arXiv subjects

Xiaoming Zhao

Publications and source records attributed to Xiaoming Zhao.

At least 19 recordsLinked to original sources

arXiv2026

Velox: Learning Representations of 4D Geometry and Appearance

We introduce a framework for learning latent representations of 4D objects which are descriptive, faithfully capturing object geometry and appearance; compressive, aiding in downstream efficiency; and accessible, requiring minimal input, i.e., an unstructured dynamic point cloud, to construct. Specifically, Velox trains an encoder to compress spatiotemporal color point clouds into a set of dynamic shape tokens. These tokens are supervised using two complementary decoders: a 4D surface decoder, which models the time-varying surface distribution capturing the geometry; and a Gaussian decoder, which maps the tokens to 3D Gaussians, helping learn appearance. To demonstrate the utility of our representation, we evaluate it across three downstream tasks -- video-to-4D generation, 3D tracking, and cloth simulation via image-to-4D generation -- and observe strong performances in all settings.

cs.CV↗

arXiv2026

LiTo: Surface Light Field Tokenization

We propose a 3D latent representation that jointly models object geometry and view-dependent appearance. Most prior works focus on either reconstructing 3D geometry or predicting view-independent diffuse appearance, and thus struggle to capture realistic view-dependent effects. Our approach leverages that RGB-depth images provide samples of a surface light field. By encoding random subsamples of this surface light field into a compact set of latent vectors, our model learns to represent both geometry and appearance within a unified 3D latent space. This representation reproduces view-dependent effects such as specular highlights and Fresnel reflections under complex lighting. We further train a latent flow matching model on this representation to learn its distribution conditioned on a single input image, enabling the generation of 3D objects with appearances consistent with the lighting and materials in the input. Experiments show that our approach achieves higher visual quality and better input fidelity than existing methods.

cs.CV↗

arXiv2026

High-Resolution Electron Paramagnetic Resonance

Electron paramagnetic resonance (EPR) is a valuable tool for physics, chemistry, biology and medicine, providing complementary spectroscopic information to NMR. It has long been known that EPR at high magnetic fields offers greater spectral resolution, but limitations in the THz instrumentation have prevented the full realization of these opportunities. Here we describe an EPR spectrometer at the high magnetic field of 14 T using 396 GHz excitation, which adapts techniques from liquid-state NMR to obtain sharp EPR resonances with a width of 210 ppb (full-width half-maximum). We use this to measure resonance positions, and hence g-factors, with a precision that reaches $\pm$16 ppb. Our use of in-situ liquid-state NMR of our solvent within the same sample improves the accuracy of these measurements: it allows us to reference our EPR measurement back to dilute gas 3He NMR for which quantum calculations are accurate. We measure the g-factor of N@C60 in deuterated toluene as g = 2.002 099 09 (3), where the 3 in brackets means that the uncertainty on the last digit is $\pm$3.

physics.ins-det↗

arXiv2025

Studying Classifier(-Free) Guidance From a Classifier-Centric Perspective

Classifier-free guidance has become a staple for conditional generation with denoising diffusion models. However, a comprehensive understanding of classifier-free guidance is still missing. In this work, we carry out an empirical study to provide a fresh perspective on classifier-free guidance. Concretely, instead of solely focusing on classifier-free guidance, we trace back to the root, i.e., classifier guidance, pinpoint the key assumption for the derivation, and conduct a systematic study to understand the role of the classifier. On 1D data, we find that both classifier guidance and classifier-free guidance achieve conditional generation by pushing the denoising diffusion trajectories away from decision boundaries, i.e., areas where conditional information is usually entangled and is hard to learn. To validate this classifier-centric perspective on high-dimensional data, we assess whether a flow-matching postprocessing step that is designed to narrow the gap between a pre-trained diffusion model's learned distribution and the real data distribution, especially near decision boundaries, can improve the performance. Experiments on various datasets verify our classifier-centric understanding.

cs.CV↗

arXiv2025

3D Shape Tokenization via Latent Flow Matching

We introduce a latent 3D representation that models 3D surfaces as probability density functions in 3D, i.e., p(x,y,z), with flow-matching. Our representation is specifically designed for consumption by machine learning models, offering continuity and compactness by construction while requiring only point clouds and minimal data preprocessing. Despite being a data-driven method, our use of flow matching in the 3D space enables interesting geometry properties, including the capabilities to perform zero-shot estimation of surface normal and deformation field. We evaluate with several machine learning tasks, including 3D-CLIP, unconditional generative models, single-image conditioned generative model, and intersection-point estimation. Across all experiments, our models achieve competitive performance to existing baselines, while requiring less preprocessing and auxiliary information from training data.

