- 标题
- 摘要
- 关键词
- 实验方案
- 产品
过滤筛选
- 2019
- 2018
- pattern recognition
- image
- partial discharge
- convolutional neural network(CNN)
- Conditional Random Fields (CRF)
- Convolutional Neural Network (CNN)
- Fine Classification
- Airborne hyperspectral
- green tide
- Elegant End-to-End Fully Convolutional Network (E3FCN)
- Optoelectronic Information Science and Engineering
- Shanghai Jiao Tong University
- Ocean University of China
- University of Oulu
- Wuhan University
- Central South University
- Hubei University
-
Industrial Optical Character Recognition System in Printing Quality Control of Hot-Rolled Coils Identification
摘要： This work presents a system designed to detect printing errors and misidentifications on steel coils that could lead to tracking problems and even guide to the delivery of the wrong product to the final client. An optical character recognition system is proposed to extract the printed identification of steel coils from images captured by a fixed camera in an industrial environment. The method considers different digital image processing techniques to deal with the significant lighting and printing variation observed, followed by a segmentation process that extracts and aligns the characters originally printed in an arch form, ending with a classification routine based on a convolutional neural network. The proposed system presents an approach to treat lighting variations in images, covering low contrast, darker and brighter images. Experiment carried out on a data set with approximately 20,000 images achieved an accuracy higher than 98%, supporting the validity of the proposed method.
关键词： Digital image processing,Convolutional neural networks,Optical character recognition,Intelligent manufacturing
更新于2025-09-12 10:27:22
-
Degradation Mechanism Detection in Photovoltaic Backsheets by Fully Convolutional Neural Network
摘要： Materials and devices age with time. Material aging and degradation has important implications for lifetime performance of materials and systems. While consensus exists that materials should be studied and designed for degradation, materials inspection during operation is typically performed manually by technicians. the manual inspection makes studies prone to errors and uncertainties due to human subjectivity. in this work, we focus on automating the process of degradation mechanism detection through the use of a fully convolutional deep neural network architecture (f-cnn). We demonstrate that f-cnn architecture allows for automated inspection of cracks in polymer backsheets from photovoltaic (pV) modules. the developed f-cnn architecture enabled an end-to-end semantic inspection of the pV module backsheets by applying a contracting path of convolutional blocks (encoders) followed by an expansive path of decoding blocks (decoders). first, the hierarchy of contextual features is learned from the input images by encoders. next, these features are reconstructed to the pixel-level prediction of the input by decoders. the structure of the encoder and the decoder networks are thoroughly investigated for the multi-class pixel-level degradation type prediction for pV module backsheets. the developed f-cnn framework is validated by reporting degradation type prediction accuracy for the pixel level prediction at the level of 92.8%.
关键词： photovoltaic backsheets,automated inspection,degradation mechanism,fully convolutional neural network,semantic segmentation
更新于2025-09-12 10:27:22
-
2D Image-To-3D Model: Knowledge-Based 3D Building Reconstruction (3DBR) Using Single Aerial Images and Convolutional Neural Networks (CNNs)
摘要： In this study, a deep learning (DL)-based approach is proposed for the detection and reconstruction of buildings from a single aerial image. The pre-required knowledge to reconstruct the 3D shapes of buildings, including the height data as well as the linear elements of individual roofs, is derived from the RGB image using an optimized multi-scale convolutional–deconvolutional network (MSCDN). The proposed network is composed of two feature extraction levels to ?rst predict the coarse features, and then automatically re?ne them. The predicted features include the normalized digital surface models (nDSMs) and linear elements of roofs in three classes of eave, ridge, and hip lines. Then, the prismatic models of buildings are generated by analyzing the eave lines. The parametric models of individual roofs are also reconstructed using the predicted ridge and hip lines. The experiments show that, even in the presence of noises in height values, the proposed method performs well on 3D reconstruction of buildings with di?erent shapes and complexities. The average root mean square error (RMSE) and normalized median absolute deviation (NMAD) metrics are about 3.43 m and 1.13 m, respectively for the predicted nDSM. Moreover, the quality of the extracted linear elements is about 91.31% and 83.69% for the Potsdam and Zeebrugge test data, respectively. Unlike the state-of-the-art methods, the proposed approach does not need any additional or auxiliary data and employs a single image to reconstruct the 3D models of buildings with the competitive precision of about 1.2 m and 0.8 m for the horizontal and vertical RMSEs over the Potsdam data and about 3.9 m and 2.4 m over the Zeebrugge test data.
