arXiv subjects

Peng Wang

Publications and source records attributed to Peng Wang.

At least 991 records · Page 55Linked to original sources

arXiv2016

DOC: Deep OCclusion Estimation From a Single Image

Recovering the occlusion relationships between objects is a fundamental human visual ability which yields important information about the 3D world. In this paper we propose a deep network architecture, called DOC, which acts on a single image, detects object boundaries and estimates the border ownership (i.e. which side of the boundary is foreground and which is background). We represent occlusion relations by a binary edge map, to indicate the object boundary, and an occlusion orientation variable which is tangential to the boundary and whose direction specifies border ownership by a left-hand rule. We train two related deep convolutional neural networks, called DOC, which exploit local and non-local image cues to estimate this representation and hence recover occlusion relations. In order to train and test DOC we construct a large-scale instance occlusion boundary dataset using PASCAL VOC images, which we call the PASCAL instance occlusion dataset (PIOD). This contains 10,000 images and hence is two orders of magnitude larger than existing occlusion datasets for outdoor images. We test two variants of DOC on PIOD and on the BSDS occlusion dataset and show they outperform state-of-the-art methods. Finally, we perform numerous experiments investigating multiple settings of DOC and transfer between BSDS and PIOD, which provides more insights for further study of occlusion estimation.

cs.CV↗

arXiv2016

Visual Question Answering: A Survey of Methods and Datasets

Visual Question Answering (VQA) is a challenging task that has received increasing attention from both the computer vision and the natural language processing communities. Given an image and a question in natural language, it requires reasoning over visual elements of the image and general knowledge to infer the correct answer. In the first part of this survey, we examine the state of the art by comparing modern approaches to the problem. We classify methods by their mechanism to connect the visual and textual modalities. In particular, we examine the common approach of combining convolutional and recurrent neural networks to map images and questions to a common feature space. We also discuss memory-augmented and modular architectures that interface with structured knowledge bases. In the second part of this survey, we review the datasets available for training and evaluating VQA systems. The various datatsets contain questions at different levels of complexity, which require different capabilities and types of reasoning. We examine in depth the question/answer pairs from the Visual Genome project, and evaluate the relevance of the structured annotations of images with scene graphs for VQA. Finally, we discuss promising future directions for the field, in particular the connection to structured knowledge bases and the use of natural language processing models.

cs.CV↗

arXiv2016

Willmore surfaces in spheres via loop groups II: a coarse classification of Willmore two-spheres by potentials

Applying the DPW version of the theory developed by Burstall and Guest for harmonic maps of finite uniton type, we derive a coarse classification of Willmore two-spheres in $S^{n+2}$ in terms of the normalized potential of their (harmonic) conformal Gauss maps. Moreover, for the case of $S^6$, some geometric properties of the corresponding Willmore two-spheres are discussed. The classical classification of Willmore two-spheres in $S^4$ are also derived as a corollary.

math.DG↗

arXiv2016

Representing Extended Finite State Machines for SDL by A Novel Control Model of Discrete Event Systems

This paper discusses EFSM for SDL and transforms EFSM into a novel control model of discrete event systems. We firstly propose a control model of discrete event systems, where the event set is made up of several conflicting pairs and control is implemented to select one event of the pair. Then we transform EFSM for SDL to the control model to clarify the control mechanism functioning in SDL flow graphs. This work views the EFSM for SDL in the perspective of supervisory control theory, and this contributes to the field of software cybernetics, which explores the theoretically justified interplay of software and the control.

cs.FL↗

arXiv2016

Where to Focus: Query Adaptive Matching for Instance Retrieval Using Convolutional Feature Maps

Instance retrieval requires one to search for images that contain a particular object within a large corpus. Recent studies show that using image features generated by pooling convolutional layer feature maps (CFMs) of a pretrained convolutional neural network (CNN) leads to promising performance for this task. However, due to the global pooling strategy adopted in those works, the generated image feature is less robust to image clutter and tends to be contaminated by the irrelevant image patterns. In this article, we alleviate this drawback by proposing a novel reranking algorithm using CFMs to refine the retrieval result obtained by existing methods. Our key idea, called query adaptive matching (QAM), is to first represent the CFMs of each image by a set of base regions which can be freely combined into larger regions-of-interest. Then the similarity between the query and a candidate image is measured by the best similarity score that can be attained by comparing the query feature and the feature pooled from a combined region. We show that the above procedure can be cast as an optimization problem and it can be solved efficiently with an off-the-shelf solver. Besides this general framework, we also propose two practical ways to create the base regions. One is based on the property of the CFM and the other one is based on a multi-scale spatial pyramid scheme. Through extensive experiments, we show that our reranking approaches bring substantial performance improvement and by applying them we can outperform the state of the art on several instance retrieval benchmarks.

