Xiaolong Wang

UC San Diego Health System

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Trends

Author Order

Document Type

Co-Authors

Yuling He

North China Electric Power University

Sifei Liu

Nvidia (United States)

Abhinav Gupta

Carnegie Mellon University

Jan Kautz

Nvidia (United States)

Chandra Kambhamettu

University of Delaware

Guiji Tang

North China Electric Power University

Trevor Darrell

University of California, Berkeley

Guiji Tang

North China Electric Power University

Shalini De Mello

Nvidia (United States)

David Gerada

University of Nottingham

Cooperative Institutions

Chinese Academy of Sciences

253

University of Chinese Academy of Sciences

142

Institute of High Energy Physics

140

Tsinghua University

North China Electric Power University

Shanghai Jiao Tong University

Peking University

University of Science and Technology of China

Shandong University

Harbin Institute of Technology

Author Statistics

Papers

Citation

H-Index

i-10 index

Research Field

ActorsNeRF: Animatable Few-shot Human Rendering with Generalizable NeRFs

2021 IEEE/CVF International Conference on Computer Vision (ICCV) (2023)

Jiteng Mu Shen Sang Nuno Vasconcelos Xiaolong Wang

While NeRF-based human representations have shown impressive novel view synthesis results, most methods still rely on a large number of images / views for training. In this work, we propose a novel animatable NeRF called ActorsNeRF. It is first pre-trained on diverse human subjects, and then adapted with few-shot monocular video frames for a new actor with unseen poses. Building on previous generalizable NeRFs with parameter sharing using a ConvNet encoder, ActorsNeRF further adopts two human priors to capture the large human appearance, shape, and pose variations. Specifically, in the encoded feature space, we will first align different human subjects in a category-level canonical space, and then align the same human from different frames in an instance-level canonical space for rendering. We quantitatively and qualitatively demonstrate that ActorsNeRF significantly outperforms the existing state-of-the-art on few-shot generalization to new people and poses on multiple datasets. Project page: https://jitengmu.github.io/ActorsNeRF/.

Monocular

10.1109/iccv51070.2023.01686

Cite

Citations (11)

Dual-layer control strategy based on economic characterization of lifetime state and frequency regulation limit partition of hybrid energy storage

Journal of Energy Storage (2024)

Cuiping Li Xiaolong Wang Junhui Li Xingxu Zhu Jia Chen

Characterization

10.1016/j.est.2024.113326

Cite

Citations (1)

Recognition Algorithm for Digital Image Based on Template Matching

Ordnance Industry Automation (2005)

Xiaolong Wang

Aiming at the shortage of the traditional template matching method such as only identifying simple numbers and characters, several kinds of new algorithm, namely, match method of vector weighting, extraction method of vector feature, invariable center moment was presented to improve the template matching method. There is a great of advantages in digital image recognition for improved recognition algorithm of digital image.

Template matching

Feature (linguistics)

Economic shortage

Source

Cite

Citations (0)

AnyTeleop: A General Vision-Based Dexterous Robot Arm-Hand Teleoperation System

Yuzhe Qin Wei Yang Binghao Huang Karl Van Wyk Hao Su

Yuzhe Qin was an intern at NVIDIA during the project.experiments, AnyTeleop can outperform a previous system that was designed for a specific robot hardware with a higher success rate, using the same robot.For teleoperation in simulation, AnyTeleop leads to better imitation learning performance, compared with a previous system that is particularly designed for that simulator.Robot 1 Controlled by Operator #1 On Computer #1 Robot 2 Controlled by Operator #2 On Computer #2

Teleoperation

Robotic hand

Telerobotics

10.15607/rss.2023.xix.015

Cite

Citations (11)

Automatic Diagnosis of Rolling Element Bearing Under Different Conditions Based on RVMD and Envelope Order Capture

IEEE Access (2019)

Yingjie Wu Zhongzhong Hu Weilun An Xiaoming Li Xiaolong Wang

Fault characteristic frequency is the main basis for rolling element bearing diagnostics but finding a suitable frequency band for demodulation and searching for the fault characteristic frequencies consume a lot of time and manpower in practice. A data-driven method based on recursive variational mode decomposition (RVMD), and an envelope order capture is proposed to realize the automatic fault diagnosis of bearing under different operating conditions. The process starts with a new proposed RVMD of the vibration signal, where the mode with maximum kurtosis of the unbiased autocorrelation of the envelope is selected to get envelope order spectrum. Thereafter, an order capture algorithm is designed to automatically search for the fault characteristic orders in theory, which are used for constructing feature vectors for diagnosis. The proposed method is tested on two test-beds which both contain the same type of bearing (SKF6205) but operate in different conditions, and gets good performance in bearing diagnosis. In addition, the fault diagnosis of test-bed two using training samples that are from test-bed one is investigated. This method reveals well generalization capability in the fault diagnosis of the same type of rolling element bearing under different operating conditions.

