arXiv subjects

Lili Wang

Publications and source records attributed to Lili Wang.

At least 37 records · Page 2Linked to original sources

arXiv2026

RFHNet: Relational and Frequency-Aware Hashing Network for Large-Scale Fine-Grained Food Image Retrieval

Fine-grained food image retrieval is a key task in computational gastronomy, with applications in food traceability, dietary monitoring, and smart catering systems. Although hashing-based retrieval is attractive for large-scale search due to its storage efficiency and fast Hamming-distance computation, existing methods often perform poorly in fine-grained food scenarios, where subtle local semantics and frequency-sensitive visual cues are essential. To address this challenge, we propose RFHNet, a cascaded hierarchical hashing network that captures both global structure and fine-grained local details through multi-level representations. RFHNet includes three components: (1) Fine-grained Relation Modeling (FRM) to capture subtle visual differences among similar food components; (2) Multi-Frequency Modulated Fusion (MFMF) to extract informative multi-frequency features; and (3) Hierarchical Semantic Synergy (HSS) to adaptively integrate multi-level representations and generate discriminative hash codes. Experiments on six food-specific benchmarks show that RFHNet consistently outperforms state-of-the-art hashing methods, with mAP gains of 4.44\% to 17.20\% at 12 bits. These results validate the effectiveness of RFHNet for large-scale visual food retrieval and smart catering applications. The source code will be released upon publication.

cs.CV↗

arXiv2026

Leaderless Collective Motion in Affine Formation Control over the Complex Plane

We propose a method for the collective maneuvering of affine formations in the plane by modifying the original weights of the Laplacian matrix used to achieve static formations of robot swarms. Specifically, the resulting collective motion is characterized as a time-varying affine transformation of a reference configuration, or shape. Unlike the traditional leader-follower strategy, our leaderless scheme allows agents to maintain distinct and possibly time-varying velocities, enabling a broader range of collective motions, including all the linear combinations of translations, rotations, scaling and shearing of a reference shape. Our analysis provides the analytic solution governing the resulting collective motion, explicitly designing the eigenvectors and eigenvalues that define this motion as a function of the modified weights in the new Laplacian matrix. To facilitate a more tractable analysis and design of affine formations in 2D, we propose the use of complex numbers to represent all relevant information. Simulations with up to 20 agents validate the theoretical results.

cs.RO↗

arXiv2026

LatentStealth: Unnoticeable and Efficient Adversarial Attacks on Expressive Human Pose and Shape Estimation

Expressive human pose and shape estimation (EHPS) plays a central role in digital human generation, particularly in live-streaming applications. However, most existing EHPS models focus primarily on minimizing estimation errors, with limited attention on potential security vulnerabilities, such as generating inappropriate content, violent actions, or racially offensive gestures and expressions. Current adversarial attacks on EHPS models often generate visually conspicuous perturbations, limiting their practicality and ability to expose real-world security threats. To address this limitation, we propose an unnoticeable adversarial method, termed \textbf{LatentStealth}, specifically tailored for EHPS models. The key idea is to exploit the structured latent representations of natural images as the medium for crafting perturbations. Instead of injecting noise directly into the pixel space, our method projects inputs into the latent space, where adversarial patterns are generated and progressively refined along optimized directions. This latent-space manipulation enables the attack to maintain high imperceptibility while preserving its effectiveness. Furthermore, as the optimization process is guided by only a small number of model output queries, the framework achieves competitive attack performance with low computational overhead, making it both practical and efficient for real-world scenarios. Extensive experiments on the 3DPW and UBody datasets demonstrate the superiority of LatentStealth, revealing critical vulnerabilities in current systems. These findings highlight the urgent need to address and mitigate security risks in digital human generation technologies.

cs.CV↗

arXiv2026

HVG-3D: Bridging Real and Simulation Domains for 3D-Conditional Hand-Object Interaction Video Synthesis

Recent methods have made notable progress in the visual quality of hand-object interaction video synthesis. However, most approaches rely on 2D control signals that lack spatial expressiveness and limit the utilization of synthetic 3D conditional data. To address these limitations, we propose HVG-3D, a unified framework for 3D-aware hand-object interaction (HOI) video synthesis conditioned on explicit 3D representations. Specifically, we develop a diffusion-based architecture augmented with a 3D ControlNet, which encodes geometric and motion cues from 3D inputs to enable explicit 3D reasoning during video synthesis. To achieve high-quality synthesis, HVG-3D is designed with two core components: (i) a 3D-aware HOI video generation diffusion architecture that encodes geometric and motion cues from 3D inputs for explicit 3D reasoning; and (ii) a hybrid pipeline for constructing input and condition signals, enabling flexible and precise control during both training and inference. During inference, given a single real image and a 3D control signal from either simulation or real data, HVG-3D generates high-fidelity, temporally consistent videos with precise spatial and temporal control. Experiments on the TASTE-Rob dataset demonstrate that HVG-3D achieves state-of-the-art spatial fidelity, temporal coherence, and controllability, while enabling effective utilization of both real and simulated data.