cs.CV↗

arXiv2016

Photoluminescence of InGaAs/GaAsBi/InGaAs type-II quantum well grown by gas source molecular beam epitaxy

InGaAs/GaAsBi/InGaAs quantum wells (QWs) were grown on GaAs substrates by gas source molecular beam epitaxy for realizing the type II band-edge line-up. Both type I and type II transitions were observed in the Bi containing W QWs and the photoluminescence intensity was enhanced in the sample with a high Bi content, which is mainly due to the improvement of carrier confinement. Blue-shift of type II transitions at high excitation power density was observed and ascribed to the band-bending effect. The calculated transition energies based on 8 band k.p model fit well with the experiment results. The experimental and theoretical results show that the type-II QW design is a new promising candidate for realizing long wavelength GaAs-based light emitting devices near 1.3 um.

cond-mat.mtrl-sci↗

arXiv2016

A Low-Complexity Transceiver Design in Sparse Multipath Massive MIMO Channels

In this letter, we develop a low-complexity transceiver design, referred to as semi-random beam pairing (SRBP), for sparse multipath massive MIMO channels. By exploring a sparse representation of the MIMO channel in the virtual angular domain, we generate a set of transmit-receive beam pairs in a semi-random way to support the simultaneous transmission of multiple data streams. These data streams can be easily separated at the receiver via a successive interference cancelation (SIC) technique, and the power allocation among them are optimized based on the classical waterfilling principle. The achieved degree of freedom (DoF) and capacity of the proposed approach are analyzed. Simulation results show that, compared to the conventional singular value decomposition (SVD)-based method, the proposed transceiver design can achieve near-optimal DoF and capacity with a significantly lower computational complexity.

cs.IT↗

arXiv2016

Pushing the Limits of Deep CNNs for Pedestrian Detection

Compared to other applications in computer vision, convolutional neural networks have under-performed on pedestrian detection. A breakthrough was made very recently by using sophisticated deep CNN models, with a number of hand-crafted features, or explicit occlusion handling mechanism. In this work, we show that by re-using the convolutional feature maps (CFMs) of a deep convolutional neural network (DCNN) model as image features to train an ensemble of boosted decision models, we are able to achieve the best reported accuracy without using specially designed learning algorithms. We empirically identify and disclose important implementation details. We also show that pixel labelling may be simply combined with a detector to boost the detection performance. By adding complementary hand-crafted features such as optical flow, the DCNN based detector can be further improved. We set a new record on the Caltech pedestrian dataset, lowering the log-average miss rate from $11.7\%$ to $8.9\%$, a relative improvement of $24\%$. We also achieve a comparable result to the state-of-the-art approaches on the KITTI dataset.

cs.CV↗

arXiv2016

Fabrication of polarization-independent single-mode waveguides in lithium niobate crystal with femtosecond laser pulses

We report on fabrication of depressed cladding optical waveguides buried in lithium niobate crystal with shaped femtosecond laser pulses. Depressed cladding waveguides of variable mode-field sizes are fabricated by forming the four sides of the cladding using a slit-beam shaping technique. We show that the waveguides fabricated by our technique allows single-mode propagation of the light polarized in both vertical and horizontal directions.

physics.optics↗

arXiv2016

Benchmarking NFV Software Dataplanes

A key enabling technology of NFV is software dataplane, which has attracted much attention in both academia and industry recently. Yet, till now there is little understanding about its performance in practice. In this paper, we make a benchmark measurement study of NFV software dataplanes in terms of packet processing capability, one of the most fundamental and critical performance metrics. Specifically, we compare two state-of-the-art open-source NFV dataplanes, SoftNIC and ClickOS, using commodity 10GbE NICs under various typical workloads. Our key observations are that (1) both dataplanes have performance issues processing small (<=128B) packets; (2) it is not always best to put all VMs of a service chain on one server due to NUMA effect. We propose resource allocation strategies to remedy the problems, including carefully adding CPU cores and vNICs to VMs, and spreading VMs of a service chain to separate servers. To fundamentally address these problems and scale their performance, SoftNIC and ClickOS could improve the support for NIC queues and multiple cores.

