arXiv subjects

Fan Zhang

Publications and source records attributed to Fan Zhang.

At least 523 records · Page 29Linked to original sources

arXiv2024

Speech-driven Personalized Gesture Synthetics: Harnessing Automatic Fuzzy Feature Inference

Speech-driven gesture generation is an emerging field within virtual human creation. However, a significant challenge lies in accurately determining and processing the multitude of input features (such as acoustic, semantic, emotional, personality, and even subtle unknown features). Traditional approaches, reliant on various explicit feature inputs and complex multimodal processing, constrain the expressiveness of resulting gestures and limit their applicability. To address these challenges, we present Persona-Gestor, a novel end-to-end generative model designed to generate highly personalized 3D full-body gestures solely relying on raw speech audio. The model combines a fuzzy feature extractor and a non-autoregressive Adaptive Layer Normalization (AdaLN) transformer diffusion architecture. The fuzzy feature extractor harnesses a fuzzy inference strategy that automatically infers implicit, continuous fuzzy features. These fuzzy features, represented as a unified latent feature, are fed into the AdaLN transformer. The AdaLN transformer introduces a conditional mechanism that applies a uniform function across all tokens, thereby effectively modeling the correlation between the fuzzy features and the gesture sequence. This module ensures a high level of gesture-speech synchronization while preserving naturalness. Finally, we employ the diffusion model to train and infer various gestures. Extensive subjective and objective evaluations on the Trinity, ZEGGS, and BEAT datasets confirm our model's superior performance to the current state-of-the-art approaches. Persona-Gestor improves the system's usability and generalization capabilities, setting a new benchmark in speech-driven gesture synthesis and broadening the horizon for virtual human technology. Supplementary videos and code can be accessed at https://zf223669.github.io/Diffmotion-v2-website/

cs.SD↗

arXiv2024

Gradient-Aware Logit Adjustment Loss for Long-tailed Classifier

In the real-world setting, data often follows a long-tailed distribution, where head classes contain significantly more training samples than tail classes. Consequently, models trained on such data tend to be biased toward head classes. The medium of this bias is imbalanced gradients, which include not only the ratio of scale between positive and negative gradients but also imbalanced gradients from different negative classes. Therefore, we propose the Gradient-Aware Logit Adjustment (GALA) loss, which adjusts the logits based on accumulated gradients to balance the optimization process. Additionally, We find that most of the solutions to long-tailed problems are still biased towards head classes in the end, and we propose a simple and post hoc prediction re-balancing strategy to further mitigate the basis toward head class. Extensive experiments are conducted on multiple popular long-tailed recognition benchmark datasets to evaluate the effectiveness of these two designs. Our approach achieves top-1 accuracy of 48.5\%, 41.4\%, and 73.3\% on CIFAR100-LT, Places-LT, and iNaturalist, outperforming the state-of-the-art method GCL by a significant margin of 3.62\%, 0.76\% and 1.2\%, respectively. Code is available at https://github.com/lt-project-repository/lt-project.

cs.CV↗

arXiv2024

Towards a Consistent Calculation of the Lunar Response to Gravitational Waves

The recent increasing interest in detecting gravitational waves (GWs) by lunar seismic measurement urges us to have a clear understanding of the response of the moon to passing GWs. In this paper, we clarify the relationship between two seemly different response functions which have been derived previously using two different methods, one taking the field-theory approach and the other using the tidal force induced by GWs. We revisit their derivation and prove, by both analytical arguments and numerical calculations, that the two response functions are equivalent. Their apparent difference can be attributed to the choice of different coordinates. Using the correct response function, we calculate the sensitivities (to GWs) of several designed lunar seismometers, and find that the sensitivity curves between $10^{-3}$ and $0.1$ Hz are much flatter than the previous calculations based on normal-mode model. Our results will help clarifying the scientific objectives of lunar GW observation, as well as provide important constraints on the design of lunar GW detectors.

