arXiv subjects

Su Xu

Publications and source records attributed to Su Xu.

9 recordsLinked to original sources

arXiv2026

USV: Towards Understanding the User-generated Short-form Videos

Several large-scale video datasets have been published these years and have advanced the area of video understanding. However, the newly emerged user-generated short-form videos have rarely been studied. This paper presents USV, the User-generated Short-form Video dataset for high-level semantic video understanding. The dataset contains around 224K videos collected from UGC platforms by label queries without extra manual verification and trimming. Although video understanding has achieved plausible improvement these years, most works focus on instance-level recognition, which is not sufficient for learning the representation of the high-level semantic information of videos. Therefore, we further establish two tasks: topic recognition and video-text retrieval on USV. We propose two unified and effective baseline methods Multi-Modality Fusion Network (MMF-Net) and Video-Text Contrastive Learning (VTCL), to tackle the topic recognition task and video-text retrieval respectively, and carry out comprehensive benchmarks to facilitate future research. Our project page is https://usvdataset.github.io.

cs.CV↗

arXiv2023

Reconfigurable Meta-Radiator Based on Flexible Mechanically Controlled Current Distribution in Three-dimensional Space

In this paper, we provide an experimental proof-of-concept of this dynamic 3D current manipulation through a 3D-printed reconfigurable meta-radiator with periodically slotted current elements. By utilizing the working frequency and the mechanical configuration comprehensively, the radiation pattern can be switched among 12 states. Inspired by maximum likelihood method in digital communications, a robustness-analysis method is proposed to evaluate the potential error ratio between ideal cases and practice. Our work provides a previously unidentified model for next-generation information distribution and terahertz-infrared wireless communications.

physics.app-ph↗

arXiv2022

Generalizable Neural Performer: Learning Robust Radiance Fields for Human Novel View Synthesis

This work targets at using a general deep learning framework to synthesize free-viewpoint images of arbitrary human performers, only requiring a sparse number of camera views as inputs and skirting per-case fine-tuning. The large variation of geometry and appearance, caused by articulated body poses, shapes and clothing types, are the key bottlenecks of this task. To overcome these challenges, we present a simple yet powerful framework, named Generalizable Neural Performer (GNR), that learns a generalizable and robust neural body representation over various geometry and appearance. Specifically, we compress the light fields for novel view human rendering as conditional implicit neural radiance fields from both geometry and appearance aspects. We first introduce an Implicit Geometric Body Embedding strategy to enhance the robustness based on both parametric 3D human body model and multi-view images hints. We further propose a Screen-Space Occlusion-Aware Appearance Blending technique to preserve the high-quality appearance, through interpolating source view appearance to the radiance fields with a relax but approximate geometric guidance. To evaluate our method, we present our ongoing effort of constructing a dataset with remarkable complexity and diversity. The dataset GeneBody-1.0, includes over 360M frames of 370 subjects under multi-view cameras capturing, performing a large variety of pose actions, along with diverse body shapes, clothing, accessories and hairdos. Experiments on GeneBody-1.0 and ZJU-Mocap show better robustness of our methods than recent state-of-the-art generalizable methods among all cross-dataset, unseen subjects and unseen poses settings. We also demonstrate the competitiveness of our model compared with cutting-edge case-specific ones. Dataset, code and model will be made publicly available.

cs.CV↗

arXiv2021

Polarization-Orthogonal Nondegenerate Plasmonic Higher-Order Topological States

Photonic topological states, providing light-manipulation approaches in robust manners, have attracted intense attention. Connecting photonic topological states with far-field degrees of freedom(DoFs) has given rise to fruitful phenomena. Recently emerged higher-order topological insulators (HOTIs), hosting boundary states two or more dimensions lower than those of bulk, offer new paradigms to localize/transport light topologically in extended dimensionalities. However, photonic HOTIs have not been related to DoFs of radiation fields yet. Here, we report the observation of polarization-orthogonal second-order topological corner states at different frequencies on a designer-plasmonic Kagome metasurface in the far field.Such phenomenon stands on two mechanisms, i.e., projecting the far-field polarizations to the intrinsic parity DoFs of lattice modes and the parity splitting of the plasmonic corner states in spectra. We theoretically and numerically show that the parity splitting originates from the underlying inter-orbital coupling. Both near-field and far-field experiments verify the polarization-orthogonal nondegeneratesecond-order topological corner states. These results promise applications in robust optical single photon emitters and multiplexed photonic devices.

