arXiv subjects

Hong Zhou

Publications and source records attributed to Hong Zhou.

At least 73 records · Page 4Linked to original sources

arXiv2022

Polygon-free: Unconstrained Scene Text Detection with Box Annotations

Although a polygon is a more accurate representation than an upright bounding box for text detection, the annotations of polygons are extremely expensive and challenging. Unlike existing works that employ fully-supervised training with polygon annotations, this study proposes an unconstrained text detection system termed Polygon-free (PF), in which most existing polygon-based text detectors (e.g., PSENet [33],DB [16]) are trained with only upright bounding box annotations. Our core idea is to transfer knowledge from synthetic data to real data to enhance the supervision information of upright bounding boxes. This is made possible with a simple segmentation network, namely Skeleton Attention Segmentation Network (SASN), that includes three vital components (i.e., channel attention, spatial attention and skeleton attention map) and one soft cross-entropy loss. Experiments demonstrate that the proposed Polygonfree system can combine general detectors (e.g., EAST, PSENet, DB) to yield surprisingly high-quality pixel-level results with only upright bounding box annotations on a variety of datasets (e.g., ICDAR2019-Art, TotalText, ICDAR2015). For example, without using polygon annotations, PSENet achieves an 80.5% F-score on TotalText [3] (vs. 80.9% of fully supervised counterpart), 31.1% better than training directly with upright bounding box annotations, and saves 80%+ labeling costs. We hope that PF can provide a new perspective for text detection to reduce the labeling costs. The code can be found at https://github.com/weijiawu/Unconstrained-Text-Detection-with-Box-Supervisionand-Dynamic-Self-Training.

cs.CV↗

arXiv2022

The Observability in Unobservable Systems

In this paper, we introduce the concept of observability of targeted state variables for systems that may not be fully observable. For their estimation, we introduce and exemplify a deep filter, which is a neural network specifically designed for the estimation of targeted state variables without computing the trajectory of the entire system. The observability definition is quantitative rather than a yes or no answer so that one can compare the level of observability between different sensor locations.

eess.SY↗

arXiv2021

A Bilingual, OpenWorld Video Text Dataset and End-to-end Video Text Spotter with Transformer

Most existing video text spotting benchmarks focus on evaluating a single language and scenario with limited data. In this work, we introduce a large-scale, Bilingual, Open World Video text benchmark dataset(BOVText). There are four features for BOVText. Firstly, we provide 2,000+ videos with more than 1,750,000+ frames, 25 times larger than the existing largest dataset with incidental text in videos. Secondly, our dataset covers 30+ open categories with a wide selection of various scenarios, e.g., Life Vlog, Driving, Movie, etc. Thirdly, abundant text types annotation (i.e., title, caption or scene text) are provided for the different representational meanings in video. Fourthly, the BOVText provides bilingual text annotation to promote multiple cultures live and communication. Besides, we propose an end-to-end video text spotting framework with Transformer, termed TransVTSpotter, which solves the multi-orient text spotting in video with a simple, but efficient attention-based query-key mechanism. It applies object features from the previous frame as a tracking query for the current frame and introduces a rotation angle prediction to fit the multiorient text instance. On ICDAR2015(video), TransVTSpotter achieves the state-of-the-art performance with 44.1% MOTA, 9 fps. The dataset and code of TransVTSpotter can be found at github:com=weijiawu=BOVText and github:com=weijiawu=TransVTSpotter, respectively.