cs.NI↗

arXiv2016

On the Björling problem for Willmore surfaces

We solve the analogue of Björling's problem for Willmore surfaces via a harmonic map representation. For the umbilic-free case the problem and solution are as follows: given a real analytic curve $y_0$ in $S^3$, together with the prescription of the values of the surface normal and the dual Willmore surface along the curve, lifted to the light cone in Minkowski $5$-space $\mathbb{R}^5_1$, we prove, using isotropic harmonic maps, that there exists a unique pair of dual Willmore surfaces $y$ and $\hat y$ satisfying the given values along the curve. We give explicit formulae for the generalized Weierstrass data for the surface pair. For the three dimensional target, we use the solution to explicitly describe the Weierstrass data, in terms of geometric quantities, for all equivariant Willmore surfaces. For the case that the surface has umbilic points, we apply the more general half-isotropic harmonic maps introduced by Hélein to derive a solution: in this case the map $\hat y$ is not necessarily the dual surface, and the additional data of a derivative of $\hat y$ must be prescribed. This solution is generalized to higher codimensions.

math.DG↗

arXiv2016

Analytical Model for Outdoor Millimeter Wave Channels using Geometry-Based Stochastic Approach

The severe bandwidth shortage in conventional microwave bands has spurred the exploration of the millimeter wave (MMW) spectrum for the next revolution in wireless communications. However, there is still lack of proper channel modeling for the MMW wireless propagation, especially in the case of outdoor environments. In this paper, we develop a geometry-based stochastic channel model to statistically characterize the effect of all the first-order reflection paths between the transmitter and receiver. These first-order reflections are generated by the single-bounce of signals reflected from the walls of randomly distributed buildings. Based on this geometric model, a closed-form expression for the power delay profile (PDP) contributed by all the first-order reflection paths is obtained and then used to evaluate their impact on the MMW outdoor propagation characteristics. Numerical results are provided to validate the accuracy of the proposed model under various channel parameter settings. The findings in this paper provide a promising step towards more complex and practical MMW propagation channel modeling.

stat.AP↗

arXiv2016

Transverse writing of three-dimensional tubular optical waveguides in glass with slit-shaped femtosecond laser beams

We report on fabrication of tubular optical waveguides buried in ZBLAN glass based on transverse femtosecond laser direct writing. Irradiation in ZBLAN with focused femtosecond laser pulses leads to decrease of refractive index in the modified region. Tubular optical waveguides of variable mode areas are fabricated by forming the four sides of the cladding with slit-shaped femtosecond laser pulses, ensuring single mode waveguiding with a mode field dimension as small as ~ 4 μm.

physics.optics↗

arXiv2016

Large-scale Binary Quadratic Optimization Using Semidefinite Relaxation and Applications

In computer vision, many problems such as image segmentation, pixel labelling, and scene parsing can be formulated as binary quadratic programs (BQPs). For submodular problems, cuts based methods can be employed to efficiently solve large-scale problems. However, general nonsubmodular problems are significantly more challenging to solve. Finding a solution when the problem is of large size to be of practical interest, however, typically requires relaxation. Two standard relaxation methods are widely used for solving general BQPs--spectral methods and semidefinite programming (SDP), each with their own advantages and disadvantages. Spectral relaxation is simple and easy to implement, but its bound is loose. Semidefinite relaxation has a tighter bound, but its computational complexity is high, especially for large scale problems. In this work, we present a new SDP formulation for BQPs, with two desirable properties. First, it has a similar relaxation bound to conventional SDP formulations. Second, compared with conventional SDP methods, the new SDP formulation leads to a significantly more efficient and scalable dual optimization approach, which has the same degree of complexity as spectral methods. We then propose two solvers, namely, quasi-Newton and smoothing Newton methods, for the dual problem. Both of them are significantly more efficiently than standard interior-point methods. In practice, the smoothing Newton solver is faster than the quasi-Newton solver for dense or medium-sized problems, while the quasi-Newton solver is preferable for large sparse/structured problems. Our experiments on a few computer vision applications including clustering, image segmentation, co-segmentation and registration show the potential of our SDP formulation for solving large-scale BQPs.

