arXiv subjects

Weixin Li

Publications and source records attributed to Weixin Li.

At least 37 records · Page 2Linked to original sources

arXiv2025

EmboMatrix: A Scalable Training-Ground for Embodied Decision-Making

Embodied decision-making enables agents to translate high-level goals into executable actions through continuous interactions within the physical world, forming a cornerstone of general-purpose embodied intelligence. Large language models (LLMs), with their general decision-making capabilities, offer a promising path to realize this potential; however, LLMs trained solely on language lack exposure to physical environments, limiting their true embodied understanding. To bridge this gap, we propose the concept of a training ground: a comprehensive infrastructure that provides task and scene simulation, embodied interaction, and feedback signals, offering a one-stop solution for LLM acquire genuine embodied decision-making skills. In this work, we present EmboMatrix, the first training ground of its kind, providing massive and diverse tasks with efficient simulation and precise rewards. EmboMatrix incorporates a series of novel techniques: a multi-agent data engine for large-scale task and scene generation, a distributed heterogeneous-hardware system for scalable simulation, and a multi-level reward architecture for precise supervision. Leveraging EmboMatrix, we cultivate EmboBrain, an LLM whose embodied decision-making abilities emerge from extensive embodied interactions. Experiments show that EmboBrain-7B surpasses the 671B DeepSeek-R1 baseline by 9.5\% on two challenging embodied decision-making benchmarks, demonstrating the power of interactive, environment-grounded learning for building truly intelligent embodied agents.

cs.AI↗

arXiv2025

Video Demoireing using Focused-Defocused Dual-Camera System

Moire patterns, unwanted color artifacts in images and videos, arise from the interference between spatially high-frequency scene contents and the spatial discrete sampling of digital cameras. Existing demoireing methods primarily rely on single-camera image/video processing, which faces two critical challenges: 1) distinguishing moire patterns from visually similar real textures, and 2) preserving tonal consistency and temporal coherence while removing moire artifacts. To address these issues, we propose a dual-camera framework that captures synchronized videos of the same scene: one in focus (retaining high-quality textures but may exhibit moire patterns) and one defocused (with significantly reduced moire patterns but blurred textures). We use the defocused video to help distinguish moire patterns from real texture, so as to guide the demoireing of the focused video. We propose a frame-wise demoireing pipeline, which begins with an optical flow based alignment step to address any discrepancies in displacement and occlusion between the focused and defocused frames. Then, we leverage the aligned defocused frame to guide the demoireing of the focused frame using a multi-scale CNN and a multi-dimensional training loss. To maintain tonal and temporal consistency, our final step involves a joint bilateral filter to leverage the demoireing result from the CNN as the guide to filter the input focused frame to obtain the final output. Experimental results demonstrate that our proposed framework largely outperforms state-of-the-art image and video demoireing methods.

cs.CV↗

arXiv2025

Multi-Grained Compositional Visual Clue Learning for Image Intent Recognition

In an era where social media platforms abound, individuals frequently share images that offer insights into their intents and interests, impacting individual life quality and societal stability. Traditional computer vision tasks, such as object detection and semantic segmentation, focus on concrete visual representations, while intent recognition relies more on implicit visual clues. This poses challenges due to the wide variation and subjectivity of such clues, compounded by the problem of intra-class variety in conveying abstract concepts, e.g. "enjoy life". Existing methods seek to solve the problem by manually designing representative features or building prototypes for each class from global features. However, these methods still struggle to deal with the large visual diversity of each intent category. In this paper, we introduce a novel approach named Multi-grained Compositional visual Clue Learning (MCCL) to address these challenges for image intent recognition. Our method leverages the systematic compositionality of human cognition by breaking down intent recognition into visual clue composition and integrating multi-grained features. We adopt class-specific prototypes to alleviate data imbalance. We treat intent recognition as a multi-label classification problem, using a graph convolutional network to infuse prior knowledge through label embedding correlations. Demonstrated by a state-of-the-art performance on the Intentonomy and MDID datasets, our approach advances the accuracy of existing methods while also possessing good interpretability. Our work provides an attempt for future explorations in understanding complex and miscellaneous forms of human expression.

