Xinhao Wang
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Xinhao Wang has a total of 18 co-authors. Here are the top co-authors by number of shared publications:
- Keelan Evanini
Educational Testing Service Princeton NJ
4 shared publications - Jiaying Liu
Peking University
3 shared publications - Wenjing Wang
Peking University, Beijing, China
3 shared publications - Klaus Zechner
Educational Testing Service
2 shared publications - Shuai Yang
Wangxuan Institute of Computer Technology, Peking University, Beijing, China
2 shared publications - Yao Qian
Microsoft USA
2 shared publications
Latest Publications
- 27 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Magazine
- Vol. 39 (6) , 10-17
Artistic Text Style Transfer: An overview of state-of-the-art methods and datasets [SP Forum]
Word art, which is text rendered with properly designed appealing artistic effects, has been a popular form of art throughout human history. Artistic text effects are of great aesthetic value and symbolic significance. Decorating with appropriate effects not only makes text more attractive but also significantly enhances the atmosphere of a scene. Thus, artistic text effects are widely used in publicity and advertising. Some text effects are simple, such as colors and shadows, while some can be complex, such as the burning flames in Figure 1 and exquisite decorations in Figure 2 . Manually creating vivid text effects requires lots of time and a series of complicated operations: observing the target glyphs, designing appropriate artistic effects, warping the texture to match the character shapes, and so on. It consumes hours of time even for well-trained designers. To produce word art more conveniently and efficiently, artistic text style transfer has been proposed recently to automatically render text with given artistic effects.
- 27 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Signal Processing Magazine
- Vol. 39 (6) , 10-17
Artistic Text Style Transfer: An overview of state-of-the-art methods and datasets [SP Forum]
Word art, which is text rendered with properly designed appealing artistic effects, has been a popular form of art throughout human history. Artistic text effects are of great aesthetic value and symbolic significance. Decorating with appropriate effects not only makes text more attractive but also significantly enhances the atmosphere of a scene. Thus, artistic text effects are widely used in publicity and advertising. Some text effects are simple, such as colors and shadows, while some can be complex, such as the burning flames in Figure 1 and exquisite decorations in Figure 2 . Manually creating vivid text effects requires lots of time and a series of complicated operations: observing the target glyphs, designing appropriate artistic effects, warping the texture to match the character shapes, and so on. It consumes hours of time even for well-trained designers. To produce word art more conveniently and efficiently, artistic text style transfer has been proposed recently to automatically render text with given artistic effects.
- 25 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 31 (10577149) , 6761-6772
CLAST: Contrastive Learning for Arbitrary Style Transfer
Arbitrary style transfer aims at migrating the style of a reference style painting to a target content image. Existing methods find it challenging to achieve good content fidelity and style migration at the same time. Moreover, they all rely on manually defined content and style, which is of limited universality and robustness. In this paper, we propose to introduce contrastive learning into style transfer, instructing the network to automatically learn to model the structural content and artistic style based on natural contrastive relationships in style transfer. Compared with existing methods, our learned modeling of content and style is more robust and universal. In addition, we further propose instance-wise contrastive style losses and a patch-wise contrastive content loss to guide style transfer. Combining the proposed contrastive losses and two self-reconstruction strategies, we develop a new style transfer framework, which is pluggable and can be flexibly applied to various style transfer modules. Experimental results demonstrate that our method has strong flexibility and synthesizes stylized images with higher quality.
- 25 October 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Image Processing
- Vol. 31 (10577149) , 6761-6772
CLAST: Contrastive Learning for Arbitrary Style Transfer
Arbitrary style transfer aims at migrating the style of a reference style painting to a target content image. Existing methods find it challenging to achieve good content fidelity and style migration at the same time. Moreover, they all rely on manually defined content and style, which is of limited universality and robustness. In this paper, we propose to introduce contrastive learning into style transfer, instructing the network to automatically learn to model the structural content and artistic style based on natural contrastive relationships in style transfer. Compared with existing methods, our learned modeling of content and style is more robust and universal. In addition, we further propose instance-wise contrastive style losses and a patch-wise contrastive content loss to guide style transfer. Combining the proposed contrastive losses and two self-reconstruction strategies, we develop a new style transfer framework, which is pluggable and can be flexibly applied to various style transfer modules. Experimental results demonstrate that our method has strong flexibility and synthesizes stylized images with higher quality.
