arXiv subjects

Wenqing Zhang

Publications and source records attributed to Wenqing Zhang.

At least 55 records · Page 3Linked to original sources

arXiv2022

I2C2W: Image-to-Character-to-Word Transformers for Accurate Scene Text Recognition

Leveraging the advances of natural language processing, most recent scene text recognizers adopt an encoder-decoder architecture where text images are first converted to representative features and then a sequence of characters via `sequential decoding'. However, scene text images suffer from rich noises of different sources such as complex background and geometric distortions which often confuse the decoder and lead to incorrect alignment of visual features at noisy decoding time steps. This paper presents I2C2W, a novel scene text recognition technique that is tolerant to geometric and photometric degradation by decomposing scene text recognition into two inter-connected tasks. The first task focuses on image-to-character (I2C) mapping which detects a set of character candidates from images based on different alignments of visual features in an non-sequential way. The second task tackles character-to-word (C2W) mapping which recognizes scene text by decoding words from the detected character candidates. The direct learning from character semantics (instead of noisy image features) corrects falsely detected character candidates effectively which improves the final text recognition accuracy greatly. Extensive experiments over nine public datasets show that the proposed I2C2W outperforms the state-of-the-art by large margins for challenging scene text datasets with various curvature and perspective distortions. It also achieves very competitive recognition performance over multiple normal scene text datasets.

cs.CV↗

arXiv2022

FIS-GAN: GAN with Flow-based Importance Sampling

Generative Adversarial Networks (GAN) training process, in most cases, apply Uniform or Gaussian sampling methods in the latent space, which probably spends most of the computation on examples that can be properly handled and easy to generate. Theoretically, importance sampling speeds up stochastic optimization in supervised learning by prioritizing training examples. In this paper, we explore the possibility of adapting importance sampling into adversarial learning. We use importance sampling to replace Uniform and Gaussian sampling methods in the latent space and employ normalizing flow to approximate latent space posterior distribution by density estimation. Empirically, results on MNIST and Fashion-MNIST demonstrate that our method significantly accelerates GAN's optimization while retaining visual fidelity in generated samples.

cs.LG↗

arXiv2022

Discovery of a metallic oxide with ultralow thermal conductivity

A compound with metallic electrical conductivity usually has a considerable total thermal conductivity because both electrons and photons contribute to thermal transport. Here, we show an exceptional example of iridium oxide, Bi3Ir3O11, that concurrently displays metallic electrical conductivity and ultralow thermal conductivity approaching 0.61 W m-1 K-1 at 300 K. The compound crystallizes into a cubic structural framework with space group Pn-3. The edge- and corner-sharing IrO6 octahedra with a mixed Ir4.33+ charge state favor metallic electrical transport. Bi3Ir3O11 exhibits an extremely low lattice thermal conductivity close to the minimum limit in theory owing to its tunnel-like structure with filled heavy atoms Bi rattling inside. Theoretical calculations reveal the underlying mechanisms for the extraordinary compatibility between metallic electrical conductivity and ultralow thermal conductivity. This study may establish a new avenue for designing and developing unprecedented heat-insulation metals.

cond-mat.mtrl-sci↗

arXiv2022

A topological transition-induced giant transverse thermoelectric effect in polycrystalline Dirac semimetal Mg3Bi2

To achieve thermoelectric energy conversion, a large transverse thermoelectric effect in topological materials is crucial. However, the general relationship between topological electronic structures and transverse thermoelectric effect remains unclear, restricting the rational design of novel transverse thermoelectric materials. Herein, we demonstrate a topological transition-induced giant transverse thermoelectric effect in polycrystalline Mn-doped Mg3+δBi2 material, which has a competitively large transverse thermopower (617 uV/K), power factor (20393 uWm-1K-2), magnetoresistance (16600%), and electronic mobility (35280cm2V-1S-1). The high performance is triggered by the modulation of chemical pressure and disorder effects in the presence of Mn doping, which induces the transition from a topological insulator to a Dirac semimetal. The high-performance polycrystalline Mn-doped Mg3+δ Bi2 described in this work robustly boosts transverse thermoelectric effect through topological phase transition, paving a new avenue for the material design of transverse thermoelectricity.

