is estimated from the estimated OCVs and SOCs to detect the ISCr, and used to update the model; this process yields accurate estimates of OCV and R-ISCf. Then the next R-ISCf is estimated and used to update the model iteratively. Simulation data from a MATLAB/Simulink model and experimental data verify that this algorithm shows high accuracy of R-ISCf estimates to detect the ISCr, thereby helping the battery management system to fulfill early detection of the ISCr.119sciescopu
Unified deep neural networks for end-to-end recognition of multi-oriented billet identification number
No full textIn this study, a novel framework for the recognition of a billet identification number (BIN) using deep learning is proposed. Because a billet, which is a semi-finished product, could be rolled, the BIN may be rotated at various angles. Most product numbers, including BIN, are a combination of individual characters. Such product numbers are determined based on the class of each character and its order (or the positioning). In addition, the two pieces of information are constant even if the product number is rotated. Inspired by this concept, the proposed framework of deep neural networks has two outputs. One is for the class of an individual character, and the other is the order of an individual character within BIN. Compared with a previous study, the proposed network requires an additional annotation but does not require additional labor for labeling. The multi-task learning for two annotations has a positive role in the representation learning of a network, which is shown in the experiment results. Furthermore, to achieve a good performance of the BIN identification, we analyzed various networks using the proposed framework. The proposed algorithm was then compared with a conventional algorithm to evaluate the performance of the BIN identification.11Nsciescopu
- …
