Search

Your search keyword '"Chen, Xiaotian"' showing total 714 results
Start Over Author "Chen, Xiaotian"
714 results on '"Chen, Xiaotian"'

9. S2R-DepthNet: Learning a Generalizable Depth-specific Structural Representation

Author

Chen, Xiaotian, Wang, Yuwang, Chen, Xuejin, and Zeng, Wenjun

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Human can infer the 3D geometry of a scene from a sketch instead of a realistic image, which indicates that the spatial structure plays a fundamental role in understanding the depth of scenes. We are the first to explore the learning of a depth-specific structural representation, which captures the essential feature for depth estimation and ignores irrelevant style information. Our S2R-DepthNet (Synthetic to Real DepthNet) can be well generalized to unseen real-world data directly even though it is only trained on synthetic data. S2R-DepthNet consists of: a) a Structure Extraction (STE) module which extracts a domaininvariant structural representation from an image by disentangling the image into domain-invariant structure and domain-specific style components, b) a Depth-specific Attention (DSA) module, which learns task-specific knowledge to suppress depth-irrelevant structures for better depth estimation and generalization, and c) a depth prediction module (DP) to predict depth from the depth-specific representation. Without access of any real-world images, our method even outperforms the state-of-the-art unsupervised domain adaptation methods which use real-world images of the target domain for training. In addition, when using a small amount of labeled real-world data, we achieve the state-ofthe-art performance under the semi-supervised setting. The code and trained models are available at https://github.com/microsoft/S2R-DepthNet., Comment: Accepted by CVPR2021(oral)

Published
2021

12. Three-dimensional eddy current pulsed thermography and its applications

Author

Chen, Xiaotian

Abstract

The measurement and quantification of defects is a challenge for Non-DestructiveTesting and Evaluation (NDT&E). Such challenges include the precise localisation and detection of surface and sub-surface defects, as well as the quantification of such defects. This work first reports a three-dimensional (3D) Eddy Current Pulsed Thermography (ECPT) system via integration with an RGB-D camera. Then, various quantitative measurements and analyses of defects are carried out based on the 3D ECPT system. The ECPT system at Newcastle University has been prooven to be an effective nondestructive testing (NDT) method in surface and sub-surface detection over the past few years. Based on the different numerical or analytical models, it has achieved precise defect detection on the rail tracks, wind turbines, carbon fibre reinforced plastic (CFRP) and so on. The ECPT system has the advantage of fast inspection and a large lift-off range. However, it involves a trade-off between detectable defect size and inspection area compared with other NDT methods. In addition, there are challenges of defect detection in a complex structure. Thus, the quantification of defects gives a higher requirement of the measurement the object geometry information. Furthermore, the analysis of thermal diffusion requires a precise 3D model. For this reason, a 3D ECPT system is proposed that adds each heat pixel with an exact X-Y-Z coordinate. In this work, first, the 3D ECPT system is built. A feature-based automatic calibration of the infrared camera and the RGB-D camera is proposed. Second, the software platform is built. A fast 3D visualization is completed with multi-threading technology and the Point Cloud Library. Lastly, various studies of defect localization, quantification and thermal tomography reconstruction are carried out.

Published
2021

17. Structure-Aware Residual Pyramid Network for Monocular Depth Estimation

Author

Chen, Xiaotian, Chen, Xuejin, and Zha, Zheng-Jun

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Monocular depth estimation is an essential task for scene understanding. The underlying structure of objects and stuff in a complex scene is critical to recovering accurate and visually-pleasing depth maps. Global structure conveys scene layouts, while local structure reflects shape details. Recently developed approaches based on convolutional neural networks (CNNs) significantly improve the performance of depth estimation. However, few of them take into account multi-scale structures in complex scenes. In this paper, we propose a Structure-Aware Residual Pyramid Network (SARPN) to exploit multi-scale structures for accurate depth prediction. We propose a Residual Pyramid Decoder (RPD) which expresses global scene structure in upper levels to represent layouts, and local structure in lower levels to present shape details. At each level, we propose Residual Refinement Modules (RRM) that predict residual maps to progressively add finer structures on the coarser structure predicted at the upper level. In order to fully exploit multi-scale image features, an Adaptive Dense Feature Fusion (ADFF) module, which adaptively fuses effective features from all scales for inferring structures of each scale, is introduced. Experiment results on the challenging NYU-Depth v2 dataset demonstrate that our proposed approach achieves state-of-the-art performance in both qualitative and quantitative evaluation. The code is available at https://github.com/Xt-Chen/SARPN., Comment: 7pages, 7figures, Accepted by the 28th International Joint Conference on Artificial Intelligence (IJCAI-2019)

Published
2019

19. Pilot Fatigue Evaluation Based on Eye-Movement Index and Performance Index

Author

Xue, Hongjun, Chen, Xiaotian, Zhang, Xiaoyan, Ma, Ziqi, Qiu, Kai, Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Harris, Don, editor, and Li, Wen-Chin, editor