physics.optics↗

arXiv2020

Reconfigurable Slotted Antenna Inspired by Multidimensional Modulation

Multidimensional modulation was widely studied in the past decades due to the explosive development of modern wireless communication. Here, we propose a spirally reconfigurable slotted antenna inspired by the multidimensional modulation. The amplitude, phase, and frequency-shift modulation are analog by integrating three-dimensional mechanical switching to a spiral slotted antenna. The maximum gain of the reconfigurable antenna can be adjusted in the type of 2 encoding bits at 9.5 GHz. Our work may pave the way to a high-performance reconfigurable antenna for 6G communication.

physics.app-ph↗

arXiv2018

Controlling high-Q trapped modes in polarization-insensitive all-dielectric metasurfaces

We reveal peculiarities of the trapped (dark) mode excitation in a polarization-insensitive all-dielectric metasurface, whose unit super-cell is constructed by particularly arranging four cylindrical dielectric particles. Involving group-theoretical description we discuss in detail the effect of different orientations of particles within the super-cell on characteristics of the trapped mode. The theoretical predictions are confirmed by numerical simulations and experimental investigations. Since the metasurface is realized from simple dielectric particles without the use of any metallic components, they are feasibly scalable to both micro- and nanometer-size structures, and they can be employed in flat-optics platforms for realizing efficient light-matter interaction for multiple hotspot light localization, optical sensing, and highly-efficient light trapping.

physics.optics↗

arXiv2018

3D-Printed Chiral Metasurface as Dichroic Dual-Band Polarization Converter

We propose a novel design of a true 3D chiral metasurface behaving as a spatial polarization converter with asymmetric transmission. The metasurface is made of a lattice of metallic sesquialteral (one and a half pitch) helical particles. Each particle contains six rectangular bars arranged in a series one above the another creating a helix The proposed metasurface exhibits a dual-band asymmetric transmission accompanied by the effect of a complete polarization conversion in the response on the particular distributions of currents induced in the particle's bars by an incident wave. Regarding circularly polarized waves the metasurface demonstrates a strong circular dichroism. A prototype of the metasurface is manufactured for the microwave experiment by using 3D-printing technique utilizing Cobalt-Chromium alloy, which exhibits good performances against thermal fatigue and corrosion at high temperatures. Our work paves the way to find an industrial solution on fabricating communication components with efficient polarization conversion for extreme environments.

physics.acc-ph↗

arXiv2017

High-quality trapped modes in all-dielectric metamaterials

A planar all-dielectric metamaterial made of a double-periodic lattice whose unit cell consists of a single subwavelength dielectric particle having the form of a disk possessing a penetrating hole is considered. The resonant states in the transmitted spectra of the metamaterial are identified considering modes inherent to the individual cylindrical dielectric resonator. A correlation between the asymmetry in particle's geometry, which arises from the off-centered displacement of the hole, and formation of the Mie-type and trapped modes is established. The advantages of using a coaxial-sector notch instead of a round hole for the trapped mode excitation are explained.

physics.class-ph↗

arXiv2012

Experimental demonstration of a free space cylindrical cloak without superluminal propagation

We experimentally demonstrated an alternative approach of invisibility cloaking that can combine technical advantages of all current major cloaking strategies in a unified manner and thus can solve bottlenecks of individual strategies. A broadband cylindrical invisibility cloak in free space is designed based on scattering cancellation (the approach of previous plasmonic cloaking), and implemented with anisotropic metamaterials (a fundamental property of singular-transformation cloaks). Particularly, non-superluminal propagation of electromagnetic waves, a superior advantage of non-Euclidian-transformation cloaks constructed with complex branch cuts, is inherited in this design, and thus is the reason of its relatively broad bandwidth. This demonstration provides the possibility for future practical implementation of cloaking devices at large scales in free space.

physics.class-ph↗