cs.CV↗

arXiv2024

Generating Editable Head Avatars with 3D Gaussian GANs

Generating animatable and editable 3D head avatars is essential for various applications in computer vision and graphics. Traditional 3D-aware generative adversarial networks (GANs), often using implicit fields like Neural Radiance Fields (NeRF), achieve photorealistic and view-consistent 3D head synthesis. However, these methods face limitations in deformation flexibility and editability, hindering the creation of lifelike and easily modifiable 3D heads. We propose a novel approach that enhances the editability and animation control of 3D head avatars by incorporating 3D Gaussian Splatting (3DGS) as an explicit 3D representation. This method enables easier illumination control and improved editability. Central to our approach is the Editable Gaussian Head (EG-Head) model, which combines a 3D Morphable Model (3DMM) with texture maps, allowing precise expression control and flexible texture editing for accurate animation while preserving identity. To capture complex non-facial geometries like hair, we use an auxiliary set of 3DGS and tri-plane features. Extensive experiments demonstrate that our approach delivers high-quality 3D-aware synthesis with state-of-the-art controllability. Our code and models are available at https://github.com/liguohao96/EGG3D.

cs.CV↗

arXiv2024

Consistent Diffusion: Denoising Diffusion Model with Data-Consistent Training for Image Restoration

In this work, we address the limitations of denoising diffusion models (DDMs) in image restoration tasks, particularly the shape and color distortions that can compromise image quality. While DDMs have demonstrated a promising performance in many applications such as text-to-image synthesis, their effectiveness in image restoration is often hindered by shape and color distortions. We observe that these issues arise from inconsistencies between the training and testing data used by DDMs. Based on our observation, we propose a novel training method, named data-consistent training, which allows the DDMs to access images with accumulated errors during training, thereby ensuring the model to learn to correct these errors. Experimental results show that, across five image restoration tasks, our method has significant improvements over state-of-the-art methods while effectively minimizing distortions and preserving image fidelity.

cs.CV↗

arXiv2024

Leveraging Predicate and Triplet Learning for Scene Graph Generation

Scene Graph Generation (SGG) aims to identify entities and predict the relationship triplets \textit{\textless subject, predicate, object\textgreater } in visual scenes. Given the prevalence of large visual variations of subject-object pairs even in the same predicate, it can be quite challenging to model and refine predicate representations directly across such pairs, which is however a common strategy adopted by most existing SGG methods. We observe that visual variations within the identical triplet are relatively small and certain relation cues are shared in the same type of triplet, which can potentially facilitate the relation learning in SGG. Moreover, for the long-tail problem widely studied in SGG task, it is also crucial to deal with the limited types and quantity of triplets in tail predicates. Accordingly, in this paper, we propose a Dual-granularity Relation Modeling (DRM) network to leverage fine-grained triplet cues besides the coarse-grained predicate ones. DRM utilizes contexts and semantics of predicate and triplet with Dual-granularity Constraints, generating compact and balanced representations from two perspectives to facilitate relation recognition. Furthermore, a Dual-granularity Knowledge Transfer (DKT) strategy is introduced to transfer variation from head predicates/triplets to tail ones, aiming to enrich the pattern diversity of tail classes to alleviate the long-tail problem. Extensive experiments demonstrate the effectiveness of our method, which establishes new state-of-the-art performance on Visual Genome, Open Image, and GQA datasets. Our code is available at \url{https://github.com/jkli1998/DRM}

cs.CV↗

arXiv2024

ReWiTe: Realistic Wide-angle and Telephoto Dual Camera Fusion Dataset via Beam Splitter Camera Rig

