Search

Your search keyword '"Wang, Xionghui"' showing total 19 results
Start Over Author "Wang, Xionghui"
19 results on '"Wang, Xionghui"'

1. ByteEdit: Boost, Comply and Accelerate Generative Image Editing

Author

Ren, Yuxi, Wu, Jie, Lu, Yanzuo, Kuang, Huafeng, Xia, Xin, Wang, Xionghui, Wang, Qianqian, Zhu, Yixing, Xie, Pan, Wang, Shiyin, Xiao, Xuefeng, Wang, Yitong, Zheng, Min, and Fu, Lean

Subjects
Computer Science - Computer Vision and Pattern Recognition
Abstract

Recent advancements in diffusion-based generative image editing have sparked a profound revolution, reshaping the landscape of image outpainting and inpainting tasks. Despite these strides, the field grapples with inherent challenges, including: i) inferior quality; ii) poor consistency; iii) insufficient instrcution adherence; iv) suboptimal generation efficiency. To address these obstacles, we present ByteEdit, an innovative feedback learning framework meticulously designed to Boost, Comply, and Accelerate Generative Image Editing tasks. ByteEdit seamlessly integrates image reward models dedicated to enhancing aesthetics and image-text alignment, while also introducing a dense, pixel-level reward model tailored to foster coherence in the output. Furthermore, we propose a pioneering adversarial and progressive feedback learning strategy to expedite the model's inference speed. Through extensive large-scale user evaluations, we demonstrate that ByteEdit surpasses leading generative image editing products, including Adobe, Canva, and MeiTu, in both generation quality and consistency. ByteEdit-Outpainting exhibits a remarkable enhancement of 388% and 135% in quality and consistency, respectively, when compared to the baseline model. Experiments also verfied that our acceleration models maintains excellent performance results in terms of quality and consistency.

Published
2024

2. ByteEdit: Boost, Comply and Accelerate Generative Image Editing

Author

Ren, Yuxi, Wu, Jie, Lu, Yanzuo, Kuang, Huafeng, Xia, Xin, Wang, Xionghui, Wang, Qianqian, Zhu, Yixing, Xie, Pan, Wang, Shiyin, Xiao, Xuefeng, Wang, Yitong, Zheng, Min, Fu, Lean, Goos, Gerhard, Series 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, Leonardis, Aleš, editor, Ricci, Elisa, editor, Roth, Stefan, editor, Russakovsky, Olga, editor, Sattler, Torsten, editor, and Varol, Gül, editor

Published
2025
Full Text
View/download PDF

3. AlignDet: Aligning Pre-training and Fine-tuning in Object Detection

Author

Li, Ming, Wu, Jie, Wang, Xionghui, Chen, Chen, Qin, Jie, Xiao, Xuefeng, Wang, Rui, Zheng, Min, and Pan, Xin

Subjects
Computer Science - Computer Vision and Pattern Recognition, Computer Science - Artificial Intelligence, Computer Science - Machine Learning
Abstract

The paradigm of large-scale pre-training followed by downstream fine-tuning has been widely employed in various object detection algorithms. In this paper, we reveal discrepancies in data, model, and task between the pre-training and fine-tuning procedure in existing practices, which implicitly limit the detector's performance, generalization ability, and convergence speed. To this end, we propose AlignDet, a unified pre-training framework that can be adapted to various existing detectors to alleviate the discrepancies. AlignDet decouples the pre-training process into two stages, i.e., image-domain and box-domain pre-training. The image-domain pre-training optimizes the detection backbone to capture holistic visual abstraction, and box-domain pre-training learns instance-level semantics and task-aware concepts to initialize the parts out of the backbone. By incorporating the self-supervised pre-trained backbones, we can pre-train all modules for various detectors in an unsupervised paradigm. As depicted in Figure 1, extensive experiments demonstrate that AlignDet can achieve significant improvements across diverse protocols, such as detection algorithm, model backbone, data setting, and training schedule. For example, AlignDet improves FCOS by 5.3 mAP, RetinaNet by 2.1 mAP, Faster R-CNN by 3.3 mAP, and DETR by 2.3 mAP under fewer epochs., Comment: Camera Ready Version on ICCV 2023. Code and Models are publicly available. Project Page: https://liming-ai.github.io/AlignDet

Published
2023

12. Contributions of Emphysema and Functional Small Airway Disease on Intrapulmonary Vascular Volume in COPD

Author

Huang,Xiaoqi, Yin,Weiling, Shen,Min, Wang,Xionghui, Ren,Tao, Wang,Lei, Liu,Min, Guo,Youmin, Huang,Xiaoqi, Yin,Weiling, Shen,Min, Wang,Xionghui, Ren,Tao, Wang,Lei, Liu,Min, and Guo,Youmin