gr-qc↗

arXiv2024

Miniature narrow-linewidth 1 μm Laser

Self-injection locking scheme has the potential to narrow the linewidth of lasers in a compact setup. Here, we report a narrow linewidth laser source near 1 μm by self-injection locking scheme using a Fabry-Perot (FP) hollow resonator with a high-quality factor (Q>10^8). The measured fundamental linewidth of the laser is 41 Hz, and a coarse tuning range over 5.5 nm is achieved by changing the driving current of the laser source. Meanwhile, a fine-tuning range of 373 MHz is achieved without mode hops by changing the voltage applied to the PZT on the resonator. More importantly, benefiting from the low thermal refractive noise and low thermal expansion of the FP hollow resonator, the beat-note linewidth and the frequency Allan deviation are measured to be 510.3 Hz in and 10^-11 (1s averaging time), respectively, by using a fully stabilized frequency comb as reference. Such a high-performance laser is fully integrated with a palm-sized package (52.3 mL) for field-deployable applications.

physics.optics↗

arXiv2024

Chlorine and zinc co-doping effects on the electronic structure and optical properties of γ-CuI

The effects of chlorine (Cl) and zinc (Zn) co-doping on the electronic structure and optical properties of the zinc blende (γ) phase of copper iodide (γ-CuI) scintillator material are investigated by using first-principles density functional theory calculations. The band structure, density of states, dielectric function, absorption coefficients, and reflectivity were analyzed before and after doping. Results show co-doping significantly modifies the band structure, reduces the band gap, and generates impurity energy levels. Cl doping enhances absorption in the high energy region while reducing visible light absorption. Zn doping induces a redshift in absorption and n-type conductivity at high concentrations. With suitable co-doping ratios, the absorption coefficient and reflectivity of γ-CuI can be optimized in the visible range to improve scintillation light yield. The calculations provide guidance for co-doping γ-CuI scintillators to achieve superior detection performance. The n-type conductivity also makes doped γ-CuI promising for optoelectronic applications.

cond-mat.mtrl-sci↗

arXiv2024

BVI-Artefact: An Artefact Detection Benchmark Dataset for Streamed Videos

Professionally generated content (PGC) streamed online can contain visual artefacts that degrade the quality of user experience. These artefacts arise from different stages of the streaming pipeline, including acquisition, post-production, compression, and transmission. To better guide streaming experience enhancement, it is important to detect specific artefacts at the user end in the absence of a pristine reference. In this work, we address the lack of a comprehensive benchmark for artefact detection within streamed PGC, via the creation and validation of a large database, BVI-Artefact. Considering the ten most relevant artefact types encountered in video streaming, we collected and generated 480 video sequences, each containing various artefacts with associated binary artefact labels. Based on this new database, existing artefact detection methods are benchmarked, with results showing the challenging nature of this tasks and indicating the requirement of more reliable artefact detection methods. To facilitate further research in this area, we have made BVI-Artifact publicly available at https://chenfeng-bristol.github.io/BVI-Artefact/

cs.CV↗

arXiv2024

RankDVQA-mini: Knowledge Distillation-Driven Deep Video Quality Assessment

Deep learning-based video quality assessment (deep VQA) has demonstrated significant potential in surpassing conventional metrics, with promising improvements in terms of correlation with human perception. However, the practical deployment of such deep VQA models is often limited due to their high computational complexity and large memory requirements. To address this issue, we aim to significantly reduce the model size and runtime of one of the state-of-the-art deep VQA methods, RankDVQA, by employing a two-phase workflow that integrates pruning-driven model compression with multi-level knowledge distillation. The resulting lightweight full reference quality metric, RankDVQA-mini, requires less than 10% of the model parameters compared to its full version (14% in terms of FLOPs), while still retaining a quality prediction performance that is superior to most existing deep VQA methods. The source code of the RankDVQA-mini has been released at https://chenfeng-bristol.github.io/RankDVQA-mini/ for public evaluation.

eess.IV↗

arXiv2024

Giant Tunability of Intersubband Transitions and Quantum Hall Quartets in Few-Layer InSe Quantum Wells