Published
2022
Full Text
View/download PDF

20. Study on the Mechanism of Pilot Fatigue on Performance

Author

Xue, Hongjun, Ma, Ziqi, Zhang, Xiaoyan, Chen, Xiaotian, Qiu, Kai, Li, Jie, Goos, Gerhard, Founding Editor, Hartmanis, Juris, Founding Editor, Bertino, Elisa, Editorial Board Member, Gao, Wen, Editorial Board Member, Steffen, Bernhard, Editorial Board Member, Yung, Moti, Editorial Board Member, Harris, Don, editor, and Li, Wen-Chin, editor

Published
2022
Full Text
View/download PDF

32. Automated Analysis of Doppler Echocardiographic Videos as a Screening Tool for Valvular Heart Diseases

Author

Yang, Feifei, Chen, Xiaotian, Lin, Xixiang, Chen, Xu, Wang, Wenjun, Liu, Bohan, Li, Yao, Pu, Haitao, Zhang, Liwei, Huang, Dangsheng, Zhang, Meiqing, Li, Xin, Wang, Hui, Wang, Yueheng, Guo, Huayuan, Deng, Yujiao, Zhang, Lu, Zhong, Qin, Li, Zongren, Yu, Liheng, Duan, Yongjie, Zhang, Peifang, Wu, Zhenzhou, Burkhoff, Daniel, Wang, Qiushuang, and He, Kunlun

Published
2022
Full Text
View/download PDF

34. The role of sarcopenic obesity for the prediction of prognosis of patients with gastrointestinal cancer.

Author

Chen, Wenqing, Yuan, Qinggang, Li, Xiangrui, Yao, Jiashu, Yuan, Lihua, Chen, Xiaotian, and Gao, Bo

Subjects
CANCER prognosis, RECEIVER operating characteristic curves, GASTROINTESTINAL cancer, PROPORTIONAL hazards models, BODY composition
Abstract

Background: Sarcopenic obesity (SO) in patients with gastrointestinal cancer is associated with a poor prognosis. We aimed to investigate the prognostic impact of SO in patients with gastrointestinal cancer, as well as the diagnostic cut‐off value of SO in patients with gastrointestinal cancer among Chinese population. Methods: We conducted a consecutive cohort study. Between January 2017 and January 2019, 289 patients diagnosed with gastrointestinal cancer were included in our study. Skeletal muscle area, total fat area, and subcutaneous fat area were measured by CT scan. All patients were followed up for 5 years. Receiver operating characteristic curves (ROC) were adopted to determine the cut‐off values of visceral fat obesity for the prediction of sarcopenia. Based on the cut‐off values, patients with sarcopenia combined with visceral fat obesity were divided into the SO group, and the others were divided into the non‐sarcopenic obesity (NSO) group. Kaplan–Meier curves and univariate and multivariate Cox proportional hazard models were employed to explore the associations of body composition profiles with 5‐year overall survival and disease‐specific survival. Results: Obtained from Youden's Index for ROC for the prediction of 5‐year survival, skeletal muscle mass index (SMI) ≤40.02 cm2/m2 with VFA ≥ 126.30 cm2 in men and SMI ≤32.05 cm2/m2 with VFA ≥72.42 cm2 in women indicate a risk of poor prognosis in patients diagnosed with gastrointestinal cancer. Patients with SO had poorer 5‐year overall survival (OS) than patients with NSO (6.74% vs. 82.84%, p < 0.001), and poorer 5‐year DFS (6.74% vs. 81.82%, p < 0.001). In multivariate analysis, we found that the long‐term mortality risk was approximately 13‐fold higher among patients in the SO group compared to those with no conditions. Conclusions: Preoperative assessment of SO is useful not only for monitoring nutritional status but also for predicting 5‐year OS in gastrointestinal cancer patients. [ABSTRACT FROM AUTHOR]

Published
2024
Full Text
View/download PDF

47. SFX Miscellaneous Free Ejournals Target: Usage Survey among the SFX Community

Author

Renaville, François, Branse, Yosef, Chen, Xiaotian, and Needleman, Mark

Subjects
Computer Science - Digital Libraries
Abstract

The number of free or open access articles is increasing rapidly, and their retrieval with library indexes and OpenURL link resolvers has been a challenge. In June 2014, the SFX MISCELLANEOUS FREE EJOURNALS target contained more than 24,000 portfolios of all kinds. The SFX Knowledge Base Advisory Board (KBAB) carried out an international survey to get an overview of the usage of this target by the SFX community and to precisely identify what could be done to improve it. The target is widely used among the community. However, many respondents complained about three major problems: (a) incorrect links, (b) full texts actually not free, and (c) incorrect or missing thresholds (years and volumes information).

Published
2015
Full Text
View/download PDF

Catalog

Books, media, physical & digital resources
See catalog results