Search

We generate abstractions of buildings, reflecting the essential aspects of their geometry and structure, by learning to invert procedural models. We first build a dataset of abstract procedural building models paired with simulated point clouds and then learn the inverse mapping through a transformer. Given a point cloud, the trained transformer then infers the corresponding abstracted building in terms of a programmatic language description. This approach leverages expressive procedural models developed for gaming and animation, and thereby retains desirable properties such as efficient rendering of the inferred abstractions and strong priors for regularity and symmetry. Our approach achieves good reconstruction accuracy in terms of geometry and structure, as well as structurally consistent inpainting., Comment: 4 pages, 3 figures

This work presents a non-intrusive reduced-order modeling framework for dynamical systems with spatially localized features characterized by slow singular value decay. The proposed approach builds upon two existing methodologies for reduced and full-order non-intrusive modeling, namely Operator Inference (OpInf) and sparse Full-Order Model (sFOM) inference. We decompose the domain into two complementary subdomains which exhibit fast and slow singular value decay. The dynamics of the subdomain exhibiting slow singular value decay are learned with sFOM while the dynamics with intrinsically low dimensionality on the complementary subdomain are learned with OpInf. The resulting, coupled OpInf-sFOM formulation leverages the computational efficiency of OpInf and the high resolution of sFOM, and thus enables fast non-intrusive predictions for conditions beyond those sampled in the training data set. A novel regularization technique with a closed-form solution based on the Gershgorin disk theorem is introduced to promote stable sFOM and OpInf models. We also provide a data-driven indicator for the subdomain selection and ensure solution smoothness over the interface via a post-processing interpolation step. We evaluate the efficiency of the approach in terms of offline and online speedup through a quantitative, parametric computational cost analysis. We demonstrate the coupled OpInf-sFOM formulation for two test cases: a one-dimensional Burgers' model for which accurate predictions beyond the span of the training snapshots are presented, and a two-dimensional parametric model for the Pine Island Glacier ice thickness dynamics, for which the OpInf-sFOM model achieves an average prediction error on the order of $1 \%$ with an online speedup factor of approximately $8\times$ compared to the numerical simulation.

Advances in IoT technologies combined with new algorithms have enabled the collection and processing of high-rate multi-source data streams that quantify human behavior in a fine-grained level and can lead to deeper insights on individual behaviors as well as on the interplay between behaviors and the environment. In this paper, we present an integrated system that collects and extracts multiple behavioral and environmental indicators, aiming at improving public health policies for tackling obesity. Data collection takes place using passive methods based on smartphone and smartwatch applications that require minimal interaction with the user. Our goal is to present a detailed account of the design principles, the implementation processes, and the evaluation of integrated algorithms, especially given the challenges we faced, in particular (a) integrating multiple technologies, algorithms, and components under a single, unified system, and (b) large scale (big data) requirements. We also present evaluation results of the algorithms on datasets (public for most cases) such as an absolute error of 8-9 steps when counting steps, 0.86 F1-score for detecting visited locations, and an error of less than 12 mins for gross sleep time. Finally, we also briefly present studies that have been materialized using our system, thus demonstrating its potential value to public authorities and individual researchers., Comment: 15 pages, 4 figures, 6 tables, journal

Efficiently reconstructing accurate 3D models from monocular video is a key challenge in computer vision, critical for advancing applications in virtual reality, robotics, and scene understanding. Existing approaches typically require pre-computed camera parameters and frame-by-frame reconstruction pipelines, which are prone to error accumulation and entail significant computational overhead. To address these limitations, we introduce VideoLifter, a novel framework that leverages geometric priors from a learnable model to incrementally optimize a globally sparse to dense 3D representation directly from video sequences. VideoLifter segments the video sequence into local windows, where it matches and registers frames, constructs consistent fragments, and aligns them hierarchically to produce a unified 3D model. By tracking and propagating sparse point correspondences across frames and fragments, VideoLifter incrementally refines camera poses and 3D structure, minimizing reprojection error for improved accuracy and robustness. This approach significantly accelerates the reconstruction process, reducing training time by over 82% while surpassing current state-of-the-art methods in visual fidelity and computational efficiency., Comment: project page: https://videolifter.github.io