cs.CV↗

arXiv2021

Broadband beam steering for misaligned multi-mode OAM communication systems

Orbital angular momentum (OAM) at radio frequency (RF) has attracted more and more attention as a novel approach of multiplexing a set of orthogonal OAM modes on the same frequency channel to achieve high spectral efficiency (SE). However, the precondition for maintaining the orthogonality among different OAM modes is perfect alignment of the transmit and receive uniform circular arrays (UCAs), which is difficult to be satisfied in practical wireless communication scenario. Therefore, to achieve available multi-mode OAM broadband wireless communication, we first investigate the effect of oblique angles on the transmission performance of the multi-mode OAM broadband system in the non-parallel misalignment case. Then, we compare the UCA-based RF analog and baseband digital transceiver structures and corresponding beam steering schemes. Mathematical analysis and numerical simulations validate that the SE of the misaligned multi-mode OAM broadband system is quite low, while analog and digital beam steering both can significantly improve the SE of the system. However, digital beam steering can obtain higher SE than analog beam steering especially when the bandwidth and the number of array elements are large, which validates that baseband digital transceiver with digital beam steering is more suitable for multi-mode OAM broadband wireless communication systems in practice.

cs.IT↗

arXiv2021

Divide-and-Assemble: Learning Block-wise Memory for Unsupervised Anomaly Detection

Reconstruction-based methods play an important role in unsupervised anomaly detection in images. Ideally, we expect a perfect reconstruction for normal samples and poor reconstruction for abnormal samples. Since the generalizability of deep neural networks is difficult to control, existing models such as autoencoder do not work well. In this work, we interpret the reconstruction of an image as a divide-and-assemble procedure. Surprisingly, by varying the granularity of division on feature maps, we are able to modulate the reconstruction capability of the model for both normal and abnormal samples. That is, finer granularity leads to better reconstruction, while coarser granularity leads to poorer reconstruction. With proper granularity, the gap between the reconstruction error of normal and abnormal samples can be maximized. The divide-and-assemble framework is implemented by embedding a novel multi-scale block-wise memory module into an autoencoder network. Besides, we introduce adversarial learning and explore the semantic latent representation of the discriminator, which improves the detection of subtle anomaly. We achieve state-of-the-art performance on the challenging MVTec AD dataset. Remarkably, we improve the vanilla autoencoder model by 10.1% in terms of the AUROC score.

cs.CV↗

arXiv2020

A Local Search Framework for Experimental Design

We present a local search framework to design and analyze both combinatorial algorithms and rounding algorithms for experimental design problems. This framework provides a unifying approach to match and improve all known results in D/A/E-design and to obtain new results in previously unknown settings. For combinatorial algorithms, we provide a new analysis of the classical Fedorov's exchange method. We prove that this simple local search algorithm works well as long as there exists an almost optimal solution with good condition number. Moreover, we design a new combinatorial local search algorithm for E-design using the regret minimization framework. For rounding algorithms, we provide a unified randomized exchange algorithm to match and improve previous results for D/A/E-design. Furthermore, the algorithm works in the more general setting to approximately satisfy multiple knapsack constraints, which can be used for weighted experimental design and for incorporating fairness constraints into experimental design.

cs.DS↗

arXiv2020

A Spectral Approach to Network Design

We present a spectral approach to design approximation algorithms for network design problems. We observe that the underlying mathematical questions are the spectral rounding problems, which were studied in spectral sparsification and in discrepancy theory. We extend these results to incorporate additional non-negative linear constraints, and show that they can be used to significantly extend the scope of network design problems that can be solved. Our algorithm for spectral rounding is an iterative randomized rounding algorithm based on the regret minimization framework. In some settings, this provides an alternative spectral algorithm to achieve constant factor approximation for the classical survivable network design problem, and partially answers a question of Bansal about survivable network design with concentration property. We also show many other applications of the spectral rounding results, including weighted experimental design and additive spectral sparsification.