cs.CV↗

arXiv2025

Tentaclelike spectra and bound states in Hatano-Nelson chain with long-range impurity coupling

In non-Hermitian systems, the energy spectra and eigenstates exhibit high sensitivity to boundary conditions, lattice geometries, and local impurities. In this paper, we study the effect of long-range impurity coupling, located far from the boundaries, on the paradigmatic non-Hermitian Hatano-Nelson model. Through exact analytical treatment, we reveal the intriguing tentacle-like spectral structures that emerge from the otherwise Bloch or non-Bloch spectra under periodic or open boundary conditions, respectively. We show that these spectral tentacles are associated with emergent bound states near the impurity, with their number determined by the coupling range. We further determine the localization length of these tentacled states using the transfer matrix. Our work indicates that the long-range impurity coupling cannot be treated as a mere perturbative effect and holds promise for state manipulations in non-Hermitian systems.

cond-mat.dis-nn↗

arXiv2024

PDC & DM-SFT: A Road for LLM SQL Bug-Fix Enhancing

Code Large Language Models (Code LLMs), such as Code llama and DeepSeek-Coder, have demonstrated exceptional performance in the code generation tasks. However, most existing models focus on the abilities of generating correct code, but often struggle with bug repair. We introduce a suit of methods to enhance LLM's SQL bug-fixing abilities. The methods are mainly consisted of two parts: A Progressive Dataset Construction (PDC) from scratch and Dynamic Mask Supervised Fine-tuning (DM-SFT). PDC proposes two data expansion methods from the perspectives of breadth first and depth first respectively. DM-SFT introduces an efficient bug-fixing supervised learning approach, which effectively reduce the total training steps and mitigate the "disorientation" in SQL code bug-fixing training. In our evaluation, the code LLM models trained with two methods have exceeds all current best performing model which size is much larger.

cs.CL↗

arXiv2024

IllumiNeRF: 3D Relighting Without Inverse Rendering

Existing methods for relightable view synthesis -- using a set of images of an object under unknown lighting to recover a 3D representation that can be rendered from novel viewpoints under a target illumination -- are based on inverse rendering, and attempt to disentangle the object geometry, materials, and lighting that explain the input images. Furthermore, this typically involves optimization through differentiable Monte Carlo rendering, which is brittle and computationally-expensive. In this work, we propose a simpler approach: we first relight each input image using an image diffusion model conditioned on target environment lighting and estimated object geometry. We then reconstruct a Neural Radiance Field (NeRF) with these relit images, from which we render novel views under the target lighting. We demonstrate that this strategy is surprisingly competitive and achieves state-of-the-art results on multiple relighting benchmarks. Please see our project page at https://illuminerf.github.io/.

cs.CV↗

arXiv2024

Navigation-grade interferometric air-core antiresonant fibre optic gyroscope with enhanced thermal stability

We present a groundbreaking navigation-grade interferometric air-core fibre optic gyroscope (IFOG) using a quadrupolar-wound coil of four-tube truncated double nested antiresonant nodeless fibre (tDNANF). This state-of-the-art tDNANF simultaneously achieves low loss, low bend loss, single-spatial-mode operation, and exceptional linear polarization purity over a broad wavelength range. Our 469 m tDNANF coil demonstrated a polarization extinction ratio (PER) of ~20 dB when illuminated by an amplified spontaneous emission (ASE) source spanning 1525-1565 nm. Under these conditions, the gyro archives an angular random walk (ARW) of 0.0038 deg h-1/2 and a bias-stability (BS) drift over 8500 s of 0.0014 deg h-1, marking the first instance of navigation-grade performance in air-core FOGs. Additionally, we validated the low thermal sensitivity of air-core FOGs, with reductions of 9.24/10.68/6.82 compared to that of conventional polarization-maintaining solid-core FOGs of the same size across various temperature ranges. These results represent a significant step towards long-standing promise of high-precision inertial navigation applications with superior environmental adaptability.