Fluctuations in the stock market rapidly shape the economic world and consumer markets, impacting millions of individuals. Hence, accurately forecasting it is essential for mitigating risks, including those associated with inactivity. Although research shows that hybrid models of Deep Learning (DL) and Machine Learning (ML) yield promising results, their computational requirements often exceed the capabilities of average personal computers, rendering them inaccessible to many. In order to address this challenge in this paper we optimize LightGBM (an efficient implementation of gradient-boosted decision trees (GBDT)) for maximum performance, while maintaining low computational requirements. We introduce novel feature engineering techniques including indicator-price slope ratios and differences of close and open prices divided by the corresponding 14-period Exponential Moving Average (EMA), designed to capture market dynamics and enhance predictive accuracy. Additionally, we test seven different feature and target variable transformation methods, including returns, logarithmic returns, EMA ratios and their standardized counterparts as well as EMA difference ratios, so as to identify the most effective ones weighing in both efficiency and accuracy. The results demonstrate Log Returns, Returns and EMA Difference Ratio constitute the best target variable transformation methods, with EMA ratios having a lower percentage of correct directional forecasts, and standardized versions of target variable transformations requiring significantly more training time. Moreover, the introduced features demonstrate high feature importance in predictive performance across all target variable transformation methods. This study highlights an accessible, computationally efficient approach to stock market forecasting using LightGBM, making advanced forecasting techniques more widely attainable., Comment: 20 pages, 11 figures

For $c\in(1,2)$ we consider the following operators \[ \mathcal{C}_{c}f(x) = \sup_{\lambda \in [-1/2,1/2)}\bigg| \sum_{n \neq 0}f(x-n) \frac{e^{2\pi i\lambda \lfloor |n|^{c} \rfloor}}{n}\bigg|\text{,}\quad \mathcal{C}^{\mathsf{sgn}}_{c}f(x) = \sup_{\lambda \in [-1/2,1/2)}\bigg| \sum_{n \neq 0}f(x-n) \frac{e^{2\pi i\lambda \mathsf{sign(n)} \lfloor |n|^{c} \rfloor}}{n}\bigg| \text{,} \] and prove that both extend boundedly on $\ell^p(\mathbb{Z})$, $p\in(1,\infty)$. The second main result is establishing almost everywhere pointwise convergence for the following ergodic averages \[ A_Nf(x)=\frac{1}{N}\sum_{n=1}^Nf(T^nS^{\lfloor n^c\rfloor}x)\text{,} \] where $T,S\colon X\to X$ are commuting measure-preserving transformations on a $\sigma$-finite measure space $(X,\mu)$, and $f\in L_{\mu}^p(X)$, $p\in(1,\infty)$. The point of departure for both proofs is the study of exponential sums with phases $\xi_2 \lfloor |n^c|\rfloor+ \xi_1n$ through the use of a simple variant of the circle method., Comment: 22 pages, no figures

In an educational setting, an estimate of the difficulty of multiple-choice questions (MCQs), a commonly used strategy to assess learning progress, constitutes very useful information for both teachers and students. Since human assessment is costly from multiple points of view, automatic approaches to MCQ item difficulty estimation are investigated, yielding however mixed success until now. Our approach to this problem takes a different angle from previous work: asking various Large Language Models to tackle the questions included in two different MCQ datasets, we leverage model uncertainty to estimate item difficulty. By using both model uncertainty features as well as textual features in a Random Forest regressor, we show that uncertainty features contribute substantially to difficulty prediction, where difficulty is inversely proportional to the number of students who can correctly answer a question. In addition to showing the value of our approach, we also observe that our model achieves state-of-the-art results on the BEA publicly available dataset., Comment: 14 pages,7 figures

We establish pointwise convergence for nonconventional ergodic averages taken along $\lfloor p^c\rfloor$, where $p$ is a prime number and $c\in(1,4/3)$ on $L^r$, $r\in(1,\infty)$. In fact, we consider averages along more general sequences $\lfloor h(p)\rfloor$, where $h$ belongs in a wide class of functions, the so-called $c$-regularly varying functions. We also establish uniform multiparameter oscillation estimates for our ergodic averages and the corresponding multiparameter pointwise ergodic theorem in the spirit of Dunford and Zygmund. A key ingredient of our approach are certain exponential sum estimates, which we also use for establishing a Waring-type result. Assuming that the Riemann zeta function has any zero-free strip upgrades our exponential sum estimates to polynomially saving ones and this makes a conditional result regarding the behavior of our ergodic averages on $L^1$ to not seem entirely out of reach., Comment: 38 pages, no figures

