arXiv subjects

Bo Wan

Publications and source records attributed to Bo Wan.

24 records · Page 2Linked to original sources

arXiv2019

Learning Cross-modal Context Graph for Visual Grounding

Visual grounding is a ubiquitous building block in many vision-language tasks and yet remains challenging due to large variations in visual and linguistic features of grounding entities, strong context effect and the resulting semantic ambiguities. Prior works typically focus on learning representations of individual phrases with limited context information. To address their limitations, this paper proposes a language-guided graph representation to capture the global context of grounding entities and their relations, and develop a cross-modal graph matching strategy for the multiple-phrase visual grounding task. In particular, we introduce a modular graph neural network to compute context-aware representations of phrases and object proposals respectively via message propagation, followed by a graph-based matching module to generate globally consistent localization of grounding phrases. We train the entire graph neural network jointly in a two-stage strategy and evaluate it on the Flickr30K Entities benchmark. Extensive experiments show that our method outperforms the prior state of the arts by a sizable margin, evidencing the efficacy of our grounding framework. Code is available at "https://github.com/youngfly11/LCMCG-PyTorch".

cs.CV↗

arXiv2019

Pose-aware Multi-level Feature Network for Human Object Interaction Detection

Reasoning human object interactions is a core problem in human-centric scene understanding and detecting such relations poses a unique challenge to vision systems due to large variations in human-object configurations, multiple co-occurring relation instances and subtle visual difference between relation categories. To address those challenges, we propose a multi-level relation detection strategy that utilizes human pose cues to capture global spatial configurations of relations and as an attention mechanism to dynamically zoom into relevant regions at human part level. Specifically, we develop a multi-branch deep network to learn a pose-augmented relation representation at three semantic levels, incorporating interaction context, object features and detailed semantic part cues. As a result, our approach is capable of generating robust predictions on fine-grained human object interactions with interpretable outputs. Extensive experimental evaluations on public benchmarks show that our model outperforms prior methods by a considerable margin, demonstrating its efficacy in handling complex scenes.

cs.CV↗

arXiv2018

Rules for Phase Shifts of Quantum Oscillations in Topological Nodal-line Semimetals

Nodal-line semimetals are topological semimetals in which band touchings form nodal lines or rings. Around a loop that encloses a nodal line, an electron can accumulate a nontrivial $π$ Berry phase, so the phase shift in the Shubnikov-de Haas (SdH) oscillation may give a transport signature for the nodal-line semimetals. However, different experiments have reported contradictory phase shifts, in particular, in the WHM nodal-line semimetals (W=Zr/Hf, H=Si/Ge, M=S/Se/Te). For a generic model of nodal-line semimetals, we present a systematic calculation for the SdH oscillation of resistivity under a magnetic field normal to the nodal-line plane. From the analytical result of the resistivity, we extract general rules to determine the phase shifts for arbitrary cases and apply them to ZrSiS and Cu$_3$PdN systems. Depending on the magnetic field directions, carrier types, and cross sections of the Fermi surface, the phase shift shows rich results, quite different from those for normal electrons and Weyl fermions. Our results may help exploring transport signatures of topological nodal-line semimetals and can be generalized to other topological phases of matter.

cond-mat.mes-hall↗

arXiv2017

Cleavage Tendency of Anisotropic Two Dimensional Materials: ReX2 (X=S, Se) and WTe2

With unique distorted 1T structure and the associated in-plane anisotropic properties, mono- and few-layer ReX2 (X=S, Se) have recently attracted particular interest. Based on experiment and first-principles calculations, we investigate the fracture behavior of ReX2. We find that the cleaved edges of ReX2 flakes usually form an angle of ~120° or ~60°. In order to understand such phenomenon, we perform comprehensive investigations on the uniaxial tensile stress-strain relation of monolayer and multi-layer ReX2 sheets. Our numerical calculation shows that the particular cleaved edges of ReX2 flakes are caused by unique anisotropic ultimate tensile strengths and critical strains. We also calculate the stress-strain relation of WTe2, which explains why their cleaved edges are not corresponding to the principle axes. Our proposed mechanism about the fracture angle has also been supported by the calculated cleavage energies and surface energies for different edge surfaces.

cond-mat.mtrl-sci↗

arXiv2015

Study of digital pulse shape discrimination method for n-γ separation of EJ-301 liquid scintillation detector

A digital pulse shape discrimination system based on a programmable module NI-5772 has been established and tested with EJ-301 liquid scintillation detector. The module was operated by means of running programs developed in LabVIEW with the sampling frequency up to 1.6GS/s. Standard gamma sources 22Na, 137Cs and 60Co were used to calibrate the EJ-301 liquid scintillation detector, and the gamma response function has been obtained. Digital algorithms for charge comparison method and zero-crossing method have been developed. The experimental results showed that both digital signal processing (DSP) algorithms could discriminate neutrons from gamma-rays. Moreover, the zero-crossing method shows better n-γ discrimination at 80 keVee and lower, whereas the charge comparison method gives better results at higher thresholds. In addition, the figure-of-merit (FOM) of two different dimension detectors were extracted at 9 energy thresholds, and it was found that the smaller one presented a better n-γ separation property for fission neutrons.

physics.ins-det↗

arXiv2014

Dirac and Weyl Semimetal in XYBi (X=Ba, Eu; Y=Cu, Ag and Au)

Weyl and Dirac semimetals recently stimulate intense research activities due to their novel properties. Combining first-principles calculations and effective model analysis, we predict that nonmagnetic compounds BaYBi (Y=Au, Ag and Cu) are Dirac semimetals. As for the magnetic compound EuYBi, although the time reversal symmetry is broken, their long-range magnetic ordering cannot split the Dirac point into pairs of Weyl points. However, we propose that partially substitute Eu ions by Ba ions will realize the Weyl semimetal.

cond-mat.mtrl-sci↗