cs.CV↗

arXiv2016

Ask Me Anything: Free-form Visual Question Answering Based on Knowledge from External Sources

We propose a method for visual question answering which combines an internal representation of the content of an image with information extracted from a general knowledge base to answer a broad range of image-based questions. This allows more complex questions to be answered using the predominant neural network-based approach than has previously been possible. It particularly allows questions to be asked about the contents of an image, even when the image itself does not contain the whole answer. The method constructs a textual representation of the semantic content of an image, and merges it with textual information sourced from a knowledge base, to develop a deeper understanding of the scene viewed. Priming a recurrent neural network with this combined information, and the submitted question, leads to a very flexible visual question answering approach. We are specifically able to answer questions posed in natural language, that refer to information not contained in the image. We demonstrate the effectiveness of our model on two publicly available datasets, Toronto COCO-QA and MS COCO-VQA and show that it produces the best reported results in both cases.

cs.CV↗

arXiv2016

Studying on Opinion Evolution by Hamilton-Jacobi Equation

A physical description of an opinion evolution is conducted based on the Hamilton-Jacobi equation derived from a generalized potential and the corresponding Langevin equation. The investigation mainly focuses on the heterogeneities such as age, connection circle and overall quality of the participants involved in the opinion exchange process. The evolutionary patterns of opinion can be described by solution of the Hamilton-Jacobi equation, information entropy. The results show that the overall qualities of the participants play critical roles in forming an opinion. The higher the overall quality is, the easier the consensus can reach. The solution also demonstrates that the age and the connection circle of the agents play equally important roles in forming an opinion. The essence of the age, overall quality, and connection circle corresponds to the maturity of thought (opinion inertia), reason and intelligence, influence strength of the environment, respectively. So the information entropy distributes in the phase planes (two variable combinations out of the three---age, connection circle and overall quality) exhibit the evolution rules of an opinion.

physics.soc-ph↗

arXiv2016

Willmore surfaces in spheres via loop groups $I$: generic cases and some examples

In this paper we deal with the global properties of Willmore surfaces in spheres via the harmonic conformal Gauss map using loop groups. We first derive a global description of those harmonic maps which can be realized as conformal Gauss maps of some Willmore surfaces (Theorem 3.4, Theorem 3.11 and Theorem 3.18). Then we introduce the DPW procedure for these harmonic maps, and state appropriate versions of the Iwasawa decomposition and the Birkhoff decomposition Theorems. In particular, we show how the harmonic maps associated with Willmore surfaces can be constructed in terms of loop groups. The third main result, which has many implications for the case of Willmore surfaces in spheres, shows that every harmonic map into some non-compact inner symmetric space $G/K$ induces a harmonic map into the compact dual inner symmetric space $U/{(U \cap K^\mathbb{C})}$. From this correspondence we obtain additional information about the global properties of harmonic maps into non-compact inner symmetric spaces. As an illustration of the theory developed in this paper we list examples (some of which were worked out in separate papers by following the theory of the present paper). In particular, we present an explicit, unbranched (isotropic) Willmore sphere in $S^6$ which is not S-Willmore, and thus does not have a dual Willmore surface. This example gives a negative answer to a long open problem (originally posed by Ejiri).

math.DG↗

arXiv2016

Construction of Willmore two-spheres via harmonic maps into $SO^+(1,n+3)/(SO^+(1,1)\times SO(n+2))$

This paper aims to provide a description of totally isotropic Willmore two-spheres and their adjoint transforms. We first recall the isotropic harmonic maps which are introduced by Hélein, Xia-Shen and Ma for the study of Willmore surfaces. Then we derive a description of the normalized potential (some Lie algebra valued meromorphic 1-forms) of totally isotropic Willmore two-spheres in terms of the isotropic harmonic maps. In particular, the corresponding isotropic harmonic maps are of finite uniton type. The proof also contains a concrete way to construct examples of totally isotropic Willmore two-spheres and their adjoint transforms. As illustrations, two kinds of examples are obtained this way.

math.DG↗