arXiv subjects

Xiangyu Xu

Publications and source records attributed to Xiangyu Xu.

64 records · Page 4Linked to original sources

arXiv2020

GLEAN: Generative Latent Bank for Large-Factor Image Super-Resolution

We show that pre-trained Generative Adversarial Networks (GANs), e.g., StyleGAN, can be used as a latent bank to improve the restoration quality of large-factor image super-resolution (SR). While most existing SR approaches attempt to generate realistic textures through learning with adversarial loss, our method, Generative LatEnt bANk (GLEAN), goes beyond existing practices by directly leveraging rich and diverse priors encapsulated in a pre-trained GAN. But unlike prevalent GAN inversion methods that require expensive image-specific optimization at runtime, our approach only needs a single forward pass to generate the upscaled image. GLEAN can be easily incorporated in a simple encoder-bank-decoder architecture with multi-resolution skip connections. Switching the bank allows the method to deal with images from diverse categories, e.g., cat, building, human face, and car. Images upscaled by GLEAN show clear improvements in terms of fidelity and texture faithfulness in comparison to existing methods.

cs.CV↗

arXiv2020

3D Human Shape and Pose from a Single Low-Resolution Image with Self-Supervised Learning

3D human shape and pose estimation from monocular images has been an active area of research in computer vision, having a substantial impact on the development of new applications, from activity recognition to creating virtual avatars. Existing deep learning methods for 3D human shape and pose estimation rely on relatively high-resolution input images; however, high-resolution visual content is not always available in several practical scenarios such as video surveillance and sports broadcasting. Low-resolution images in real scenarios can vary in a wide range of sizes, and a model trained in one resolution does not typically degrade gracefully across resolutions. Two common approaches to solve the problem of low-resolution input are applying super-resolution techniques to the input images which may result in visual artifacts, or simply training one model for each resolution, which is impractical in many realistic applications. To address the above issues, this paper proposes a novel algorithm called RSC-Net, which consists of a Resolution-aware network, a Self-supervision loss, and a Contrastive learning scheme. The proposed network is able to learn the 3D body shape and pose across different resolutions with a single model. The self-supervision loss encourages scale-consistency of the output, and the contrastive learning scheme enforces scale-consistency of the deep features. We show that both these new training losses provide robustness when learning 3D shape and pose in a weakly-supervised manner. Extensive experiments demonstrate that the RSC-Net can achieve consistently better results than the state-of-the-art methods for challenging low-resolution images.

cs.CV↗

arXiv2020

A Lightweight and Accurate Localization Algorithm Using Multiple Inertial Measurement Units

This paper proposes a novel inertial-aided localization approach by fusing information from multiple inertial measurement units (IMUs) and exteroceptive sensors. IMU is a low-cost motion sensor which provides measurements on angular velocity and gravity compensated linear acceleration of a moving platform, and widely used in modern localization systems. To date, most existing inertial-aided localization methods exploit only one single IMU. While the single-IMU localization yields acceptable accuracy and robustness for different use cases, the overall performance can be further improved by using multiple IMUs. To this end, we propose a lightweight and accurate algorithm for fusing measurements from multiple IMUs and exteroceptive sensors, which is able to obtain noticeable performance gain without incurring additional computational cost. To achieve this, we first probabilistically map measurements from all IMUs onto a virtual IMU. This step is performed by stochastic estimation with least-square estimators and probabilistic marginalization of inter-IMU rotational accelerations. Subsequently, the propagation model for both state and error state of the virtual IMU is also derived, which enables the use of the classical filter-based or optimization-based sensor fusion algorithms for localization. Finally, results from both simulation and real-world tests are provided, which demonstrate that the proposed algorithm outperforms competing algorithms by noticeable margins.

cs.RO↗

arXiv2019

Quadratic video interpolation

Video interpolation is an important problem in computer vision, which helps overcome the temporal limitation of camera sensors. Existing video interpolation methods usually assume uniform motion between consecutive frames and use linear models for interpolation, which cannot well approximate the complex motion in the real world. To address these issues, we propose a quadratic video interpolation method which exploits the acceleration information in videos. This method allows prediction with curvilinear trajectory and variable velocity, and generates more accurate interpolation results. For high-quality frame synthesis, we develop a flow reversal layer to estimate flow fields starting from the unknown target frame to the source frame. In addition, we present techniques for flow refinement. Extensive experiments demonstrate that our approach performs favorably against the existing linear models on a wide variety of video datasets.

cs.CV↗

arXiv2019

Discrete Laplace Operator Estimation for Dynamic 3D Reconstruction

We present a general paradigm for dynamic 3D reconstruction from multiple independent and uncontrolled image sources having arbitrary temporal sampling density and distribution. Our graph-theoretic formulation models the Spatio-temporal relationships among our observations in terms of the joint estimation of their 3D geometry and its discrete Laplace operator. Towards this end, we define a tri-convex optimization framework that leverages the geometric properties and dependencies found among a Euclideanshape-space and the discrete Laplace operator describing its local and global topology. We present a reconstructability analysis, experiments on motion capture data and multi-view image datasets, as well as explore applications to geometry-based event segmentation and data association.