The fusion of images from dual camera systems featuring a wide-angle and a telephoto camera has become a hotspot problem recently. By integrating simultaneously captured wide-angle and telephoto images from these systems, the resulting fused image achieves a wide field of view (FOV) coupled with high-definition quality. Existing approaches are mostly deep learning methods, and predominantly rely on supervised learning, where the training dataset plays a pivotal role. However, current datasets typically adopt a data synthesis approach generate input pairs of wide-angle and telephoto images alongside ground-truth images. Notably, the wide-angle inputs are synthesized rather than captured using real wide-angle cameras, and the ground-truth image is captured by wide-angle camera whose quality is substantially lower than that of input telephoto images captured by telephoto cameras. To address these limitations, we introduce a novel hardware setup utilizing a beam splitter to simultaneously capture three images, i.e. input pairs and ground-truth images, from two authentic cellphones equipped with wide-angle and telephoto dual cameras. Specifically, the wide-angle and telephoto images captured by cellphone 2 serve as the input pair, while the telephoto image captured by cellphone 1, which is calibrated to match the optical path of the wide-angle image from cellphone 2, serves as the ground-truth image, maintaining quality on par with the input telephoto image. Experiments validate the efficacy of our newly introduced dataset, named ReWiTe, significantly enhances the performance of various existing methods for real-world wide-angle and telephoto dual image fusion tasks.

cs.CV↗

arXiv2024

BMLP: Behavior-aware MLP for Heterogeneous Sequential Recommendation

In real recommendation scenarios, users often have different types of behaviors, such as clicking and buying. Existing research methods show that it is possible to capture the heterogeneous interests of users through different types of behaviors. However, most multi-behavior approaches have limitations in learning the relationship between different behaviors. In this paper, we propose a novel multilayer perceptron (MLP)-based heterogeneous sequential recommendation method, namely behavior-aware multilayer perceptron (BMLP). Specifically, it has two main modules, including a heterogeneous interest perception (HIP) module, which models behaviors at multiple granularities through behavior types and transition relationships, and a purchase intent perception (PIP) module, which adaptively fuses subsequences of auxiliary behaviors to capture users' purchase intent. Compared with mainstream sequence models, MLP is competitive in terms of accuracy and has unique advantages in simplicity and efficiency. Extensive experiments show that BMLP achieves significant improvement over state-of-the-art algorithms on four public datasets. In addition, its pure MLP architecture leads to a linear time complexity.

cs.IR↗

arXiv2023

View Transition based Dual Camera Image Fusion

The dual camera system of wide-angle ($\bf{W}$) and telephoto ($\bf{T}$) cameras has been widely adopted by popular phones. In the overlap region, fusing the $\bf{W}$ and $\bf{T}$ images can generate a higher quality image. Related works perform pixel-level motion alignment or high-dimensional feature alignment of the $\bf{T}$ image to the view of the $\bf{W}$ image and then perform image/feature fusion, but the enhancement in occlusion area is ill-posed and can hardly utilize data from $\bf{T}$ images. Our insight is to minimize the occlusion area and thus maximize the use of pixels from $\bf{T}$ images. Instead of insisting on placing the output in the $\bf{W}$ view, we propose a view transition method to transform both $\bf{W}$ and $\bf{T}$ images into a mixed view and then blend them into the output. The transformation ratio is kept small and not apparent to users, and the center area of the output, which has accumulated a sufficient amount of transformation, can directly use the contents from the T view to minimize occlusions. Experimental results show that, in comparison with the SOTA methods, occlusion area is largely reduced by our method and thus more pixels of the $\bf{T}$ image can be used for improving the quality of the output image.