A two-dimensional (2D) quantum electron system is characterized by the quantized energy levels, or subbands, in the out-of-plane direction. Populating higher subbands and controlling the inter-subband transitions have wide technological applications such as optical modulators and quantum cascade lasers. In conventional materials, however, the tunability of intersubband spacing is limited. Here we demonstrate electrostatic population and characterization of the second subband in few-layer InSe quantum wells, with giant tunability of its energy, population, and spin-orbit coupling strength, via the control of not only layer thickness but also out-of-plane displacement field. A modulation of as much as 350% or over 250 meV is achievable, underscoring the promise of InSe for tunable infrared and THz sources, detectors and modulators.

cond-mat.mes-hall↗

arXiv2024

Immersive Video Compression using Implicit Neural Representations

Recent work on implicit neural representations (INRs) has evidenced their potential for efficiently representing and encoding conventional video content. In this paper we, for the first time, extend their application to immersive (multi-view) videos, by proposing MV-HiNeRV, a new INR-based immersive video codec. MV-HiNeRV is an enhanced version of a state-of-the-art INR-based video codec, HiNeRV, which was developed for single-view video compression. We have modified the model to learn a different group of feature grids for each view, and share the learnt network parameters among all views. This enables the model to effectively exploit the spatio-temporal and the inter-view redundancy that exists within multi-view videos. The proposed codec was used to compress multi-view texture and depth video sequences in the MPEG Immersive Video (MIV) Common Test Conditions, and tested against the MIV Test model (TMIV) that uses the VVenC video codec. The results demonstrate the superior performance of MV-HiNeRV, with significant coding gains (up to 72.33\%) over TMIV. The implementation of MV-HiNeRV is published for further development and evaluation.

eess.IV↗

arXiv2024

Compressing Deep Image Super-resolution Models

Deep learning techniques have been applied in the context of image super-resolution (SR), achieving remarkable advances in terms of reconstruction performance. Existing techniques typically employ highly complex model structures which result in large model sizes and slow inference speeds. This often leads to high energy consumption and restricts their adoption for practical applications. To address this issue, this work employs a three-stage workflow for compressing deep SR models which significantly reduces their memory requirement. Restoration performance has been maintained through teacher-student knowledge distillation using a newly designed distillation loss. We have applied this approach to two popular image super-resolution networks, SwinIR and EDSR, to demonstrate its effectiveness. The resulting compact models, SwinIRmini and EDSRmini, attain an 89% and 96% reduction in both model size and floating-point operations (FLOPs) respectively, compared to their original versions. They also retain competitive super-resolution performance compared to their original models and other commonly used SR approaches. The source code and pre-trained models for these two lightweight SR approaches are released at https://pikapi22.github.io/CDISM/.

eess.IV↗

arXiv2024

Solving the Einstein Constraints Numerically on Compact Three-Manifolds Using Hyperbolic Relaxation

The effectiveness of the hyperbolic relaxation method for solving the Einstein constraint equations numerically is studied here on a variety of compact orientable three-manifolds. Convergent numerical solutions are found using this method on manifolds admitting negative Ricci scalar curvature metrics, i.e. those from the $H^3$ and the $H^2\times S^1$ geometrization classes. The method fails to produce solutions, however, on all the manifolds examined here admitting non-negative Ricci scalar curvatures, i.e. those from the $S^3$, $S^2\times S^1$, and the $E^3$ classes. This study also finds that the accuracy of the convergent solutions produced by hyperbolic relaxation can be increased significantly by performing fairly low-cost standard elliptic solves using the hyperbolic relaxation solutions as initial guesses.

gr-qc↗

arXiv2024

Magnetic-field-free nonreciprocal transport in graphene multi-terminal Josephson junctions

