Search

Your search keyword '"Shen, Fei"' showing total 197 results
Start Over Author "Shen, Fei" Publisher ieee
197 results on '"Shen, Fei"'

29. Exploring Spatial Significance via Hybrid Pyramidal Graph Network for Vehicle Re-Identification.

Author

Shen, Fei, Zhu, Jianqing, Zhu, Xiaobin, Xie, Yi, and Huang, Jingchang

Abstract

Existing vehicle re-identification methods commonly use spatial pooling operations to aggregate feature maps extracted via off-the-shelf backbone networks, such as visual geometry group network (VGGNet), Google network (GoogLeNet) and residual network (ResNet). They ignore exploring the spatial significance of feature maps, eventually degrading the vehicle re-identification performance. In this paper, firstly, an innovative spatial graph network (SGN) is proposed to elaborately explore the spatial significance of feature maps. The SGN stacks multiple spatial graphs (SGs). Each SG assigns feature map’s elements as nodes and utilizes spatial neighborhood relationships to determine edges among nodes. During the SGN’s propagation, each node and its spatial neighbors on an SG are aggregated to the next SG. On the next SG, each aggregated node is re-weighted with a learnable parameter to find the significance at the corresponding location. Secondly, a novel pyramidal graph network (PGN) is designed to comprehensively explore the spatial significance of feature maps at multiple scales. The PGN organizes multiple SGNs in a pyramidal manner and makes each SGN handles feature maps of a specific scale. Finally, a hybrid pyramidal graph network (HPGN) is developed by embedding the PGN behind a ResNet-50 based backbone network. Extensive experiments on three large scale vehicle databases (i.e., VeRi776, VehicleID, and VeRi-Wild) demonstrate that the proposed HPGN is superior to state-of-the-art vehicle re-identification approaches in terms of accuracy, parameter cost, and computation cost. In addition, experiments show that the proposed PGN is universal to various backbone networks. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

31. A New Intermediate-Domain SVM-Based Transfer Model for Rolling Bearing RUL Prediction.

Author

Shen, Fei and Yan, Ruqiang

Abstract

Various working conditions and bearing structures make remaining useful life (RUL) prediction more challenging. This article presents a new intermediate-domain support vector machine (SVM) based transfer model for rolling bearing RUL prediction. This transfer model aims to solve the problem when the source degradation indices are too poor to be applied by introducing the new intermediate domain. At first, the high-quality feature degradation indices are selected using the joint evaluation index and the principal component analysis algorithm. Then, a maximum correlated kurtosis deconvolution algorithm is carried out to obtain the demarcation point between the healthy and degradation stages. After selecting the high-quality domain, the objective function of intermediate-domain SVM is designed based on the classical-domain-independent SVM to optimize “source-to-intermediate” and “intermediate-to-target” transfer processes simultaneously. Finally, experiments using both ball and conical bearing datasets indicate that the proposed method has higher RUL prediction performance than the existing models, which proves the advantage of multioptimization transfer learning. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

36. Inverse Problem in Resonant Ultrasound Spectroscopy With Sampling and Optimization: A Comparative Study on Human Cortical Bone.

Author

Shen, Fei, Fan, Fan, Wang, Rui, Wang, Yue, Yang, Songyan, Wu, Qiong, Laugier, Pascal, Cai, Xiran, and Niu, Haijun

Subjects
*BONE mechanics, *RESONANT ultrasound spectroscopy, *INVERSE problems, *COMPACT bone, *MARKOV chain Monte Carlo, *PARTICLE swarm optimization, *STIFFNESS (Mechanics)
Abstract

The Bayesian inference with prior knowledge has been proposed recently to solve the inverse problem in resonant ultrasound spectroscopy. It allows inferring the elastic properties of high damping materials, such as cortical bone with less dependence on the initial guessed values. In this method, the estimation of the stiffness coefficients is expressed as a probabilistic solution to the inverse problem, which can be achieved by sampling or optimization methods. However, the detailed performance comparison of these two strategies applied to high damping materials has not been fully studied. In this work, the full stiffness tensor of 52 transversely isotropic cortical bone specimens was obtained using Markov chain Monte Carlo (MCMC) sampling and particle swarm optimization (PSO), respectively. Results showed that the local probability distributions of stiffness coefficients estimated by the two methods are consistent. Compared with MCMC, the average calculation speed of PSO is ten times faster [614 s ± 59 s (MCMC) versus 53 s ± 22 s (PSO)]. The mean standard error between theoretical and experimental resonant frequencies was slightly smaller for PSO compared with MCMC. In conclusion, PSO, a global optimization strategy, is suitable to solve the inverse problem for high damping materials. [ABSTRACT FROM AUTHOR]

Published
2022
Full Text
View/download PDF

49. A Novel Pixel-Wise Defect Inspection Method Based on Stable Background Reconstruction.

Author

Lv, Chengkan, Shen, Fei, Zhang, Zhengtao, Xu, De, and He, Yonghao

Subjects
*GENERATIVE adversarial networks, *PIXELS, *SURFACE texture, *INDUSTRIAL goods, *INSPECTION & review, *GALLIUM nitride
Abstract

In this article, an anomaly detection method based on background reconstruction is proposed to perform defect inspection on the texture surface of the industrial products. This method consists of two modules: 1) an autoencoder integrated with a generative adversarial network is utilized to reconstruct the textured background of the original image as a defect-free reference. Specifically, extra anomalous images are introduced and a mapping method of anomaly is given to improve the stability of reconstruction. 2) A U-net based inspection network is trained to perform pixel-wise analysis of the differences between the original and the reconstructed defect-free image. During these processes, only artificial synthesized defective images are utilized to train the model without any real defective samples. A series of experiments are conducted on several texture image data sets and the industrial production line. The experimental results reveal the effectiveness and versatility of the proposed method. [ABSTRACT FROM AUTHOR]

Published
2021
Full Text
View/download PDF

50. Towards Autonomous Driving Technology: A Method to Enhance Visibility in Fog Based on Low-Position Road Lighting

Author

Wei-Sheng Li, Song-Bai Yang, Guanhua Lai, Bolin Xu, Hai-Tian Zhao, Bing Zhang, and Shen-Fei Chen

Subjects
010309 optics, Visual distance, Position (vector), 021105 building & construction, 0103 physical sciences, 0211 other engineering and technologies, Environmental science, 02 engineering and technology, Visibility, 01 natural sciences, Remote sensing
Abstract

To see through the thick fog is an important challenge in the field of autonomous driving technology. This experiment developed a new lighting method to improve visibility under thick fog based on the principle of visual acuity in biology of vision. The experiment was conducted in a fog equipment that is 1.5 meters * 1.2 meters * 4.0 meters in volume, with an accurate ratio of 1:15 compared with realistic road. The experiment increased the visual distance under thick fog from 1.5 meters to 3.0 meters, increased by one time. The research results are of great significance to the development of autonomous driving technology.

Published
2018

Catalog

Books, media, physical & digital resources
See catalog results