cs.CV↗

arXiv2023

Zero-Shot Scene Graph Generation via Triplet Calibration and Reduction

Scene Graph Generation (SGG) plays a pivotal role in downstream vision-language tasks. Existing SGG methods typically suffer from poor compositional generalizations on unseen triplets. They are generally trained on incompletely annotated scene graphs that contain dominant triplets and tend to bias toward these seen triplets during inference. To address this issue, we propose a Triplet Calibration and Reduction (T-CAR) framework in this paper. In our framework, a triplet calibration loss is first presented to regularize the representations of diverse triplets and to simultaneously excavate the unseen triplets in incompletely annotated training scene graphs. Moreover, the unseen space of scene graphs is usually several times larger than the seen space since it contains a huge number of unrealistic compositions. Thus, we propose an unseen space reduction loss to shift the attention of excavation to reasonable unseen compositions to facilitate the model training. Finally, we propose a contextual encoder to improve the compositional generalizations of unseen triplets by explicitly modeling the relative spatial relations between subjects and objects. Extensive experiments show that our approach achieves consistent improvements for zero-shot SGG over state-of-the-art methods. The code is available at https://github.com/jkli1998/T-CAR.

cs.CV↗

arXiv2020

Elastic and Fracture Behavior of Three-Dimensional Ply-to-Ply Angle Interlock Woven Composites: Through-Thickness, Size Effect, and Multiaxial Tests

This work presents a comprehensive investigation of the elastic and fracture behavior of ply-to-ply angle interlock three-dimensional woven composites. The research investigated novel splitting and wedge-driven out-of-plane fracture tests to shed light on the tensile fracture behavior in the thickness direction and to provide estimates of the out-of-plane tensile strength and fracture energy. In addition, size effect tests on geometrically-scaled Single Edge Notch Tension (SENT) specimens were performed to fully characterize the intra-laminar fracture energy of the material and to study the scaling of structural strength in this type of three-dimensional composites. The results confirmed that size effect in the structural strength of these materials is significant. In fact, even if the range of sizes investigated was broader than in any previous size effect study on traditional laminated composites and two-dimensional textile composites, all the experimental data fell in the transition zone between quasi-ductile and brittle behavior. This implies strong damage tolerance of the investigated three-dimensional composites. The analysis of the data via Bazant's Type II Size Effect Law (SEL) enabled the objective characterization of the intra-laminar fracture energy of three-dimensional composites for the first time. Finally, Arcan rig tests combined with X-ray micro-computed tomography allowed unprecedented insights on the different damage mechanisms under multi-axial nominal loading conditions, particularly tension-dominated and shear-dominated conditions.

cond-mat.mtrl-sci↗

arXiv2018

Branching of Hydraulic Cracks in Gas or Oil Shale with Closed Natural Fractures: How to Master Permeability

While the hydraulic fracturing technology, aka fracking (or fraccing, frac), has become highly developed and astonishingly successful, a consistent formulation of the associated fracture mechanics that would not conflict with some observations is still unavailable. It is attempted here. Classical fracture mechanics, as well as the current commercial softwares, predict vertical cracks to propagate without branching from the perforations of the horizontal well casing, which are typically spaced at 10 m or more. However, to explain the gas production rate at the wellhead, the crack spacing would have to be only about 0.1 m, which would increase the overall gas permeability of shale mass about 10,000$\times$. This permeability increase has generally been attributed to a preexisting system of orthogonal natural cracks, whose spacing is about 0.1 m. But their average age is about 100 million years, and a recent analysis indicated that these cracks must have been completely closed by secondary creep of shale in less than a million years. Here it is considered that the tectonic events that produced the natural cracks in shale must have also created weak layers with nano- or micro-cracking damage. It is numerically demonstrated that a greatly enhanced permeability along the weak layers, with a greatly increased transverse Biot coefficient, must cause the fracking to engender lateral branching and the opening of hydraulic cracks along the weak layers, even if these cracks are initially almost closed. A finite element crack band model, based on recently developed anisotropic spherocylindrical microplane constitutive law, demonstrates these findings.

physics.geo-ph↗

arXiv2018

Multiphysics Lattice Discrete Particle Modeling (M-LDPM) for the Simulation of Shale Fracture Permeability