cond-mat.mtrl-sci↗

arXiv2022

Language Matters: A Weakly Supervised Vision-Language Pre-training Approach for Scene Text Detection and Spotting

Recently, Vision-Language Pre-training (VLP) techniques have greatly benefited various vision-language tasks by jointly learning visual and textual representations, which intuitively helps in Optical Character Recognition (OCR) tasks due to the rich visual and textual information in scene text images. However, these methods cannot well cope with OCR tasks because of the difficulty in both instance-level text encoding and image-text pair acquisition (i.e. images and captured texts in them). This paper presents a weakly supervised pre-training method, oCLIP, which can acquire effective scene text representations by jointly learning and aligning visual and textual information. Our network consists of an image encoder and a character-aware text encoder that extract visual and textual features, respectively, as well as a visual-textual decoder that models the interaction among textual and visual features for learning effective scene text representations. With the learning of textual features, the pre-trained model can attend texts in images well with character awareness. Besides, these designs enable the learning from weakly annotated texts (i.e. partial texts in images without text bounding boxes) which mitigates the data annotation constraint greatly. Experiments over the weakly annotated images in ICDAR2019-LSVT show that our pre-trained model improves F-score by +2.5\% and +4.8\% while transferring its weights to other text detection and spotting networks, respectively. In addition, the proposed method outperforms existing pre-training techniques consistently across multiple public datasets (e.g., +3.2\% and +1.3\% for Total-Text and CTW1500).

cs.CV↗

arXiv2022

Data-based Polymer-Unit Fingerprint (PUFp): A Newly Accessible Expression of Polymer Organic Semiconductors for Machine Learning

In the process of finding high-performance organic semiconductors (OSCs), it is of paramount importance in material development to identify important functional units that play key roles in material performance and subsequently establish substructure-property relationships. Herein, we describe a polymer-unit fingerprint (PUFp) generation framework. Machine learning (ML) models can be used to determine structure-mobility relationships by using PUFp information as structural input with 678 pieces of collected OSC data. A polymer-unit library consisting of 445 units is constructed, and the key polymer units for the mobility of OSCs are identified. By investigating the combinations of polymer units with mobility performance, a scheme for designing polymer OSC materials by combining ML approaches and PUFp information is proposed to not only passively predict OSC mobility but also actively provide structural guidance for new high-mobility OSC material design. The proposed scheme demonstrates the ability to screen new materials through pre-evaluation and classification ML steps and is an alternative methodology for applying ML in new high-mobility OSC discovery.

cond-mat.mtrl-sci↗

arXiv2022

High-performance non-Fermi-liquid metallic thermoelectric materials

Searching for high-performance thermoelectric (TE) materials in the paradigm of narrow-bandgap semiconductors has lasted for nearly 70 years and is obviously hampered by a bottleneck of research now. Here we report on the discovery of a few metallic compounds, TiFexCu2x-1Sb and TiFe1.33Sb, showing the thermopower exceeding many TE semiconductors and the dimensionless figure of merits comparable with the state-of-the-art TE materials. A quasi-linear temperature (T) dependence of electrical resistivity in 2 K - 700 K and the logarithmic T-dependent electronic specific heat at low temperature are also observed to coexist with the high thermopower, highlighting the strong intercoupling of the non-Fermi-liquid (NFL) quantum critical behavior of electrons with TE transports. Electronic structure analysis reveals the existence of fluctuating Fe-eg-related local magnetic moments, Fe-Fe antiferromagnetic (AFM) interaction at the nearest 4c-4d sites, and two-fold degenerate eg orbitals antiferromagnetically coupled with the dual-type itinerant electrons close to the Fermi level, all of which infer to a competition between the AFM ordering and Kondo-like spin compensation as well as a parallel two-channel Kondo effect. These effects are both strongly meditated by the structural disorder due to the random filling of Fe/Cu at the equivalent 4c/4d sites of the Heusler crystal lattice. The magnetic susceptibility deviates from ideal antiferromagnetism but can be fitted well by x(T) = 1/(θ + BTα), seemingly being consistent with the quantum critical scenario of strong local correlation as discussed before. Our work not only breaks the dilemma that the promising TE materials should be heavily-doped semiconductors, but also demonstrates the correlation among high TE performance, NFL quantum criticality, and magnetic fluctuation, which opens up new directions for future research.