Rolling-element bearing

Envelope (radar)

SIGNAL (programming language)

Kurtosis

10.1109/access.2019.2927039

Cite

Citations (1)

Vat Photopolymerization 3D Printing Hydrogels and Bionic Adhesive Devices: A Minireview

Advanced Materials Technologies (2024)

Lei Shi Yixian Wang Xin Xu Desheng Liu Zhongying Ji

Abstract Vivid biological structures with reversible adhesive properties are widely found in nature. Although current biomimicking adhesion systems have attracted extensive attention, unmet challenges remain in terms of excellent adhesion preservation, precise structural fabrication, and good environmental compatibility. By combining advanced vat photopolymerization 3D printing technology with hydrogel materials with excellent environmental adaptability, bioinspired adhesion devices present satisfactory application prospects. Herein, a systematic overview of bioinspired hydrogel adhesion devices from the perspectives of material design, structural design, and fabrication is presented. First, the main types and application prospects of vat photopolymerization 3D printed hydrogels are introduced. Subsequently, the research progress on hydrogel‐based vat photopolymerized 3D printed bionic adhesion devices is detailed. Future developments, such as balancing unmet challenges and expected opportunities, in 3D printable multiscale bionic adhesive devices are presented.

3d printed

Biofabrication

10.1002/admt.202301853

Cite

Citations (3)

Hollywood in Homes: Crowdsourcing Data Collection for Activity Understanding

Lecture notes in computer science (2016)

Gunnar A. Sigurdsson Gül Varol Xiaolong Wang Ali Farhadi Ivan Laptev

Casual

Crowdsourcing

10.1007/978-3-319-46448-0_31

Cite

Citations (961)

2M-BAND INTERLEAVED DFT MODULATED FILTER BANKS WITH PERFECT RECONSTRUCTION

International Journal of Wavelets Multiresolution and Information Processing (2008)

Peng‐Lang Shui Xiaolong Wang

In this paper, we propose a new family of perfect reconstruction (PR) complex filter banks, named interleaved discrete Fourier transform modulated filter banks (Interleaved DFT-FBs). In the filter banks, the analysis filters are generated by interlaced exponential modulating two different analysis prototype filters, and the synthesis filters are generated by two different synthesis prototype filters via the same manner. The filter banks have a simple polyphase structure similar to DFT modulated filter banks (DFT-FBs). More importantly, the proposed Interleaved DFT-FBs can achieve critically sampled PR complex filter bank with FIR analysis and synthesis filters, which is impossible for DFT-FBs. We give and prove the PR condition for 2M-band Interleaved DFT-FBs. Utilizing the result, the design procedure of the prototype filters is presented. In addition, by the theoretic analysis and numerical examples, it is shown that the analysis and synthesis filters cannot simultaneously provide good stopband attenuation for the critically sampled PR Interleaved DFT-FBs. Although the limitation always exits, the filter banks can find applications in some subband coding systems of high bit rate.

Filter bank

Stopband

Network synthesis filters

10.1142/s021969130800246x

Cite

Citations (2)

VideoINR: Learning Video Implicit Neural Representation for Continuous Space-Time Super-Resolution

arXiv (Cornell University) (2022)

Zeyuan Chen Yinbo Chen Jingwen Liu Xingqian Xu Vidit Goel

Videos typically record the streaming and continuous visual data as discrete consecutive frames. Since the storage cost is expensive for videos of high fidelity, most of them are stored in a relatively low resolution and frame rate. Recent works of Space-Time Video Super-Resolution (STVSR) are developed to incorporate temporal interpolation and spatial super-resolution in a unified framework. However, most of them only support a fixed up-sampling scale, which limits their flexibility and applications. In this work, instead of following the discrete representations, we propose Video Implicit Neural Representation (VideoINR), and we show its applications for STVSR. The learned implicit neural representation can be decoded to videos of arbitrary spatial resolution and frame rate. We show that VideoINR achieves competitive performances with state-of-the-art STVSR methods on common up-sampling scales and significantly outperforms prior works on continuous and out-of-training-distribution scales. Our project page is at http://zeyuan-chen.com/VideoINR/ .

Interpolation

Representation

Frame rate

10.48550/arxiv.2206.04647

Cite

Citations (0)

Accelerating Proposal Generation Network for \\Fast Face Detection on Mobile Devices

arXiv (Cornell University) (2019)

Heming Zhang Xiaolong Wang Jingwen Zhu C.‐C. Jay Kuo

Face detection is a widely studied problem over the past few decades. Recently, significant improvements have been achieved via the deep neural network, however, it is still challenging to directly apply these techniques to mobile devices for its limited computational power and memory. In this work, we present a proposal generation acceleration framework for real-time face detection. More specifically, we adopt a popular cascaded convolutional neural network (CNN) as the basis, then apply our acceleration approach on the basic framework to speed up the model inference time. We are motivated by the observation that the computation bottleneck of this framework arises from the proposal generation stage, where each level of the dense image pyramid has to go through the network. In this work, we reduce the number of image pyramid levels by utilizing both global and local facial characteristics (i.e., global face and facial parts). Experimental results on public benchmarks WIDER-face and FDDB demonstrate the satisfactory performance and faster speed compared to the state-of-the-arts. %the comparable accuracy to state-of-the-arts with faster speed.

Pyramid (geometry)

Speedup

10.48550/arxiv.1904.12094

Cite

Citations (0)