- 18 February 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 45 (1) , 1250-1266
Unsupervised Face Detection in the Dark
Low-light face detection is challenging but critical for real-world applications, such as nighttime autonomous driving and city surveillance. Current face detection models rely on extensive annotations and lack generality and flexibility. In this paper, we explore how to learn face detectors without low-light annotations. Fully exploiting existing normal light data, we propose adapting face detectors from normal light to low light. This task is difficult because the gap between brightness and darkness is too large and complicated at the object level and pixel level. Accordingly, the performance of current low-light enhancement or adaptation methods is unsatisfactory. To solve this problem, we propose a joint High-Low Adaptation (HLA) framework. We design bidirectional low-level adaptation and multitask high-level adaptation. For low-level, we enhance the dark images and degrade the normal-light images, making both domains move toward each other. For high-level, we combine context-based and contrastive learning to comprehensively close the features on different domains. Experiments show that our HLA-Face v2 model obtains superior low-light face detection performance even without the use of low-light annotations. Moreover, our adaptation scheme can be extended to a wide range of applications, such as improving supervised learning and generic object detection. Project publicly available at: https://daooshee.github.io/HLA-Face-v2-Website/ .
- 18 February 2022
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Pattern Analysis and Machine Intelligence
- Vol. 45 (1) , 1250-1266
Unsupervised Face Detection in the Dark
Low-light face detection is challenging but critical for real-world applications, such as nighttime autonomous driving and city surveillance. Current face detection models rely on extensive annotations and lack generality and flexibility. In this paper, we explore how to learn face detectors without low-light annotations. Fully exploiting existing normal light data, we propose adapting face detectors from normal light to low light. This task is difficult because the gap between brightness and darkness is too large and complicated at the object level and pixel level. Accordingly, the performance of current low-light enhancement or adaptation methods is unsatisfactory. To solve this problem, we propose a joint High-Low Adaptation (HLA) framework. We design bidirectional low-level adaptation and multitask high-level adaptation. For low-level, we enhance the dark images and degrade the normal-light images, making both domains move toward each other. For high-level, we combine context-based and contrastive learning to comprehensively close the features on different domains. Experiments show that our HLA-Face v2 model obtains superior low-light face detection performance even without the use of low-light annotations. Moreover, our adaptation scheme can be extended to a wide range of applications, such as improving supervised learning and generic object detection. Project publicly available at: https://daooshee.github.io/HLA-Face-v2-Website/ .
- 1 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 764-771
Using Very Deep Convolutional Neural Networks to Automatically Detect Plagiarized Spoken Responses
This study focuses on the automatic plagiarism detection in the context of high-stakes spoken language proficiency assessment, in which some test takers may attempt to game the test by memorizing prepared source materials before the test and then adapting them on-the-fly during the test to produce their spoken responses. When trying to identify such instances of plagiarism, experienced human raters attempt to find salient matching expressions that appear both in potential source materials and the test responses. This motivates an approach that visualizes a grid of lexical matches between a test response and a source and then applies state-of-the-art image recognition techniques to detect patterns of matching sequences. This study employs Inception networks-very deep convolutional neural networks-to build automatic detection models. The system achieves an F1-score of 79.6% on the class of plagiarized responses outperforming a baseline system based on word sequence matching (F1-score of 74.1%).