cs.DS↗

arXiv2019

Orbital Angular Momentum Waves: Generation, Detection and Emerging Applications

Orbital angular momentum (OAM) has aroused a widespread interest in many fields, especially in telecommunications due to its potential for unleashing new capacity in the severely congested spectrum of commercial communication systems. Beams carrying OAM have a helical phase front and a field strength with a singularity along the axial center, which can be used for information transmission, imaging and particle manipulation. The number of orthogonal OAM modes in a single beam is theoretically infinite and each mode is an element of a complete orthogonal basis that can be employed for multiplexing different signals, thus greatly improving the spectrum efficiency. In this paper, we comprehensively summarize and compare the methods for generation and detection of optical OAM, radio OAM and acoustic OAM. Then, we represent the applications and technical challenges of OAM in communications, including free-space optical communications, optical fiber communications, radio communications and acoustic communications. To complete our survey, we also discuss the state of art of particle manipulation and target imaging with OAM beams.

physics.app-ph↗

arXiv2019

TextCohesion: Detecting Text for Arbitrary Shapes

In this paper, we propose a pixel-wise method named TextCohesion for scene text detection, which splits a text instance into five key components: a Text Skeleton and four Directional Pixel Regions. These components are easier to handle than the entire text instance. A confidence scoring mechanism is designed to filter characters that are similar to text. Our method can integrate text contexts intensively when backgrounds are complex. Experiments on two curved challenging benchmarks demonstrate that TextCohesion outperforms state-of-the-art methods, achieving the F-measure of 84.6% on Total-Text and bfseries86.3% on SCUT-CTW1500.

cs.CV↗

arXiv2019

Network design for s-t effective resistance

We consider a new problem of designing a network with small $s$-$t$ effective resistance. In this problem, we are given an undirected graph $G=(V,E)$, two designated vertices $s,t \in V$, and a budget $k$. The goal is to choose a subgraph of $G$ with at most $k$ edges to minimize the $s$-$t$ effective resistance. This problem is an interpolation between the shortest path problem and the minimum cost flow problem and has applications in electrical network design. We present several algorithmic and hardness results for this problem and its variants. On the hardness side, we show that the problem is NP-hard, and the weighted version is hard to approximate within a factor smaller than two assuming the small-set expansion conjecture. On the algorithmic side, we analyze a convex programming relaxation of the problem and design a constant factor approximation algorithm. The key of the rounding algorithm is a randomized path-rounding procedure based on the optimality conditions and a flow decomposition of the fractional solution. We also use dynamic programming to obtain a fully polynomial time approximation scheme when the input graph is a series-parallel graph, with better approximation ratio than the integrality gap of the convex program for these graphs.

cs.DS↗

arXiv2018

Brain Tumor Segmentation Based on Refined Fully Convolutional Neural Networks with A Hierarchical Dice Loss

As a basic task in computer vision, semantic segmentation can provide fundamental information for object detection and instance segmentation to help the artificial intelligence better understand real world. Since the proposal of fully convolutional neural network (FCNN), it has been widely used in semantic segmentation because of its high accuracy of pixel-wise classification as well as high precision of localization. In this paper, we apply several famous FCNN to brain tumor segmentation, making comparisons and adjusting network architectures to achieve better performance measured by metrics such as precision, recall, mean of intersection of union (mIoU) and dice score coefficient (DSC). The adjustments to the classic FCNN include adding more connections between convolutional layers, enlarging decoders after up sample layers and changing the way shallower layers' information is reused. Besides the structure modification, we also propose a new classifier with a hierarchical dice loss. Inspired by the containing relationship between classes, the loss function converts multiple classification to multiple binary classification in order to counteract the negative effect caused by imbalance data set. Massive experiments have been done on the training set and testing set in order to assess our refined fully convolutional neural networks and new types of loss function. Competitive figures prove they are more effective than their predecessors.

cs.CV↗

arXiv2018

High Performance Depletion/Enhancement-Mode beta-Ga2O3 on Insulator (GOOI) Field-effect Transistors with Record Drain Currents of 600/450 mA/mm