cs.CV↗

arXiv2019

Towards Real Scene Super-Resolution with Raw Images

Most existing super-resolution methods do not perform well in real scenarios due to lack of realistic training data and information loss of the model input. To solve the first problem, we propose a new pipeline to generate realistic training data by simulating the imaging process of digital cameras. And to remedy the information loss of the input, we develop a dual convolutional neural network to exploit the originally captured radiance information in raw images. In addition, we propose to learn a spatially-variant color transformation which helps more effective color corrections. Extensive experiments demonstrate that super-resolution with raw data helps recover fine details and clear structures, and more importantly, the proposed network and data generation pipeline achieve superior results for single image super-resolution in real scenarios.

eess.IV↗

arXiv2019

Learning Deformable Kernels for Image and Video Denoising

Most of the classical denoising methods restore clear results by selecting and averaging pixels in the noisy input. Instead of relying on hand-crafted selecting and averaging strategies, we propose to explicitly learn this process with deep neural networks. Specifically, we propose deformable 2D kernels for image denoising where the sampling locations and kernel weights are both learned. The proposed kernel naturally adapts to image structures and could effectively reduce the oversmoothing artifacts. Furthermore, we develop 3D deformable kernels for video denoising to more efficiently sample pixels across the spatial-temporal space. Our method is able to solve the misalignment issues of large motion from dynamic scenes. For better training our video denoising model, we introduce the trilinear sampler and a new regularization term. We demonstrate that the proposed method performs favorably against the state-of-the-art image and video denoising approaches on both synthetic and real-world data.

cs.CV↗

arXiv2015

The Scale-free Network of Passwords : Visualization and Estimation of Empirical Passwords

In this paper, we present a novel vision of large scale of empirical password sets available and improve the understanding of passwords by revealing their interconnections and considering the security on a level of the whole password set instead of one single password level. Through the visualization of Yahoo, Phpbb, 12306, etc. we, for the first time, show what the spatial structure of empirical password sets are like and take the community and clustering patterns of the passwords into account to shed lights on the definition of popularity of a password based on their frequency and degree separately. Furthermore, we propose a model of statistical guessing attack from the perspective of the data's topological space, which provide an explanation of the "cracking curve". We also give a lower bound of the minimum size of the dictionary needed to compromise arbitrary ratio of any given password set by proving that it is equivalent to the minimum dominating set problem, which is a NP-complete problem. Hence the minimal dictionary problem is also NP-complete.

cs.CR↗

arXiv2004

Effect of strain on the magnetic and transport properties of the composite manganites, La$_{2/3}$Ca$_{1/3}$MnO$_3$/yttria-stabilized zirconia

The magnetic and transport properties have been investigated for the composite polycrystalline manganites, (1-x)$La_{2/3}Ca_{1/3}MnO_3$/(x)yttria-stabilized zirconia ((1-x)LCMO/(x)YSZ), at various YSZ fractions, x, ranging from 0 % to 15 %. The ac magnetic susceptibility, $χ(T)$, DC magnetization, $M(T)$, temperature dependent resistivity, $ρ(T)$ and thermoelectric power (TEP), $S(T)$, have been measured. It was found, surprisingly, that a TEP peak showed up in the magnetic transition region for the sample with the x even as little as 0.75 %. The magnetic transition temperature reaches the minimum value as x increases from 0 % to 4.5 % and goes up as x increases further. Several possible factors such as the effect of strain, the finite size effect, and the effect of magnetic tunnelling coupling, {\it etc.}, in affecting the above physical properties of the composite manganites have been studied carefully. The strain induced by the YSZ/LCMO boundary layer (BL) was identified as the leading factor to account for the x dependence of these properties. It demonstrates that the effect of strain could be important in the bulk manganites as in the film sample.

cond-mat.other↗

arXiv2004

Thick MgB2 film with (101) oriented micro-crystals

Very thick, ~ 40 $μ$m, clean, and highly textured MgB2 film was effectively grown on an Al2O3 substrate. The fabrication technique is by the hybrid physical-chemical vapor deposition (HPCVD) using B2H6 gas and Mg ingot as the sources. The X-ray diffraction (XRD) analysis shows a highly (101)-oriented MgB2 crystal structure without any impurity detected. There is no signal from the substrate in the XRD spectrum, indicating that the film thickness exceeds the X-ray penetration length. Scanning electron microscopy (SEM) reveals that the film is composed of highly-packed MgB2 micro-crystals with a uniform size distribution of about 2 $μ$m in diameter and 0.2 $μ$m in thickness. According to the compositional analysis of energy-dispersive X-ray spectroscopy (EDX), no oxygen, hence no MgO, exists in the textured film, consistent with the XRD result. Also, the transport properties are similar to those of a single crystal, indicating a clean film of good crystallite. The zero field transition temperatures are determined as TC(onset) = 39.2 K and TC(zero) = 38.4 K, giving a sharp transition typical of a clean sample. The residual resistivity ratio (RRR) is determined as 6.4 and the magnetoreisitance (MR) is about 28 % at 40 K under the applied field of 9 T, which are similar to those of a single crystal. The zero temperature upper critical field, HC2(0), is extrapolated as 19 T from the TC(onset) at applied field up to 9 T.

cond-mat.supr-con↗