关键词： convolutional neural networks,deep learning,building reconstruction,building detection,depth prediction
更新于2025-09-12 10:27:22
-
A convolutional neural network approach on bead geometry estimation for a laser cladding system
摘要： Laser cladding is a complex manufacturing process. As the laser beam melts the feedstock powder, small changes in laser power or traverse speed reflect on deviations of the deposition’s geometry. Thus, fine-tuning these process parameters is crucial to achieve desirable results. In order to monitor and further understand the laser cladding process, an automated method for clad bead final geometry estimation is proposed. To do so, six different convolutional neural network architectures were developed to analyze the process’ molten pool image acquired by a 50-fps coaxial camera. Those networks receive both the camera image and the process parameters as inputs, yielding width and height of the clad beads as outputs. The results of the network’s performances show testing error mean values as little as 8 μm for clad beads around a millimeter in height. For the width dimension, in 95% of the cases, the error remained under 15% of the bead’s width. Plots of the target versus the estimated values show coefficients of determination over 0.95 on the testing set. The architectures are then compared, and their performances are discussed. Deeper convolutional layers far exceeded the performance of shallower ones; nonetheless, deeper densely connected layers decreased the performances of the networks when compared with shallower ones. Those results represent yet another alternative on intelligent process monitoring with potential for real-time usage, taking the researches one step further into developing a closed-loop control for this process.
关键词： Optical monitoring,Geometry estimation,Laser cladding,Convolutional neural network,Bead geometry
更新于2025-09-12 10:27:22
-
[IEEE 2019 10th Workshop on Hyperspectral Imaging and Signal Processing: Evolution in Remote Sensing (WHISPERS) - Amsterdam, Netherlands (2019.9.24-2019.9.26)] 2019 10th Workshop on Hyperspectral Imaging and Signal Processing: Evolution in Remote Sensing (WHISPERS) - Knowledge Transfer via Convolution Neural Networks for Multi-Resolution Lawn Weed Classification
摘要： Weed identi?cation and classi?cation are essential and challenging tasks for site-speci?c weed control. Object-based image analysis making use of spatial information is adopted in this study for the weed classi?cation because the spectral similarity between the weeds and crop is high. With the availability of a wide range of sensors, it is likely to capture weed imagery at various altitudes and with different speci?cations of the sensor. In this paper, we propose a novel method using transfer learning to deal with multi-resolution images from various sensors via Convolutional Neural Networks (CNN). CNN trained for a typical image data set and the trained weights are transferred to other data sets of different resolutions. In this way, the new data sets can be classi?ed by ?ne-tuning the network using a small number of training samples, which reduces the need of big data to train the model. To avoid over-?tting during the ?ne-tuning, small deep learning architecture is proposed and investigated using the parameters of the initial layers of pre-trained model. The sizes of training samples are investigated for their impact on the performance of ?ne-tuning. Experiments were conducted with ?eld data, which show that the proposed method outperforms the direct training method in terms of recognition accuracy and computation cost.
关键词： Hyperspectral images,Resolution,Convolutional Neural Network (CNN),Weed Mapping,Transfer Learning
更新于2025-09-12 10:27:22
-
[IEEE 2019 IEEE 21st International Workshop on Multimedia Signal Processing (MMSP) - Kuala Lumpur, Malaysia (2019.9.27-2019.9.29)] 2019 IEEE 21st International Workshop on Multimedia Signal Processing (MMSP) - 3D Facial Expression Recognition Based on Multi-View and Prior Knowledge Fusion
摘要： This paper presents a novel multi-view convolutional neural network (CNN) model for 3D facial expression recognition (FER). In contrast to existing deep learning-based 3D FER approaches that mainly learn the expressions from frontal facial attribute images, the proposed model incorporates multi-view and facial prior information of the observed 3D face into the learning process. This information is jointly trained in an end-to-end manner to predict the emotion of the input 3D face model. The experiments on public 3D facial expression datasets show that training the CNN with additional information from different views and facial prior knowledge would result in learning more discriminative features as against from a single view. Our model outperforms the state-of-the-art 3D FER methods in term of recognition accuracy indicating its effectiveness. Moreover, the improvement of the proposed model is displayed more clearly in the discrimination of low- intensity facial expressions.