In this letter, we report on high performance depletion/enhancement (D/E)-mode beta-Ga2O3 on insulator (GOOI) field-effect transistors (FETs) with record high drain currents (ID) of 600/450 mA/mm, which are nearly one order of magnitude higher than any other reported ID values. The threshold voltage (VT) can be modulated by varying the thickness of the beta-Ga2O3 films and the E-mode GOOI FET can be simply achieved by shrinking the beta-Ga2O3 film thickness. Benefiting from the good interface between beta-Ga2O3 and SiO2 and wide bandgap of beta-Ga2O3, a negligible transfer characteristic hysteresis, high ID on/off ratio of 10^10, and low subthreshold swing of 140 mV/dec for a 300 nm thick SiO2 are observed. E-mode GOOI FET with source to drain spacing of 0.9 um demonstrates a breakdown voltage of 185 V and an average electric field (E) of 2 MV/cm, showing the great promise of GOOI FET for future power devices.

cond-mat.mes-hall↗

arXiv2018

Steep-slope Hysteresis-free Negative Capacitance MoS2 Transistors

The so-called Boltzmann Tyranny defines the fundamental thermionic limit of the subthreshold slope (SS) of a metal-oxide-semiconductor field-effect transistor (MOSFET) at 60 mV/dec at room temperature and, therefore, precludes the lowering of the supply voltage and the overall power consumption. Adding a ferroelectric negative capacitor to the gate stack of a MOSFET may offer a promising solution to bypassing this fundamental barrier. Meanwhile, two-dimensional (2D) semiconductors, such as atomically thin transition metal dichalcogenides (TMDs) due to their low dielectric constant, and ease of integration in a junctionless transistor topology, offer enhanced electrostatic control of the channel. Here, we combine these two advantages and demonstrate for the first time a molybdenum disulfide (MoS2) 2D steep slope transistor with a ferroelectric hafnium zirconium oxide layer (HZO) in the gate dielectric stack. This device exhibits excellent performance in both on- and off-states, with maximum drain current of 510 μA/μm, sub-thermionic subthreshold slope and is essentially hysteresis-free. Negative differential resistance (NDR) was observed at room temperature in the MoS2 negative capacitance field-effect-transistors (NC-FETs) as the result of negative capacitance due to the negative drain-induced-barrier-lowering (DIBL). High on-current induced self-heating effect was also observed and studied.

cond-mat.mes-hall↗

arXiv2017

An Automatic Diagnosis Method of Facial Acne Vulgaris Based on Convolutional Neural Network

In this paper, we present a new automatic diagnosis method of facial acne vulgaris based on convolutional neural network. This method is proposed to overcome the shortcoming of classification types in previous methods. The core of our method is to extract features of images based on convolutional neural network and achieve classification by classifier. We design a binary classifier of skin-and-non-skin to detect skin area and a seven-classifier to achieve the classification of facial acne vulgaris and healthy skin. In the experiment, we compared the effectiveness of our convolutional neural network and the pre-trained VGG16 neural network on the ImageNet dataset. And we use the ROC curve and normal confusion matrix to evaluate the performance of the binary classifier and the seven-classifier. The results of our experiment show that the pre-trained VGG16 neural network is more effective in extracting image features. The classifiers based on the pre-trained VGG16 neural network achieve the skin detection and acne classification and have good robustness.

cs.CV↗

arXiv2017

Thermodynamic Studies of \b{eta}-Ga2O3 Nanomembrane Field-Effect Transistors on a Sapphire Substrate

The self-heating effect is a severe issue for high-power semiconductor devices, which degrades the electron mobility and saturation velocity, and also affects the device reliability. On applying an ultrafast and high-resolution thermoreflectance imaging technique, the direct self-heating effect and surface temperature increase phenomenon are observed on novel top-gate \b{eta}-Ga2O3 on insulator field-effect transistors. Here, we demonstrate that by utilizing a higher thermal conductivity sapphire substrate rather than a SiO2/Si substrate, the temperature rise above room temperature of \b{eta}-Ga2O3 on the insulator field-effect transistor can be reduced by a factor of 3 and thereby the self-heating effect is significantly reduced. Both thermoreflectance characterization and simulation verify that the thermal resistance on the sapphire substrate is less than 1/3 of that on the SiO2/Si substrate. Therefore, maximum drain current density of 535 mA/mm is achieved on the sapphire substrate, which is 70% higher than that on the SiO2/Si substrate due to reduced self-heating. Integration of \b{eta}-Ga2O3 channel on a higher thermal conductivity substrate opens a new route to address the low thermal conductivity issue of \b{eta}-Ga2O3 for power electronics applications.