Nonreciprocal superconducting devices have attracted growing interest in recent years as they potentially enable directional charge transport for applications in superconducting quantum circuits. Specifically, the superconducting diode effect has been explored in two-terminal devices that exhibit superconducting transport in one current direction while showing dissipative transport in the opposite direction. Here, we exploit multi-terminal Josephson junctions (MTJJs) to engineer magnetic-field-free nonreciprocity in multi-port networks. We show that when treated as a two-port electrical network, a three-terminal Josephson junction (JJ) with an asymmetric graphene region exhibits reconfigurable two-port nonreciprocity. We observe nonreciprocal (reciprocal) transport between superconducting terminals with broken (preserved) spatial mirror symmetry. We explain our observations by considering a circuit-network of JJs with different critical currents.

cond-mat.supr-con↗

arXiv2024

EVA-CLIP-18B: Scaling CLIP to 18 Billion Parameters

Scaling up contrastive language-image pretraining (CLIP) is critical for empowering both vision and multimodal models. We present EVA-CLIP-18B, the largest and most powerful open-source CLIP model to date, with 18-billion parameters. With only 6-billion training samples seen, EVA-CLIP-18B achieves an exceptional 80.7% zero-shot top-1 accuracy averaged across 27 widely recognized image classification benchmarks, outperforming its forerunner EVA-CLIP (5-billion parameters) and other open-source CLIP models by a large margin. Remarkably, we observe a consistent performance improvement with the model size scaling of EVA-CLIP, despite maintaining a constant training dataset of 2-billion image-text pairs from LAION-2B and COYO-700M. This dataset is openly available and much smaller than the in-house datasets (e.g., DFN-5B, WebLI-10B) employed in other state-of-the-art CLIP models. EVA-CLIP-18B demonstrates the potential of EVA-style weak-to-strong visual model scaling. With our model weights made publicly available, we hope to facilitate future research in vision and multimodal foundation models.

cs.CV↗

arXiv2024

HiNeRV: Video Compression with Hierarchical Encoding-based Neural Representation

Learning-based video compression is currently a popular research topic, offering the potential to compete with conventional standard video codecs. In this context, Implicit Neural Representations (INRs) have previously been used to represent and compress image and video content, demonstrating relatively high decoding speed compared to other methods. However, existing INR-based methods have failed to deliver rate quality performance comparable with the state of the art in video compression. This is mainly due to the simplicity of the employed network architectures, which limit their representation capability. In this paper, we propose HiNeRV, an INR that combines light weight layers with novel hierarchical positional encodings. We employs depth-wise convolutional, MLP and interpolation layers to build the deep and wide network architecture with high capacity. HiNeRV is also a unified representation encoding videos in both frames and patches at the same time, which offers higher performance and flexibility than existing methods. We further build a video codec based on HiNeRV and a refined pipeline for training, pruning and quantization that can better preserve HiNeRV's performance during lossy model compression. The proposed method has been evaluated on both UVG and MCL-JCV datasets for video compression, demonstrating significant improvement over all existing INRs baselines and competitive performance when compared to learning-based codecs (72.3% overall bit rate saving over HNeRV and 43.4% over DCVC on the UVG dataset, measured in PSNR).

eess.IV↗

arXiv2024

Observation of GRB 221009A early afterglow in X/$γ$-ray energy band

The early afterglow of a Gamma-ray burst (GRB) can provide critical information on the jet and progenitor of the GRB. The extreme brightness of GRB 221009A allows us to probe its early afterglow in unprecedented detail. In this letter, we report comprehensive observation results of the early afterglow of GRB 221009A (from $T_0$+660 s to $T_0$+1860 s, where $T_0$ is the \textit{Insight}-HXMT/HE trigger time) in X/$γ$-ray energy band (from 20 keV to 20 MeV) by \textit{Insight}-HXMT/HE, GECAM-C and \textit{Fermi}/GBM. We find that the spectrum of the early afterglow in 20 keV-20 MeV could be well described by a cutoff power-law with an extra power-law which dominates the low and high energy bands respectively. The cutoff power-law $E_{\rm peak}$ is $\sim$ 30 keV and the power-law photon index is $\sim$ 1.8 throughout the early afterglow phase. By fitting the light curves in different energy bands, we find that a significant achromatic break (from keV to TeV) is required at $T_0$ + 1246$^{+27}_{-26}$ s (i.e. 1021 s since the afterglow starting time $T_{\rm AG}$=$T_0$+225 s), providing compelling evidence of a jet break. Interestingly, both the pre-break and post-break decay slopes vary with energy, and these two slopes become closer in the lower energy band, making the break less identifiable. Intriguingly, the spectrum of the early afterglow experienced a slight hardening before the break and a softening after the break. These results provide new insights into the understanding of this remarkable GRB.