关键词： facial expression recognition (FER),convolutional neural network,3D face scan
更新于2025-09-12 10:27:22
-
[IEEE 2019 IEEE International Conference on Image Processing (ICIP) - Taipei, Taiwan (2019.9.22-2019.9.25)] 2019 IEEE International Conference on Image Processing (ICIP) - Prior Knowledge Guided Small Object Detection on High-Resolution Images
摘要： When applying common object detection algorithms to detect small objects on high-resolution images, the down-sampling operation of the input images is inevitable due to the limitation of GPU memory. Accordingly, the details for characterizing small objects are lost. To resolve this contradiction, a small object detection method in a coarse-to-fine manner is presented. Specifically, some rough regions of interest (ROI) are firstly computed from low-resolution images. The prior knowledge of the positions of objects is used to guide the generation of ROIs. Then the features of small ROIs are recomputed from high-resolution images, and the features of large ROIs are obtained from the feature maps used to generate ROIs. The proposed method is validated on two datasets. One is a plant phenotyping dataset and the other is a public traffic sign dataset. Experimental results convincingly show the effectiveness of the proposed method.
关键词： prior knowledge,convolutional neural network,high-resolution image,small object detection
更新于2025-09-11 14:15:04
-
[IEEE 2019 2nd International Conference on Artificial Intelligence and Big Data (ICAIBD) - Chengdu, China (2019.5.25-2019.5.28)] 2019 2nd International Conference on Artificial Intelligence and Big Data (ICAIBD) - Short-term Photovoltaic Power Forecasting Using Deep Convolutional Networks
摘要： Uncertain factors such as weather conditions make the photovoltaic power forecasting challenging. Therefore, some advanced deep learning (DL) techniques have been introduced into the field. However, the exploitation of historical data by existing DL-based models is usually limited to one-dimensional power arrays, and the information describing power varies over time is also ignored. Differently, this work proposes a model based on a deep convolutional network (DCN), which attempts to improve forecast performance by combining feature learning and multi-dimensional arrays. To match the DCN paradigm along with digging the power information indepth, we use the historical daily data to build a power tensor. We train and test the proposed model using data sets from the real world, Self-evaluation and comparison with several state-of-the-art models demonstrate the superiority of our proposal.
关键词： convolutional neural networks,power tensor,PV power forecasting,Deep learning
更新于2025-09-11 14:15:04
-
Quantitative diagnosis method of beam defects based on laser Doppler non-contact random vibration measurement
摘要： The beam structure is prone to defect damage during its use, and the rapid quantitative diagnosis of the beam structure can detect the defects of the beam in real time and quantitatively. In this article, the method of obtaining the vibration time-domain signal under random excitation of beam structure is proposed by using random vibration excitation and Laser Doppler principle. Based on this, the defect quantitative identification algorithm of beam structure is proposed based on fast Fourier, continuous wavelet transform and convolutional neural network. The random vibration of different parts of steel beams with artificial defects is measured by Laser Doppler method. The experimental results show that the defect size of the beam structure can be effectively identified only by the random vibration signal of the finite point. The method is expected to help to develop an online real-time assessment instrument for beam structure defects in service state.
关键词： Continuous wavelet transform,Beam defects,Laser Doppler,Non-contact random excitation,Quantitative diagnosis,Convolutional neural network
更新于2025-09-11 14:15:04
-
[IEEE 2018 IEEE Advanced Accelerator Concepts Workshop (AAC) - Breckenridge, CO, USA (2018.8.12-2018.8.17)] 2018 IEEE Advanced Accelerator Concepts Workshop (AAC) - Compression of Terawatt Long-Wavelength Laser Pulses Through Backward Raman Amplification
摘要： JPEG is one of the widely used lossy compression methods. JPEG-compressed images usually suffer from compression artifacts including blocking and blurring, especially at low bit-rates. Soft decoding is an effective solution to improve the quality of compressed images without changing codec or introducing extra coding bits. Inspired by the excellent performance of the deep convolutional neural networks (CNNs) on both low-level and high-level computer vision problems, we develop a dual pixel-wavelet domain deep CNNs-based soft decoding network for JPEG-compressed images, namely DPW-SDNet. The pixel domain deep network takes the four downsampled versions of the compressed image to form a 4-channel input and outputs a pixel domain prediction, while the wavelet domain deep network uses the 1-level discrete wavelet transformation (DWT) coefficients to form a 4-channel input to produce a DWT domain prediction. The pixel domain and wavelet domain estimates are combined to generate the final soft decoded result. Experimental results demonstrate the superiority of the proposed DPW-SDNet over several state-of-the-art compression artifacts reduction algorithms.
关键词： JPEG,soft decoding,deep convolutional neural networks,compression artifacts,DPW-SDNet
更新于2025-09-11 14:15:04