A three-dimensional Multiphysics Lattice Discrete Particle Model (M-LDPM) framework is formulated to investigate the fracture permeability behavior of shale. The framework features a dual lattice system mimicking the mesostructure of the material and simulates coupled mechanical and flow behavior. The mechanical lattice model simulates the granular internal structure of shale and describes heterogeneous deformation by means of discrete compatibility and equilibrium equations. The network of flow lattice elements constitutes a dual graph of the mechanical lattice system. A discrete formulation of mass balance for the flow elements is formulated to model fluid flow along cracks. The overall computational framework is implemented with a mixed explicit-implicit integration scheme and a staggered coupling method that makes use of the dual lattice topology enabling the seamless two-way coupling of the mechanical and flow behaviors. The proposed model is used for the computational analysis of shale fracture permeability behavior by simulating triaxial direct shear tests on Marcellus shale specimens under various confining pressures. The simulated mechanical response is calibrated against the experimental data, and the predicted permeability values are also compared with the experimental measurements. Furthermore, the paper presents the scaling analysis of both the mechanical response and permeability measurements based on simulations performed on geometrically similar specimens with increasing size. The simulated stress-strain curves show a significant size effect in the post-peak due to the presence of localized fractures. The scaling analysis of permeability measurements enables prediction of permeability for large specimens by extrapolating the numerical results of small ones.

physics.geo-ph↗

arXiv2018

Water-phospholipid interactions at the interface of lipid membranes: comparison of different force fields

Water-phospholipid interactions at the lipid bilayer/water interfaces are of essential importance for the dynamics, stability and function of biological membrane, and are also strongly associated with numerous biological processes at the interfaces of lipid bilayers. Various force fields, such as the united-atom Berger force field, its two improved versions by Kukol and by Poger, and the all-atom Slipid force field developed recently, can be applied to simulating the structures of lipid bilayer, with their structural predictions in good agreement with experimental data. In this work, we show that despite the similarity in structural predictions of lipid bilayers, there are observable differences in formation of hydrogen bonds and the interaction energy profiles between water and phospholipid groups at the lipid bilayer/water interfaces, when four force fields for dipalmitoylphosphatidylcholine (DPPC) phospholipids are employed in molecular dynamics simulations. In particular, the Slipid force field yields more hydrogen bonds between water and phospholipids and more symmetrical interaction energy distributions for the two carboxylic groups on their respective acyl tails, compared to the Berger and its two improved force fields. These differences are mainly attributed to the different interfacial water distributions and ability to form hydrogen bonds between interfacial water and oxygen atoms of the DPPC lipids using different force fields. These results would be helpful in understanding the behaviors of water as well as its interaction with phospholipids at the lipid bilayer/water interfaces, and provide a guide for making the appropriate choice on the force field in simulations of lipid bilayers.

cond-mat.soft↗

arXiv2017

Elastic, strength, and fracture properties of Marcellus shale

Shale, a fine-grained sedimentary rock, is the key source rock for many of the world's most important oil and natural gas deposits. A deep understanding of the mechanical properties of shale is of vital importance in various geotechnical applications, including oil and gas exploitation. In this work, deformability, strength, and fracturing properties of Marcellus shale were investigated through an experimental study. Firstly, uniaxial compression, direct tension, and Brazilian tests were performed on the Marcellus shale specimens in various bedding plane orientations with respect to loading directions to measure the static mechanical properties and their anisotropy. Furthermore, the deformability of Marcellus shale was also studied through seismic velocity measurements for comparison with the static measurements. The experimental results revealed that the transversely isotropic model is applicable for describing the elastic behaviors of Marcellus shale in pure tension and compression. The elastic properties measured from these two experiments, however, were not exactly the same. Strength results showed that differences exist between splitting (Brazilian) and direct tensile strengths, both of which varied with bedding plane orientations and loading directions and were associated with different failure modes. Finally, a series of three-point-bending tests were conducted on specimens of increasing size in three different principal notch orientations to investigate the fracture properties of the material. It was found that there exists a significant size effect on the fracture properties calculated from the measured peak loads and by using the Linear Elastic Fracture Mechanics (LEFM) theory. The fracture properties can be uniquely identified, however, by using Bazant's Size Effect Law and they were found to be anisotropic.