Abstract

Xiaoqi Huang,1,* Weiling Yin,1,* Min Shen,1 Xionghui Wang,1 Tao Ren,1 Lei Wang,1 Min Liu,2 Youmin Guo1 1Department of Radiology, Yan’an University Affiliated Hospital, Yan’an, People’s Republic of China; 2Department of Radiology, China-Japan Friendship Hospital, Beijing, People’s Republic of China*These authors contributed equally to this workCorrespondence: Youmin Guo, Department of Radiology, Yan’an University Affiliated Hospital, Yan’an, People’s Republic of China, Tel +86-13909187113, Email cjr.guoyoumin@vip.163.comBackground: Previous studies have demonstrated that there is a certain correlation between emphysema and changes in pulmonary small blood vessels in patients with chronic obstructive pulmonary disease (COPD), but most of them were limited to the investigation of the inspiratory phase. The emphysema indicators need to be further optimized. Based on the parametric response mapping (PRM) method, this study aimed to investigate the effect of emphysema and functional small airway disease on intrapulmonary vascular volume (IPVV).Methods: This retrospective study enrolled 63 healthy subjects and 47 COPD patients, who underwent both inspiratory and expiratory CT scans of the chest and pulmonary function tests (PFTs). Inspiratory and expiratory IPVV were measured by using an automatic pulmonary vessels integration segmentation approach, the ratio of emphysema volume (Emph%), functional small airway disease volume (fsAD%), and normal areas volume (Normal%) were quantified by the PRM method for biphasic CT scans. The participants were grouped according to PFTs. Analysis of variance (ANOVA) and Kruskal–Wallis H-test were used to analyze the differences in indicators between different groups. Then, Spearman’s rank correlation coefficients were used to analyze the correlation between Emph%, fsAD%, Normal%, PFTs, and IPVV. Finally, multiple linear regression was applied to analyze the effects of Emph% and fsAD% on IPVV.Resul

Published
2022

13. An edge thinning algorithm based on newly defined single‐pixel edge patterns.

Author

Ren, Lijuan, Wang, Xionghui, Wang, Nina, Zhang, Guangpeng, Li, Yongchang, and Yang, Zhijian

Subjects
*PIXELS, *CONTINUITY, *UNIFORMITY, *NOISE
Abstract

To improve the uniformity of one‐pixel width and continuity of the thinned edges, this paper proposes an edge thinning algorithm acting on grey‐scale edge images based on 24 self‐defined single‐pixel connection patterns. First, for binary or blurred grey‐scale gradient edge images, a distance–greyscale coupling algorithm is proposed to achieve gradient enhancement in the edge width direction. Then the elimination rules of noise points and gradient calculation method are given. Secondly, the marking rules of the first three pixels of each edge are given. The next pixel to be marked must meet that the new last three pixels belong to the 24 connection modes. Whether the qualified pixels are retained depends on its grey value and the local edge gradient. The algorithm is tested on four types of images. The results show that the proposed method can guarantee uniform, smooth, and connected one‐pixel‐wide lines that lie at the centre of the initial edges. The algorithm and the existing algorithms are performed on portrait image and four scenarios of the indoor datasets. Five evaluation indicators are statistically analyzed to prove the feasibility and effectiveness of the proposed algorithm. [ABSTRACT FROM AUTHOR]

Published
2023
Full Text
View/download PDF

19. Experimental Investigation of Self-Excited Instabilities in Liquid-Fueled Swirl Combustion

Author

Wang, Xionghui

Subjects
Aerospace Materials, Combustion Instability, Swirling Flow, Heat release rate oscillation, Liquid-fueled swirling combustion
Abstract

A self-excited instability of liquid-fueled swirling combustion in a laboratory scale combustor is investigated in this study.Liquid fuel is injected through a simplex nozzle and mixed with a swirling air flow generated by an in-house designed, radial-radial, counter-rotating swirler. Air is preheated to 400F (477K). Air flow rate is maintained constant during tests while fuel flow rate is varied to achieve different global equivalence ratios. Pressure perturbations were monitored and recorded by two dynamic pressure transducers (Model PCB 116B) located at 4 and 9.5 inches downstream of a dome plate. Global chemiluminescence filtered by a CH* filter is recorded by a photomultiplier tube (PMT). The data of perturbed pressure and instant chemiluminescence emission are saved simultaneously through a high-response-data-acquisition board. Instant flame image indicating spatial chemiluminescence fluctuations is recorded by a high-speed camera. Acoustic boundaries of the combustion chamber’s inlet and outlet were first characterized. For the no bias flow case, the swirler impedance exhibits classical compact-element characteristics. When bias flow is presented, the swirler impedance greatly depends on flow Mach number. The temperature dependence of the impedance is captured in the wavenumber. The chamber exit impedances were measured and curve-fitted into correlations. Then the flame response to air modulation is studied at four equivalence ratios. A loudspeaker is used to perturb plenum air in a 22.5 inches (0.5715m) long combustion chamber. The flame dynamics are expressed as flame transfer functions which are defined as the ratio of relative global heat release rate oscillations to the relative velocity fluctuations at the base of the flame, and calculated within the range of tested equivalence ratio. The behavior of the flame dynamics indicates that the combustion system is similar to a damped high-order system. At last, the instability map of the studied swirler is established. Pressure and flame chemiluminescence emission oscillations were recorded. Velocity oscillations are computed from pressure perturbations using a two-microphone method. The measurements recovered Rayleigh’s criterion which states that when pressure perturbation and unsteady heat input are in phase, combustion instability tends to occur. During tests, a sudden increase of static pressure was observed while air mass flow rates remained the same when combustion instabilities occurred, indicating the air flow temperature at the exit of the swirler increased. Meanwhile, the intensity of chemiluminescence emissions inside the swirler venturi observed from the top view of the combustion chamber became stronger at the occurrence of combustion instability. A possible theory of burning inside the venturi or burning on the fuel nozzle tip is proposed. The possible relation of burning inside the venturi and the occurrence of combustion instability is investigated. The variation of air mass distribution between the primary swirler and secondary swirler due to burning inside the venturi is checked using a numerical simulation. The increased flow Mach number at the chamber inlet altered the inlet acoustic impedance. The outcomes of burning inside the venturi are in favor of combustion instabilities. The universality of the relationship between burning inside the venturi and combustion instability requires further investigations.

Published
2017

Catalog

Books, media, physical & digital resources
See catalog results