cs.CV↗

arXiv2025

Isocapacitary constants for the $p$-Laplacian on compact manifolds

In this paper, we introduce Steklov and Neumann isocapacitary constants for the $p$-Laplacian on compact manifolds. These constants yield two-sided bounds for the $(p,α)$-Sobolev constants, which degenerate to upper and lower bounds for the first nontrivial Steklov and Neumann eigenvalues of the $p$-Laplacian when $α= 1$.

math.DG↗

arXiv2025

Blow-up suppression of the Patlak-Keller-Segel-Navier-Stokes system via Taylor-Couette flow

Motivated by the use of Taylor-Couette flow in extracorporeal circulation devices [K$\ddot{\rm o}$rfer et al., 2003, 26(4): 331-338], where it leads to an accumulation of platelets and plasma proteins in the vortex center and therefore to a decreased probability of contact between platelets and material surfaces and its protein adsorption per square unit is significantly lower than laminar flow. Increased platelet adhesion or protein adsorption on the device surface can induce platelet aggregation or thrombosis, which is analogous to the ``blow-up phenomenon" in mathematical modeling. Here we mathematically analyze this stability mechanism and demonstrate that sufficiently strong flow can prevent blow-up from occurring. In details, we investigate the two-dimensional Patlak-Keller-Segel-Navier-Stokes system in an annular domain around a Taylor-Couette flow $U(r,θ)=A\big(r+\frac{1}{r} \big)(-\sinθ, \cosθ)^{T}$ with $(r,θ)\in[1,R]\times\mathbb{S}^{1}$, and prove that the solutions are globally bounded without any smallness restriction on the initial cell mass or velocity when $A$ is large.

math.AP↗

arXiv2025

Sublattice Dichotomy in Monolayer FeSe Superconductor

The pairing mechanism behind the monolayer FeSe is one essential question for iron-based superconductors. In this work, we show the sublattice degree of freedoms of monolayer FeSe plays a special role in its pairing properties, namely the sublattice dichotomy. The high-quality monolayer FeSe samples with atomic flat $1\times1$ topography on the SrTiO$_3$(001) substrates are grown by molecular beam epitaxy. By comparing the tunneling spectra at $α$ and $β$ Fe sublattices, we find the coherence peak of $α$-Fe at the inner gap $+V_i$ is higher than $β$-Fe while the coherence peak of $β$-Fe at $-V_i$ is higher than $α$-Fe with a similar amount. We also observed a reversed effect at the outer gap $\pm V_o$. We propose the $η$-pairing mechanism between $k$ and $-k+Q$ is the key mechanism for this unconventional sublattice dichotomy effect.

cond-mat.supr-con↗

arXiv2025

Distributed Koopman Learning with Incomplete Measurements

Koopman operator theory has emerged as a powerful tool for system identification, particularly for approximating nonlinear time-invariant systems (NTIS). This paper considers a network of agents with limited observation capabilities that collaboratively estimate the dynamics of an NTIS. A distributed deep Koopman learning algorithm is developed by integrating Koopman operator theory, deep neural networks, and consensus-based coordination. In the proposed framework, each agent approximates the system dynamics using its partial measurements and lifted states exchanged with its neighbors. This cooperative scheme enables accurate reconstruction of the global dynamics despite the absence of full-state information at individual agents. Simulation results on the Lunar Lander environment from OpenAI Gym demonstrate that the proposed method achieves performance comparable to the centralized deep Koopman learning with full-state access.

eess.SY↗

arXiv2025

Global bounded solutions of the 3D Patlak-Keller-Segel-Navier-Stokes system via Couette flow and logistic source

As is well-known, the solutions to the Patlak-Keller-Segel system in 3D may blow up in finite time regardless of any initial cell mass. In this paper, we investigate the existence of global bounded solutions of the 3D Patlak-Keller-Segel-Navier-Stokes system with a large initial cell mass via Couette flow or logistic source in a finite channel. On the one hand, it is proved that as long as the Couette flow is strong enough and the initial velocity is small, the bounded solutions of the system are global in time without any limitation on the initial mass $M$ if the logistic source term exists. On the other hand, if the logistic source term vanishes, it is proved that as long as the Couette flow is strong enough and the initial velocity is small, the solutions with a finite initial mass $M$, whose lower bound is $ \left(\frac{8π}{9}\right)^-$, are global in time.