physics.geo-ph↗

arXiv2017

Characterization of Marcellus Shale Fracture Properties through Size Effect Tests and Computations

Mechanical characterization of shale-like rocks requires understanding the scaling of the measured properties to enable the extrapolation from small scale laboratory tests to field study. In this paper, the size effect of Marcellus shale was analyzed, and the fracture properties were obtained through size effect tests. A number of fracture tests were conducted on Three-Point-Bending (TPB) specimens with increasing size. Test results show that the nominal strength decreases with increasing specimen size, and can be fitted well by Bazant's Size Effect Law (SEL). It is shown that SEL accounts for the effects of both specimen size and geometry, allowing an accurate identification of the initial fracture energy of the material, Gf, and the effective Fracture Process Zone (FPZ) length, cf. The obtained fracture properties were verified by the numerical simulations of the investigated specimens using standard Finite Element technique with cohesive model. Significant anisotropy was observed in the fracture properties determined in three principal notch orientations: arrester, divider, and short-transverse. The size effect of the measured structural strength and apparent fracture toughness was discussed. Neither strength-based criterion which neglects size effect, nor classic LEFM which does not account for the finiteness of the FPZ can predict the reported size effect data, and nonlinear fracture mechanics of the quasibrittle type is instead applicable.

physics.geo-ph↗

arXiv2016

A multiscale framework for the simulation of the anisotropic mechanical behavior of shale

Shale, like many other sedimentary rocks, is typically heterogeneous, anisotropic, and is characterized by partial alignment of anisotropic clay minerals and naturally formed bedding planes. In this study, a micromechanical framework based on the Lattice Discrete Particle Model (LDPM) is formulated to capture these features. Material anisotropy is introduced through an approximated geometric description of shale internal structure, which includes representation of material property variation with orientation and explicit modeling of parallel lamination. The model is calibrated by carrying out numerical simulations to match various experimental data, including the ones relevant to elastic properties, Brazilian tensile strength, and unconfined compressive strength. Furthermore, parametric study is performed to investigate the relationship between the mesoscale parameters and the macroscopic properties. It is shown that the dependence of the elastic stiffness, strength, and failure mode on loading orientation can be captured successfully. Finally, a homogenization approach based on the asymptotic expansion of field variables is applied to upscale the proposed micromechanical model, and the properties of the homogenized model are analyzed.

cond-mat.mtrl-sci↗

arXiv2016

Elastic Microplane Formulation for Transversely Isotropic Materials

This contribution investigates the extension of the microplane formulation to the description of transversely isotropic materials such as shale rock, foams, unidirectional composites, and ceramics. Two possible approaches are considered: 1) the spectral decomposition of the stiffness tensor to define the microplane constitutive laws in terms of energetically orthogonal eigenstrains and eigenstresses; and 2) the definition of orientation-dependent microplane elastic moduli. It is shown that the first approach provides a rigorous way to tackle anisotropy within the microplane framework whereas the second approach represents an approximation which, however, makes the formulation of nonlinear constitutive equations much simpler. The efficacy of the second approach in modeling the macroscopic elastic behavior is compared to the thermodynamic restrictions of the anisotropic parameters showing that a significant range of elastic properties can be modeled with excellent accuracy. Further, it is shown that it provides a very good approximation of the microplane stresses provided by the first approach, with the advantage of a simpler formulation. It is concluded that the spectral stiffness decomposition represents the best approach in such cases as for modeling unidirectional composites, in which accurately capturing the elastic behavior is important. The introduction of orientation-dependent microplane elastic moduli provides a simpler framework for the modeling of transversely isotropic materials with remarked inelastic behavior, as in the case, for example, of shale rock.

cond-mat.mtrl-sci↗