Gaussian splatting methods are emerging as a popular approach for converting multi-view image data into scene representations that allow view synthesis. In particular, there is interest in enabling view synthesis for dynamic scenes using only monocular input data -- an ill-posed and challenging problem. The fast pace of work in this area has produced multiple simultaneous papers that claim to work best, which cannot all be true. In this work, we organize, benchmark, and analyze many Gaussian-splatting-based methods, providing apples-to-apples comparisons that prior works have lacked. We use multiple existing datasets and a new instructive synthetic dataset designed to isolate factors that affect reconstruction quality. We systematically categorize Gaussian splatting methods into specific motion representation types and quantify how their differences impact performance. Empirically, we find that their rank order is well-defined in synthetic data, but the complexity of real-world data currently overwhelms the differences. Furthermore, the fast rendering speed of all Gaussian-based methods comes at the cost of brittleness in optimization. We summarize our experiments into a list of findings that can help to further progress in this lively problem setting. Project Webpage: https://lynl7130.github.io/MonoDyGauBench.github.io/, Comment: 37 pages, 39 figures, 9 tables

Apparel is essential to human life, offering protection, mirroring cultural identities, and showcasing personal style. Yet, the creation of garments remains a time-consuming process, largely due to the manual work involved in designing them. To simplify this process, we introduce AIpparel, a large multimodal model for generating and editing sewing patterns. Our model fine-tunes state-of-the-art large multimodal models (LMMs) on a custom-curated large-scale dataset of over 120,000 unique garments, each with multimodal annotations including text, images, and sewing patterns. Additionally, we propose a novel tokenization scheme that concisely encodes these complex sewing patterns so that LLMs can learn to predict them efficiently. AIpparel achieves state-of-the-art performance in single-modal tasks, including text-to-garment and image-to-garment prediction, and enables novel multimodal garment generation applications such as interactive garment editing. The project website is at georgenakayama.github.io/AIpparel/., Comment: The project website is at georgenakayama.github.io/AIpparel/

Vision foundation models, particularly the ViT family, have revolutionized image understanding by providing rich semantic features. However, despite their success in 2D comprehension, their abilities on grasping 3D spatial relationships are still unclear. In this work, we evaluate and enhance the 3D awareness of ViT-based models. We begin by systematically assessing their ability to learn 3D equivariant features, specifically examining the consistency of semantic embeddings across different viewpoints. Our findings indicate that improved 3D equivariance leads to better performance on various downstream tasks, including pose estimation, tracking, and semantic transfer. Building on this insight, we propose a simple yet effective finetuning strategy based on 3D correspondences, which significantly enhances the 3D correspondence understanding of existing vision models. Remarkably, even finetuning on a single object for just one iteration results in substantial performance gains. All code and resources will be made publicly available to support further advancements in 3D-aware vision models. Our code is available at https://github.com/qq456cvb/3DCorrEnhance.

Text-to-image diffusion models produce impressive results but are frustrating tools for artists who desire fine-grained control. For example, a common use case is to create images of a specific instance in novel contexts, i.e., "identity-preserving generation". This setting, along with many other tasks (e.g., relighting), is a natural fit for image+text-conditional generative models. However, there is insufficient high-quality paired data to train such a model directly. We propose Diffusion Self-Distillation, a method for using a pre-trained text-to-image model to generate its own dataset for text-conditioned image-to-image tasks. We first leverage a text-to-image diffusion model's in-context generation ability to create grids of images and curate a large paired dataset with the help of a Visual-Language Model. We then fine-tune the text-to-image model into a text+image-to-image model using the curated paired dataset. We demonstrate that Diffusion Self-Distillation outperforms existing zero-shot methods and is competitive with per-instance tuning techniques on a wide range of identity-preservation generation tasks, without requiring test-time optimization., Comment: Project page: https://primecai.github.io/dsd/

Structure editors operate directly on a program's syntactic tree structure. At first glance, this allows for the exciting possibility that such an editor could enforce correctness properties: programs could be well-formed and sometimes even well-typed by construction. Unfortunately, traditional approaches to structure editing that attempt to rigidly enforce these properties face a seemingly fundamental problem, known in the literature as viscosity. Making changes to existing programs often requires temporarily breaking program structure -- but disallowing such changes makes it difficult to edit programs! In this paper, we present a scheme for structure editing which always maintains a valid program structure without sacrificing the fluidity necessary to freely edit programs. Two key pieces help solve this puzzle: first, we develop a novel generalization of selection for tree-based structures that properly generalizes text-based selection and editing, allowing users to freely rearrange pieces of code by cutting and pasting one-hole contexts; second, we type these one-hole contexts with a category of type diffs and explore the metatheory of the system that arises for maintaining well-typedness systematically. We implement our approach as an editor called Pantograph, and we conduct a study in which we successfully taught students to program with Pantograph and compare their performance against a traditional text editor., Comment: This is the extended version of the paper