physics.app-ph↗

arXiv2017

\b{eta}-Ga2O3 on Insulator Field-effect Transistors with Drain Currents Exceeding 1.5 A/mm and Their Self-heating Effect

We have demonstrated that depletion/enhancement-mode b-Ga2O3 on insulator field-effect transistors can achieve a record high drain current density of 1.5/1.0 A/mm by utilizing a highly doped b-Ga2O3 nano-membrane as the channel. b-Ga2O3 on insulator field-effect transistor (GOOI FET) shows a high on/off ratio of 1010 and low subthreshold slope of 150 mV/dec even with 300 nm thick SiO2. The enhancement-mode GOOI FET is achieved through surface depletion. An ultra-fast, high resolution thermo-reflectance imaging technique is applied to study the self-heating effect by directly measuring the local surface temperature. High drain current, low Rc, and wide bandgap make the b-Ga2O3 on insulator field-effect transistor a promising candidate for future power electronics applications.

cond-mat.mes-hall↗

arXiv2017

Controlled Growth of a Large-Size 2D Selenium Nanosheet and Its Electronic and Optoelectronic Applications

Selenium has attracted intensive attention as a promising material candidate for future optoelectronic applications. However, selenium has a strong tendency to grow into nanowire forms due to its anisotropic atomic structure, which has largely hindered the exploration of its potential applications. In this work, using a physical vapor deposition method, we have demonstrated the synthesis of large-size, high-quality 2D selenium nanosheets, the minimum thickness of which could be as thin as 5 nm. The Se nanosheet exhibits a strong in-plane anisotropic property, which is determined by angle-resolved Raman spectroscopy. Back-gating field-effect transistors based on a Se nanosheet exhibit p-type transport behaviors with on-state current density around 20 mA/mm at Vds = 3 V. Four-terminal field effect devices are also fabricated to evaluate the intrinsic hole mobility of the selenium nanosheet, and the value is determined to be 0.26 cm2 Vs at 300 K. The selenium nanosheet phototransistors show an excellent photoresponsivity of up to 263 A/W, with a rise time of 0.1 s and fall time of 0.12 s. These results suggest that crystal selenium as a 2D form of a 1D van der Waals solid opens up the possibility to explore device applications.

cond-mat.mtrl-sci↗

arXiv2017

$β$-Ga$_2$O$_3$ Nano-membrane Negative Capacitance Field-effect Transistor with Steep Subthreshold Slope for Wide Bandgap Logic Applications

Steep-slope $β$-Ga$_2$O$_3$ nano-membrane negative capacitance field-effect transistors (NC-FETs) are demonstrated with ferroelectric hafnium zirconium oxide in gate dielectric stack. Subthreshold slope less than 60 mV/dec at room temperature is obtained for both forward and reverse gate voltage sweeps with a minimum value of 34.3 mV/dec at reverse gate voltage sweep and 53.1 mV/dec at forward gate voltage sweep at $V_{DS}$=0.5 V. Enhancement-mode operation with threshold voltage ~0.4 V is achieved by tuning the thickness of $β$-Ga$_2$O$_3$ membrane. Low hysteresis of less than 0.1 V is obtained. The steep-slope, low hysteresis and enhancement-mode $β$-Ga$_2$O$_3$ NC-FETs are promising as nFET candidate for future wide bandgap CMOS logic applications.

cond-mat.mtrl-sci↗