math.AP↗

arXiv2025

On the variety generated by all semirings of order two

There are ten distinct two-element semirings up to isomorphism, denoted $ L_2, R_2, M_2, D_2, N_2, T_2, Z_2, W_2, Z_7 $, and $ Z_8 $ (see \cite{bk}). Among these, the multiplicative reductions of $ M_2, D_2, W_2 $, and $ Z_8 $ form semilattices, while the additive reductions of $ L_2, R_2, M_2, D_2, N_2 $, and $ T_2 $ are idempotent semilattices, commonly referred to as \emph{idempotent semirings}. In 2015, Vechtomov and Petrov \cite{vp} studied the variety generated by $ M_2, D_2, W_2 $, and $ Z_8 $, proving that it is finitely based. In the same year, Shao and Ren \cite{srii} examined the variety generated by the six idempotent semirings, demonstrating that every subvariety of this variety is finitely based. This paper systematically investigates the variety generated by all ten two-element semirings. We prove that this variety contains exactly 480 subvarieties, each of which is finitely based.

math.GR↗

arXiv2025

Suppression of blow-up for the 3D Patlak-Keller-Segel-Navier-Stokes system via the Couette flow

As is well known, for the 3D Patlak-Keller-Segel system, regardless of whether they are parabolic-elliptic or parabolic-parabolic forms, finite-time blow-up may occur for arbitrarily small values of the initial mass. In this paper, it is proved for the first time that one can prevent the finite-time blow-up when the initial mass is less than a certain critical threshold via the stabilizing effect of the moving Navier-Stokes flows. In details, we investigate the nonlinear stability of the Couette flow $(Ay, 0, 0)$ in the Patlak-Keller-Segel-Navier-Stokes system and show that if the Couette flow is sufficiently strong (A is large enough), then the solutions for Patlak-Keller-Segel-Navier-Stokes system are global in time provided that the initial velocity is sufficiently small and the initial cell mass is less than $\frac{24}{5} π^2$.

math.AP↗

arXiv2025

Stability threshold of Couette flow for 3D Boussinesq system in Sobolev spaces

In this paper, we investigate the nonlinear stability and transition threshold for the 3D Boussinesq system in Sobolev space under the high Reynolds number and small thermal diffusion in $\mathbb{T}\times\mathbb{R}\times\mathbb{T} $. It is proved that if the initial velocity $v_{\rm in}$ and the initial temperature $ θ_{\rm in} $ satisfy $ \|v_{\rm in}-(y,0,0)\|_{H^{2}}\leq \varepsilonν, \|θ_{\rm in}\|_{H^{2}}\leq \varepsilonν^{2} $, respectively for some $ \varepsilon>0 $ independent of the Reynolds number or thermal diffusion, then the solutions of 3D Boussinesq system are global in time.

math.AP↗

arXiv2025

Deep Spatio-Temporal Neural Network for Air Quality Reanalysis

Air quality prediction is key to mitigating health impacts and guiding decisions, yet existing models tend to focus on temporal trends while overlooking spatial generalization. We propose AQ-Net, a spatiotemporal reanalysis model for both observed and unobserved stations in the near future. AQ-Net utilizes the LSTM and multi-head attention for the temporal regression. We also propose a cyclic encoding technique to ensure continuous time representation. To learn fine-grained spatial air quality estimation, we incorporate AQ-Net with the neural kNN to explore feature-based interpolation, such that we can fill the spatial gaps given coarse observation stations. To demonstrate the efficiency of our model for spatiotemporal reanalysis, we use data from 2013-2017 collected in northern China for PM2.5 analysis. Extensive experiments show that AQ-Net excels in air quality reanalysis, highlighting the potential of hybrid spatio-temporal models to better capture environmental dynamics, especially in urban areas where both spatial and temporal variability are critical.

cs.LG↗

arXiv2025

QCS: Feature Refining from Quadruplet Cross Similarity for Facial Expression Recognition

Facial expression recognition faces challenges where labeled significant features in datasets are mixed with unlabeled redundant ones. In this paper, we introduce Cross Similarity Attention (CSA) to mine richer intrinsic information from image pairs, overcoming a limitation when the Scaled Dot-Product Attention of ViT is directly applied to calculate the similarity between two different images. Based on CSA, we simultaneously minimize intra-class differences and maximize inter-class differences at the fine-grained feature level through interactions among multiple branches. Contrastive residual distillation is utilized to transfer the information learned in the cross module back to the base network. We ingeniously design a four-branch centrally symmetric network, named Quadruplet Cross Similarity (QCS), which alleviates gradient conflicts arising from the cross module and achieves balanced and stable training. It can adaptively extract discriminative features while isolating redundant ones. The cross-attention modules exist during training, and only one base branch is retained during inference, resulting in no increase in inference time. Extensive experiments show that our proposed method achieves state-of-the-art performance on several FER datasets.