Although vital for life on Earth, solar activity poses questions and increasing threats to humanity due to the Sun's unknown dynamics, intensified by our dependence on terrestrial and space-based infrastructure. This situation is compounded by significant gaps in our understanding of space weather phenomena, the Sun's magnetic field, and the need for rapid responses to unpredicted solar events. To address these issues, an optimized heliocentric satellite constellation is proposed that leverages satellites in an Elliptical Walker Constellation. This system offers (among others) equally distributed arguments of periapsis separations and cross-coupled true anomalies with respect to the Sun-centric coordinate frame. In this paper it is also demonstrated that this strategic multi-spacecraft configuration makes it possible to distinguish spatial and temporal changes in solar wind phenomena, reconstruct, in 3D, Coronal Mass Ejections (CMEs), predict which space or ground-based infrastructure and when it will be affected by CMEs, maintain continuous coverage of the critical Sun-Earth line throughout the mission's duration, and protect future missions by providing simultaneously in-situ and remote measurements from small and cost-effective satellites.

The cost of deploying vision transformers increasingly represents a barrier to wider industrial adoption. Existing compression requires additional end-to-end fine-tuning or incurs a significant drawback to runtime, thus making them ill-suited for online inference. We introduce the $\textbf{Visual Word Tokenizer}$ (VWT), a training-free method for reducing energy costs while retaining performance and runtime. The VWT groups patches (visual subwords) that are frequently used into visual words while infrequent ones remain intact. To do so, intra-image or inter-image statistics are leveraged to identify similar visual concepts for compression. Experimentally, we demonstrate a reduction in wattage of up to 19% with only a 20% increase in runtime at most. Comparative approaches of 8-bit quantization and token merging achieve a lower or similar energy efficiency but exact a higher toll on runtime (up to $2\times$ or more). Our results indicate that VWTs are well-suited for efficient online inference with a marginal compromise on performance.

Grasping by a robot in unstructured environments is deemed a critical challenge because of the requirement for effective adaptation to a wide variation in object geometries, material properties, and other environmental factors. In this paper, we propose a novel framework for robotic grasping based on the idea of compressing high-dimensional target and gripper features in a common latent space using a set of autoencoders. Our approach simplifies grasping by using three autoencoders dedicated to the target, the gripper, and a third one that fuses their latent representations. This allows the RL agent to achieve higher learning rates at the initial stages of exploration of a new environment, as well as at non-zero shot grasp attempts. The agent explores the latent space of the third autoencoder for better quality grasp without explicit reconstruction of objects. By implementing the PoWER algorithm into the RL training process, updates on the agent's policy will be made through the perturbation in the reward-weighted latent space. The successful exploration efficiently constrains both position and pose integrity for feasible executions of grasps. We evaluate our system on a diverse set of objects, demonstrating the high success rate in grasping with minimum computational overhead. We found that approach enhances the adaptation of the RL agent by more than 35 % in simulation experiments., Comment: Submitted for review at IEEE ICRA 2025

One-shot transfer of dexterous grasps to novel scenes with object and context variations has been a challenging problem. While distilled feature fields from large vision models have enabled semantic correspondences across 3D scenes, their features are point-based and restricted to object surfaces, limiting their capability of modeling complex semantic feature distributions for hand-object interactions. In this work, we propose the \textit{neural attention field} for representing semantic-aware dense feature fields in the 3D space by modeling inter-point relevance instead of individual point features. Core to it is a transformer decoder that computes the cross-attention between any 3D query point with all the scene points, and provides the query point feature with an attention-based aggregation. We further propose a self-supervised framework for training the transformer decoder from only a few 3D pointclouds without hand demonstrations. Post-training, the attention field can be applied to novel scenes for semantics-aware dexterous grasping from one-shot demonstration. Experiments show that our method provides better optimization landscapes by encouraging the end-effector to focus on task-relevant scene regions, resulting in significant improvements in success rates on real robots compared with the feature-field-based methods.