astro-ph.HE↗

arXiv2024

Phase diagram of a square lattice model of XY Spins with direction-dependent interactions

We study a generalization of the well-known classical two-dimensional square lattice compass model of XY spins (sometimes referred to as the 90$^\circ$ compass model), which interpolates between the XY model and the compass model. Our model possesses the combined $C_4$ lattice and spin rotation symmetry of the compass model but is free of its fine-tuned subsystem symmetries. Using both field theoretic arguments and Monte Carlo simulations, we find that our model possesses a line of critical points with continuously varying exponents of the Ashkin-Teller type terminating at the four-state Potts point. Further, our Monte Carlo study uncovers that beyond the four-state Potts point, the line of phase transition is connected to the lattice-nematic Ising phase transition in the square lattice compass model through a region of first-order transitions.

cond-mat.stat-mech↗

arXiv2024

MIMIC: Mask Image Pre-training with Mix Contrastive Fine-tuning for Facial Expression Recognition

Cutting-edge research in facial expression recognition (FER) currently favors the utilization of convolutional neural networks (CNNs) backbone which is supervisedly pre-trained on face recognition datasets for feature extraction. However, due to the vast scale of face recognition datasets and the high cost associated with collecting facial labels, this pre-training paradigm incurs significant expenses. Towards this end, we propose to pre-train vision Transformers (ViTs) through a self-supervised approach on a mid-scale general image dataset. In addition, when compared with the domain disparity existing between face datasets and FER datasets, the divergence between general datasets and FER datasets is more pronounced. Therefore, we propose a contrastive fine-tuning approach to effectively mitigate this domain disparity. Specifically, we introduce a novel FER training paradigm named Mask Image pre-training with MIx Contrastive fine-tuning (MIMIC). In the initial phase, we pre-train the ViT via masked image reconstruction on general images. Subsequently, in the fine-tuning stage, we introduce a mix-supervised contrastive learning process, which enhances the model with a more extensive range of positive samples by the mixing strategy. Through extensive experiments conducted on three benchmark datasets, we demonstrate that our MIMIC outperforms the previous training paradigm, showing its capability to learn better representations. Remarkably, the results indicate that the vanilla ViT can achieve impressive performance without the need for intricate, auxiliary-designed modules. Moreover, when scaling up the model size, MIMIC exhibits no performance saturation and is superior to the current state-of-the-art methods.

cs.CV↗

arXiv2024

3D Landmark Detection on Human Point Clouds: A Benchmark and A Dual Cascade Point Transformer Framework

3D landmark detection plays a pivotal role in various applications such as 3D registration, pose estimation, and virtual try-on. While considerable success has been achieved in 2D human landmark detection or pose estimation, there is a notable scarcity of reported works on landmark detection in unordered 3D point clouds. This paper introduces a novel challenge, namely 3D landmark detection on human point clouds, presenting two primary contributions. Firstly, we establish a comprehensive human point cloud dataset, named HPoint103, designed to support the 3D landmark detection community. This dataset comprises 103 human point clouds created with commercial software and actors, each manually annotated with 11 stable landmarks. Secondly, we propose a Dual Cascade Point Transformer (D-CPT) model for precise point-based landmark detection. D-CPT gradually refines the landmarks through cascade Transformer decoder layers across the entire point cloud stream, simultaneously enhancing landmark coordinates with a RefineNet over local regions. Comparative evaluations with popular point-based methods on HPoint103 and the public dataset DHP19 demonstrate the dramatic outperformance of our D-CPT. Additionally, the integration of our RefineNet into existing methods consistently improves performance.

cs.CV↗