cond-mat.str-el↗

arXiv2022

1st Place Solution to ECCV 2022 Challenge on Out of Vocabulary Scene Text Understanding: End-to-End Recognition of Out of Vocabulary Words

Scene text recognition has attracted increasing interest in recent years due to its wide range of applications in multilingual translation, autonomous driving, etc. In this report, we describe our solution to the Out of Vocabulary Scene Text Understanding (OOV-ST) Challenge, which aims to extract out-of-vocabulary (OOV) words from natural scene images. Our oCLIP-based model achieves 28.59\% in h-mean which ranks 1st in end-to-end OOV word recognition track of OOV Challenge in ECCV2022 TiE Workshop.

cs.CV↗

arXiv2022

Runner-Up Solution to ECCV 2022 Challenge on Out of Vocabulary Scene Text Understanding: Cropped Word Recognition

This report presents our 2nd place solution to ECCV 2022 challenge on Out-of-Vocabulary Scene Text Understanding (OOV-ST) : Cropped Word Recognition. This challenge is held in the context of ECCV 2022 workshop on Text in Everything (TiE), which aims to extract out-of-vocabulary words from natural scene images. In the competition, we first pre-train SCATTER on the synthetic datasets, then fine-tune the model on the training set with data augmentations. Meanwhile, two additional models are trained specifically for long and vertical texts. Finally, we combine the output from different models with different layers, different backbones, and different seeds as the final results. Our solution achieves a word accuracy of 59.45\% when considering out-of-vocabulary words only.

cs.CV↗

arXiv2022

SeqFormer: Sequential Transformer for Video Instance Segmentation

In this work, we present SeqFormer for video instance segmentation. SeqFormer follows the principle of vision transformer that models instance relationships among video frames. Nevertheless, we observe that a stand-alone instance query suffices for capturing a time sequence of instances in a video, but attention mechanisms shall be done with each frame independently. To achieve this, SeqFormer locates an instance in each frame and aggregates temporal information to learn a powerful representation of a video-level instance, which is used to predict the mask sequences on each frame dynamically. Instance tracking is achieved naturally without tracking branches or post-processing. On YouTube-VIS, SeqFormer achieves 47.4 AP with a ResNet-50 backbone and 49.0 AP with a ResNet-101 backbone without bells and whistles. Such achievement significantly exceeds the previous state-of-the-art performance by 4.6 and 4.4, respectively. In addition, integrated with the recently-proposed Swin transformer, SeqFormer achieves a much higher AP of 59.3. We hope SeqFormer could be a strong baseline that fosters future research in video instance segmentation, and in the meantime, advances this field with a more robust, accurate, neat model. The code is available at https://github.com/wjf5203/SeqFormer.

cs.CV↗

arXiv2022

Solid-state Janus nanoprecipitation enables amorphous-like heat conduction in crystalline Mg3Sb2-based thermoelectric materials

Solid-state precipitation can be used to tailor materials properties, ranging from ferromagnets and catalysts to mechanical strengthening and energy storage. Thermoelectric properties can be modified by precipitation to enhance phonon scattering while retaining charge-carrier transmission. Here, we uncover unconventional dual Janus-type nanoprecipitates in Mg3Sb1.5Bi0.5 formed by side-by-side Bi- and Ge-rich appendages, in contrast to separate nanoprecipitate formation. These Janus nanoprecipitates result from local co-melting of Bi and Ge during sintering, enabling an amorphous-like lattice thermal conductivity. A precipitate size effect on phonon scattering is observed due to the balance between alloy-disorder and nanoprecipitate scattering. The thermoelectric figure-of-merit ZT reaches 0.6 near room temperature and 1.6 at 773 K. The Janus nanoprecipitation can be introduced into other materials and may act as a general property-tailoring mechanism.

cond-mat.mtrl-sci↗

arXiv2022

Prediction of protected band edge states and dielectric tunable quasiparticle and excitonic properties of monolayer MoSi$_2$N$_4$