- 1 December 2019
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 764-771
Using Very Deep Convolutional Neural Networks to Automatically Detect Plagiarized Spoken Responses
This study focuses on the automatic plagiarism detection in the context of high-stakes spoken language proficiency assessment, in which some test takers may attempt to game the test by memorizing prepared source materials before the test and then adapting them on-the-fly during the test to produce their spoken responses. When trying to identify such instances of plagiarism, experienced human raters attempt to find salient matching expressions that appear both in potential source materials and the test responses. This motivates an approach that visualizes a grid of lexical matches between a test response and a source and then applies state-of-the-art image recognition techniques to detect patterns of matching sequences. This study employs Inception networks-very deep convolutional neural networks-to build automatic detection models. The system achieves an F1-score of 79.6% on the class of plagiarized responses outperforming a baseline system based on word sequence matching (F1-score of 74.1%).
- 20 August 2017
- Published by International Speech Communication Association
- p. 2586-2590
Improving Sub-Phone Modeling for Better Native Language Identification with Non-Native English Speech
- 20 August 2017
- Published by International Speech Communication Association
- p. 2586-2590
Improving Sub-Phone Modeling for Better Native Language Identification with Non-Native English Speech
- 6 September 2015
- Published by International Speech Communication Association
- p. 653-657
Using F0 contours to assess nativeness in a sentence repeat task
- 6 September 2015
- Published by International Speech Communication Association
- p. 653-657
Using F0 contours to assess nativeness in a sentence repeat task
- 1 December 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 294-299
The influence of automatic speech recognition accuracy on the performance of an automated speech assessment system
The effectiveness of automated scoring systems for evaluating spoken language proficiency depends greatly on the quality of the automatic speech recognition (ASR) output that is used to calculate the features for the scoring model. In this paper, we examine the effects of ASR word error rate (WER) on the scores produced by a system for automated scoring of non-native English speaking proficiency, as well as on the scoring model features (especially content features) in order to demonstrate the impact of ASR improvements on the performance of the automated speech assessment system. Five different sets of transcriptions with varying degrees of WER ranging from 0% to 52% (including four sets of ASR hypotheses and manual transcriptions) were obtained for a dataset of spoken responses from a pilot administration of an assessment of non-native English speaking proficiency. The experimental results show that higher performing ASR leads to better performance in the automated assessment system; furthermore, the correlation between human and automated scores drops substantially with an increase in WER from 10.7% to 28.9%, whereas the correlation changes little within the following two ranges of WERs: 0% to 10.7% and 28.9% to 52%. A detailed analysis of the features used in the scoring model shows that the ASR errors have a bigger impact on the content features than the delivery and language use features.
- 1 December 2014
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 294-299
The influence of automatic speech recognition accuracy on the performance of an automated speech assessment system
The effectiveness of automated scoring systems for evaluating spoken language proficiency depends greatly on the quality of the automatic speech recognition (ASR) output that is used to calculate the features for the scoring model. In this paper, we examine the effects of ASR word error rate (WER) on the scores produced by a system for automated scoring of non-native English speaking proficiency, as well as on the scoring model features (especially content features) in order to demonstrate the impact of ASR improvements on the performance of the automated speech assessment system. Five different sets of transcriptions with varying degrees of WER ranging from 0% to 52% (including four sets of ASR hypotheses and manual transcriptions) were obtained for a dataset of spoken responses from a pilot administration of an assessment of non-native English speaking proficiency. The experimental results show that higher performing ASR leads to better performance in the automated assessment system; furthermore, the correlation between human and automated scores drops substantially with an increase in WER from 10.7% to 28.9%, whereas the correlation changes little within the following two ranges of WERs: 0% to 10.7% and 28.9% to 52%. A detailed analysis of the features used in the scoring model shows that the ASR errors have a bigger impact on the content features than the delivery and language use features.
- 21 March 2006
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 1, 602-607
Improving Chinese Text Categorization by Outlier Learning
- 21 March 2006
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- Vol. 1, 602-607
Improving Chinese Text Categorization by Outlier Learning