physics.optics↗

arXiv2024

NeRFDeformer: NeRF Transformation from a Single View via 3D Scene Flows

We present a method for automatically modifying a NeRF representation based on a single observation of a non-rigid transformed version of the original scene. Our method defines the transformation as a 3D flow, specifically as a weighted linear blending of rigid transformations of 3D anchor points that are defined on the surface of the scene. In order to identify anchor points, we introduce a novel correspondence algorithm that first matches RGB-based pairs, then leverages multi-view information and 3D reprojection to robustly filter false positives in two steps. We also introduce a new dataset for exploring the problem of modifying a NeRF scene through a single observation. Our dataset ( https://github.com/nerfdeformer/nerfdeformer ) contains 113 synthetic scenes leveraging 47 3D assets. We show that our proposed method outperforms NeRF editing methods as well as diffusion-based methods, and we also explore different methods for filtering correspondences.

cs.CV↗

arXiv2024

Robust Knowledge Distillation Based on Feature Variance Against Backdoored Teacher Model

Benefiting from well-trained deep neural networks (DNNs), model compression have captured special attention for computing resource limited equipment, especially edge devices. Knowledge distillation (KD) is one of the widely used compression techniques for edge deployment, by obtaining a lightweight student model from a well-trained teacher model released on public platforms. However, it has been empirically noticed that the backdoor in the teacher model will be transferred to the student model during the process of KD. Although numerous KD methods have been proposed, most of them focus on the distillation of a high-performing student model without robustness consideration. Besides, some research adopts KD techniques as effective backdoor mitigation tools, but they fail to perform model compression at the same time. Consequently, it is still an open problem to well achieve two objectives of robust KD, i.e., student model's performance and backdoor mitigation. To address these issues, we propose RobustKD, a robust knowledge distillation that compresses the model while mitigating backdoor based on feature variance. Specifically, RobustKD distinguishes the previous works in three key aspects: (1) effectiveness: by distilling the feature map of the teacher model after detoxification, the main task performance of the student model is comparable to that of the teacher model; (2) robustness: by reducing the characteristic variance between the teacher model and the student model, it mitigates the backdoor of the student model under backdoored teacher model scenario; (3) generic: RobustKD still has good performance in the face of multiple data models (e.g., WRN 28-4, Pyramid-200) and diverse DNNs (e.g., ResNet50, MobileNet).

cs.LG↗

arXiv2024

GoMAvatar: Efficient Animatable Human Modeling from Monocular Video Using Gaussians-on-Mesh

We introduce GoMAvatar, a novel approach for real-time, memory-efficient, high-quality animatable human modeling. GoMAvatar takes as input a single monocular video to create a digital avatar capable of re-articulation in new poses and real-time rendering from novel viewpoints, while seamlessly integrating with rasterization-based graphics pipelines. Central to our method is the Gaussians-on-Mesh representation, a hybrid 3D model combining rendering quality and speed of Gaussian splatting with geometry modeling and compatibility of deformable meshes. We assess GoMAvatar on ZJU-MoCap data and various YouTube videos. GoMAvatar matches or surpasses current monocular human modeling algorithms in rendering quality and significantly outperforms them in computational efficiency (43 FPS) while being memory-efficient (3.63 MB per subject).

cs.CV↗

arXiv2024

Pseudo-Generalized Dynamic View Synthesis from a Video

Rendering scenes observed in a monocular video from novel viewpoints is a challenging problem. For static scenes the community has studied both scene-specific optimization techniques, which optimize on every test scene, and generalized techniques, which only run a deep net forward pass on a test scene. In contrast, for dynamic scenes, scene-specific optimization techniques exist, but, to our best knowledge, there is currently no generalized method for dynamic novel view synthesis from a given monocular video. To answer whether generalized dynamic novel view synthesis from monocular videos is possible today, we establish an analysis framework based on existing techniques and work toward the generalized approach. We find a pseudo-generalized process without scene-specific appearance optimization is possible, but geometrically and temporally consistent depth estimates are needed. Despite no scene-specific appearance optimization, the pseudo-generalized approach improves upon some scene-specific methods.