The electronic structure of two-dimensional (2D) materials are inherently prone to environmental perturbations, which may pose significant challenges to their applications in electronic or optoelectronic devices. A 2D material couples with its environment through two mechanisms: local chemical coupling and nonlocal dielectric screening effects. The local chemical coupling is often difficult to predict or control experimentally. Nonlocal dielectric screening, on the other hand, can be tuned by choosing the substrates or layer thickness in a controllable manner. Therefore, a compelling 2D electronic material should offer band edge states that are robust against local chemical coupling effects. Here it is demonstrated that the recently synthesized MoSi$_2$N$_4$ is an ideal 2D semiconductor with robust band edge states protected from capricious environmental chemical coupling effects. Detailed many-body perturbation theory calculations are carried out to illustrate how the band edge states of MoSi$_2$N$_4$ are shielded from the direct chemical coupling effects, but its quasiparticle and excitonic properties can be modulated through the nonlocal dielectric screening effects. This unique property, together with the moderate band gap and the thermodynamic and mechanical stability of this material, paves the way for a range of applications of MoSi$_2$N$_4$ in areas including energy, 2D electronics, and optoelectronics.

cond-mat.mtrl-sci↗

arXiv2022

Few Could Be Better Than All: Feature Sampling and Grouping for Scene Text Detection

Recently, transformer-based methods have achieved promising progresses in object detection, as they can eliminate the post-processes like NMS and enrich the deep representations. However, these methods cannot well cope with scene text due to its extreme variance of scales and aspect ratios. In this paper, we present a simple yet effective transformer-based architecture for scene text detection. Different from previous approaches that learn robust deep representations of scene text in a holistic manner, our method performs scene text detection based on a few representative features, which avoids the disturbance by background and reduces the computational cost. Specifically, we first select a few representative features at all scales that are highly relevant to foreground text. Then, we adopt a transformer for modeling the relationship of the sampled features, which effectively divides them into reasonable groups. As each feature group corresponds to a text instance, its bounding box can be easily obtained without any post-processing operation. Using the basic feature pyramid network for feature extraction, our method consistently achieves state-of-the-art results on several popular datasets for scene text detection.

cs.CV↗

arXiv2022

Giant transverse and longitudinal magneto-thermoelectric effect in polycrystalline nodal-line semimetal Mg3Bi2

Topological semimetals provide new opportunities for exploring new thermoelectric phenomena, because of their exotic and nontrivial electronic structure topology around the Fermi surface. In this study, we report on the discovery of giant transverse and longitudinal magneto-thermoelectric (MTE) effects in Mg3Bi2, which is predicted to be a type-II nodal-line semimetal in the absence of spin-orbit coupling (SOC). The maximum transverse power factor is 2182 μWm^{-1}K^{-2} at 13.5 K and 6 Tesla. The longitudinal power factor reaches up to 3043μWm^{-1}K^{-2} at 15 K and 13 Tesla, which is 20 times higher than in a zero-strength magnetic field and is also comparable to state-of-the-art MTE materials. By compensating Mg loss in the Mg-rich conditions for turning carrier concentration, the sample obtained in this work shows a large linear non-saturating magnetoresistance of 940% under a field of 14 Tesla. This is a two-orders-of-magnitude increase with respect to the normal Mg-deficiency Mg3Bi2 sample. Using density functional calculations, we attribute the underlying mechanism to the parent nodal-line electronic structure without SOC and the anisotropic Fermi surface shape with SOC, highlighting the essential role of high carrier mobility and open electron orbits in moment space. Our work offers a new avenue toward highly efficient thermoelectric materials through the design of Fermi surfaces with special topological electronic structures in novel quantum materials.