cs.CV↗

arXiv2025

Constructing Boundary-identical Microstructures via Guided Diffusion for Fast Multiscale Topology Optimization

Hierarchical structures exhibit critical features across multiple scales. However, designing multiscale structures demands significant computational resources, and ensuring connectivity between microstructures remains a key challenge. To address these issues, \textit{\textbf{large-range, boundary-identical microstructure datasets}} are successfully constructed, where the microstructures share the same boundaries and exhibit a wide range of elastic moduli. This approach enables highly efficient multiscale topology optimization. Central to our technique adopts a deep generative model, guided diffusion, to generate microstructures under the two conditions, including the specified boundary and homogenized elastic tensor. We generate the desired datasets using active learning approaches, where microstructures with diverse elastic moduli are iteratively added to the dataset, which is then retrained. %We achieve the desired datasets by active learning approaches which are alternately adding microstructures with diverse elastic modulus constructed by the deep generative model into the dataset and retraining the deep generative model. After that, sixteen boundary-identical microstructure datasets with wide ranges of elastic modulus %high property coverage are constructed. We demonstrate the effectiveness and practicability of the obtained datasets over various multiscale design examples. Specifically, in the design of a mechanical cloak, we utilize macrostructures with $30 \times 30$ elements and microstructures filled with $256 \times 256$ elements. The entire reverse design process is completed within one minute, significantly enhancing the efficiency of the multiscale topology optimization.

cs.CE↗

arXiv2024

FFA Sora, video generation as fundus fluorescein angiography simulator

Fundus fluorescein angiography (FFA) is critical for diagnosing retinal vascular diseases, but beginners often struggle with image interpretation. This study develops FFA Sora, a text-to-video model that converts FFA reports into dynamic videos via a Wavelet-Flow Variational Autoencoder (WF-VAE) and a diffusion transformer (DiT). Trained on an anonymized dataset, FFA Sora accurately simulates disease features from the input text, as confirmed by objective metrics: Frechet Video Distance (FVD) = 329.78, Learned Perceptual Image Patch Similarity (LPIPS) = 0.48, and Visual-question-answering Score (VQAScore) = 0.61. Specific evaluations showed acceptable alignment between the generated videos and textual prompts, with BERTScore of 0.35. Additionally, the model demonstrated strong privacy-preserving performance in retrieval evaluations, achieving an average Recall@K of 0.073. Human assessments indicated satisfactory visual quality, with an average score of 1.570(scale: 1 = best, 5 = worst). This model addresses privacy concerns associated with sharing large-scale FFA data and enhances medical education.

cs.CV↗

arXiv2024

GoHD: Gaze-oriented and Highly Disentangled Portrait Animation with Rhythmic Poses and Realistic Expression

Audio-driven talking head generation necessitates seamless integration of audio and visual data amidst the challenges posed by diverse input portraits and intricate correlations between audio and facial motions. In response, we propose a robust framework GoHD designed to produce highly realistic, expressive, and controllable portrait videos from any reference identity with any motion. GoHD innovates with three key modules: Firstly, an animation module utilizing latent navigation is introduced to improve the generalization ability across unseen input styles. This module achieves high disentanglement of motion and identity, and it also incorporates gaze orientation to rectify unnatural eye movements that were previously overlooked. Secondly, a conformer-structured conditional diffusion model is designed to guarantee head poses that are aware of prosody. Thirdly, to estimate lip-synchronized and realistic expressions from the input audio within limited training data, a two-stage training strategy is devised to decouple frequent and frame-wise lip motion distillation from the generation of other more temporally dependent but less audio-related motions, e.g., blinks and frowns. Extensive experiments validate GoHD's advanced generalization capabilities, demonstrating its effectiveness in generating realistic talking face results on arbitrary subjects.

cs.CV↗

arXiv2024

Stability of a class of supercritical volume-filling chemotaxis-fluid model near Couette flow

Consider a class of chemotaxis-fluid model incorporating a volume-filling effect in the sense of Painter and Hillen (Can. Appl. Math. Q. 2002; 10(4): 501-543), which is a supercritical parabolic-elliptic Keller-Segel system. As shown by Winkler et al., for any given mass, there exists a corresponding solution of the same mass that blows up in either finite or infinite time. In this paper, we investigate the stability properties of the two dimensional Patlak-Keller-Segel-type chemotaxis-fluid model near the Couette flow $ (Ay, 0) $ in $ \mathbb{T}\times\mathbb{R}, $ and show that the solutions are global in time as long as the initial cell mass $M<\frac{2π}{\sqrt{3}} $ and the shear flow is sufficiently strong ($A$ is large enough).

math.AP↗