cs.CV↗

arXiv2024

Non-orthogonal cavity modes near exceptional points in the far field

Non-orthogonal eigenstates are a fundamental feature of non-Hermitian systems and are accompanied by the emergence of nontrivial features. However, the platforms to explore non-Hermitian mode couplings mainly measure near-field effects, and the far-field behaviour remain mostly unexplored. Here, we study how a microcavity with non-Hermitian mode coupling exhibits eigenstate non-orthogonality by investigating the spatial field and the far-field polarization of cavity modes. The non-Hermiticity arises from asymmetric backscattering, which is controlled by integrating two scatterers of different size and location into a microdisk. We observe that the spatial field overlaps of two modes increases abruptly to its maximum value, whilst different far-field elliptical polarizations of two modes coalesce when approaching an exceptional point. We demonstrate such features experimentally by measuring the far-field polarization from the fabricated microdisks. Our work reveals the non-orthogonality in the far-field degree of freedom, and the integrability of the microdisks paves a way to integrate more non-Hermitian optical properties into nanophotonic systems.

physics.optics↗

arXiv2023

ALIKED: A Lighter Keypoint and Descriptor Extraction Network via Deformable Transformation

Image keypoints and descriptors play a crucial role in many visual measurement tasks. In recent years, deep neural networks have been widely used to improve the performance of keypoint and descriptor extraction. However, the conventional convolution operations do not provide the geometric invariance required for the descriptor. To address this issue, we propose the Sparse Deformable Descriptor Head (SDDH), which learns the deformable positions of supporting features for each keypoint and constructs deformable descriptors. Furthermore, SDDH extracts descriptors at sparse keypoints instead of a dense descriptor map, which enables efficient extraction of descriptors with strong expressiveness. In addition, we relax the neural reprojection error (NRE) loss from dense to sparse to train the extracted sparse descriptors. Experimental results show that the proposed network is both efficient and powerful in various visual measurement tasks, including image matching, 3D reconstruction, and visual relocalization.

cs.CV↗

arXiv2022

Occupancy Planes for Single-view RGB-D Human Reconstruction

Single-view RGB-D human reconstruction with implicit functions is often formulated as per-point classification. Specifically, a set of 3D locations within the view-frustum of the camera are first projected independently onto the image and a corresponding feature is subsequently extracted for each 3D location. The feature of each 3D location is then used to classify independently whether the corresponding 3D point is inside or outside the observed object. This procedure leads to sub-optimal results because correlations between predictions for neighboring locations are only taken into account implicitly via the extracted features. For more accurate results we propose the occupancy planes (OPlanes) representation, which enables to formulate single-view RGB-D human reconstruction as occupancy prediction on planes which slice through the camera's view frustum. Such a representation provides more flexibility than voxel grids and enables to better leverage correlations than per-point classification. On the challenging S3D data we observe a simple classifier based on the OPlanes representation to yield compelling results, especially in difficult situations with partial occlusions due to other objects and partial visibility, which haven't been addressed by prior work.

cs.CV↗

arXiv2022

Initialization and Alignment for Adversarial Texture Optimization

While recovery of geometry from image and video data has received a lot of attention in computer vision, methods to capture the texture for a given geometry are less mature. Specifically, classical methods for texture generation often assume clean geometry and reasonably well-aligned image data. While very recent methods, e.g., adversarial texture optimization, better handle lower-quality data obtained from hand-held devices, we find them to still struggle frequently. To improve robustness, particularly of recent adversarial texture optimization, we develop an explicit initialization and an alignment procedure. It deals with complex geometry due to a robust mapping of the geometry to the texture map and a hard-assignment-based initialization. It deals with misalignment of geometry and images by integrating fast image-alignment into the texture refinement optimization. We demonstrate efficacy of our texture generation on a dataset of 11 scenes with a total of 2807 frames, observing 7.8% and 11.1% relative improvements regarding perceptual and sharpness measurements.

cs.CV↗

arXiv2022

Generative Multiplane Images: Making a 2D GAN 3D-Aware

What is really needed to make an existing 2D GAN 3D-aware? To answer this question, we modify a classical GAN, i.e., StyleGANv2, as little as possible. We find that only two modifications are absolutely necessary: 1) a multiplane image style generator branch which produces a set of alpha maps conditioned on their depth; 2) a pose-conditioned discriminator. We refer to the generated output as a 'generative multiplane image' (GMPI) and emphasize that its renderings are not only high-quality but also guaranteed to be view-consistent, which makes GMPIs different from many prior works. Importantly, the number of alpha maps can be dynamically adjusted and can differ between training and inference, alleviating memory concerns and enabling fast training of GMPIs in less than half a day at a resolution of $1024^2$. Our findings are consistent across three challenging and common high-resolution datasets, including FFHQ, AFHQv2, and MetFaces.

cs.CV↗