cond-mat.mtrl-sci↗

arXiv2022

FedOCR: Communication-Efficient Federated Learning for Scene Text Recognition

While scene text recognition techniques have been widely used in commercial applications, data privacy has rarely been taken into account by this research community. Most existing algorithms have assumed a set of shared or centralized training data. However, in practice, data may be distributed on different local devices that can not be centralized to share due to the privacy restrictions. In this paper, we study how to make use of decentralized datasets for training a robust scene text recognizer while keeping them stay on local devices. To the best of our knowledge, we propose the first framework leveraging federated learning for scene text recognition, which is trained with decentralized datasets collaboratively. Hence we name it FedOCR. To make FedCOR fairly suitable to be deployed on end devices, we make two improvements including using lightweight models and hashing techniques. We argue that both are crucial for FedOCR in terms of the communication efficiency of federated learning. The simulations on decentralized datasets show that the proposed FedOCR achieves competitive results to the models that are trained with centralized data, with fewer communication costs and higher-level privacy-preserving.

cs.CV↗

arXiv2021

Leveraging Large-scale Computational Database and Deep Learning for Accurate Prediction of Material Properties

Accurately predicting the physical and chemical properties of materials remains one of the most challenging tasks in material design, and one effective strategy is to construct a reliable data set and use it for training a machine learning model. In this study, we constructed a large-scale material genome database (Matgen) containing 76,463 materials collected from experimentally-observed database, and computed their bandgap properties through the Density functional theory (DFT) method with Perdew-Burke-Ernzehof (PBE) functional. We verified the computation method by comparing part of our results with those from the open Material Project (MP) and Open Quantum Materials Database (OQMD), all with PBE computations, and found that Matgen achieved the same computation accuracy based on both measured and computed bandgap properties. Based on the computed properties of our comprehensive dataset, we have developed a new graph-based deep learning model, namely CrystalNet, through our recently developed Communicative Message Passing Neural Network (CMPNN) framework. The model was shown to outperform other state-of-the-art prediction models. A further fine-tuning on 1716 experimental bandgap values (CrystalNet-TL) achieved a superior performance with mean absolute error (MAE) of 0.77 eV on independent test, which has outperformed the pure PBE (1.14~1.45 eV). Moreover, the model was proven applicable to hypothetical materials with MAE of 0.77 eV as referred by computations from HSE, a highly accurate quantum mechanics (QM) method, consist better than PBE (MAE=1.13eV). We also made material structures, computed properties by PBE, and the CrystalNet models publically available at https://matgen.nscc-gz.cn.

cond-mat.mtrl-sci↗

arXiv2021

Advancing COVID-19 Diagnosis with Privacy-Preserving Collaboration in Artificial Intelligence

Artificial intelligence (AI) provides a promising substitution for streamlining COVID-19 diagnoses. However, concerns surrounding security and trustworthiness impede the collection of large-scale representative medical data, posing a considerable challenge for training a well-generalised model in clinical practices. To address this, we launch the Unified CT-COVID AI Diagnostic Initiative (UCADI), where the AI model can be distributedly trained and independently executed at each host institution under a federated learning framework (FL) without data sharing. Here we show that our FL model outperformed all the local models by a large yield (test sensitivity /specificity in China: 0.973/0.951, in the UK: 0.730/0.942), achieving comparable performance with a panel of professional radiologists. We further evaluated the model on the hold-out (collected from another two hospitals leaving out the FL) and heterogeneous (acquired with contrast materials) data, provided visual explanations for decisions made by the model, and analysed the trade-offs between the model performance and the communication costs in the federated training process. Our study is based on 9,573 chest computed tomography scans (CTs) from 3,336 patients collected from 23 hospitals located in China and the UK. Collectively, our work advanced the prospects of utilising federated learning for privacy-preserving AI in digital health.

cs.AI↗

arXiv2021

Tuning color centers at a twisted interface

Color center is a promising platform for quantum technologies, but their application is hindered by the typically random defect distribution and complex mesoscopic environment. Employing cathodoluminescence, we demonstrate that an ultraviolet-emitting single photon emitter can be readily activated and controlled on-demand at the twisted interface of two hexagonal boron nitride flakes. The brightness of the color center can be enhanced by two orders of magnitude by altering the twist angle. Additionally, a brightness modulation of nearly 100% of this color center is achieved by an external voltage. Our ab-initio GW calculations suggest that the emission is correlated to nitrogen vacancies and that a twist-induced moiré potential facilitates electron-hole recombination. This mechanism is further exploited to draw nanoscale color center patterns using electron beams.

cond-mat.mtrl-sci↗