Multi-view image diffusion models have significantly advanced open-domain 3D object generation. However, most existing models rely on 2D network architectures that lack inherent 3D biases, resulting in compromised geometric consistency. To address this challenge, we introduce 3D-Adapter, a plug-in module designed to infuse 3D geometry awareness into pretrained image diffusion models. Central to our approach is the idea of 3D feedback augmentation: for each denoising step in the sampling loop, 3D-Adapter decodes intermediate multi-view features into a coherent 3D representation, then re-encodes the rendered RGBD views to augment the pretrained base model through feature addition. We study two variants of 3D-Adapter: a fast feed-forward version based on Gaussian splatting and a versatile training-free version utilizing neural fields and meshes. Our extensive experiments demonstrate that 3D-Adapter not only greatly enhances the geometry quality of text-to-multi-view models such as Instant3D and Zero123++, but also enables high-quality 3D generation using the plain text-to-image Stable Diffusion. Furthermore, we showcase the broad application potential of 3D-Adapter by presenting high quality results in text-to-3D, image-to-3D, text-to-texture, and text-to-avatar tasks., Comment: Project page: https://lakonik.github.io/3d-adapter/

Originating from the mathematical modelling of rainfall infiltration, we derive the solution of an initial-boundary value problem of a linear evolution partial differential equation, by using the Fokas method. We present numerical examples which correspond to specific physical rainfall problems. Based on this formalism we present an effective algorithm for the associated null-controllability problem, namely we numerically derive a family of boundary controls that steer the solution to the desired flat final state. Finally, a regularisation scheme allows the derivation of relatively small controls, in cases where this is necessary., Comment: 20 pages, 6 figures, 9 tables

We revisit the construction of the renormalized trace $\Theta$ of the Energy-Momentum tensor in the four-dimensional $\lambda\phi^4$ theory,using dimensional regularization in $d=4-\ve$ dimensions. We first construct several basic correlators such as $\braket{\phi^2 \phi\phi}$, $\braket{\phi^4 \phi \phi}$ to order $\lambda^2$ and from these the correlators $\braket{K_I \phi \phi}$ and $\braket{K_I K_J}$ with $K_I$ the basis of dimension $d$ operators. We then match the limit of their expressions on the Wilson-Fisher fixed point to the corresponding expressions obtained in Conformal Field Theory. Then, using the 3-point function $\braket{\Theta\phi\phi}$, we construct the operator $\Theta$ as a certain linear combination of the basis operators, using the requirements that $\Theta$ should vanish on the fixed point and that it should have zero anomalous dimension. Finally, we compute the 2-point function $\braket{\Theta\Theta}$ and we show that it obeys an eigenvalue equation that gives additional information about the internal structure of the Energy-Momentum tensor operator to what is already contained in its Callan-Symanzik equation., Comment: 84 pages, clarification comments added, typos corrected

Strongly lensed supernovae (SNe) are a rare class of transient that can offer tight cosmological constraints that are complementary to methods from other astronomical events. We present a follow-up study of one recently-discovered strongly lensed SN, the quadruply-imaged Type Ia SN 2022qmx (aka, "SN Zwicky") at z = 0.3544. We measure updated, template-subtracted photometry for SN Zwicky and derive improved time delays and magnifications. This is possible because SNe are transient, fading away after reaching their peak brightness. Specifically, we measure point spread function (PSF) photometry for all four images of SN Zwicky in three Hubble Space Telescope WFC3/UVIS passbands (F475W, F625W, F814W) and one WFC3/IR passband (F160W), with template images taken $\sim 11$ months after the epoch in which the SN images appear. We find consistency to within $2\sigma$ between lens model predicted time delays ($\lesssim1$ day), and measured time delays with HST colors ($\lesssim2$ days), including the uncertainty from chromatic microlensing that may arise from stars in the lensing galaxy. The standardizable nature of SNe Ia allows us to estimate absolute magnifications for the four images, with images A and C being elevated in magnification compared to lens model predictions by about $6\sigma$ and $3\sigma$ respectively, confirming previous work. We show that millilensing or differential dust extinction is unable to explain these discrepancies and find evidence for the existence of microlensing in images A, C, and potentially D, that may contribute to the anomalous magnification., Comment: Submitted to ApJ

Generalizable long-horizon robotic assembly requires reasoning at multiple levels of abstraction. End-to-end imitation learning (IL) has been proven a promising approach, but it requires a large amount of demonstration data for training and often fails to meet the high-precision requirement of assembly tasks. Reinforcement Learning (RL) approaches have succeeded in high-precision assembly tasks, but suffer from sample inefficiency and hence, are less competent at long-horizon tasks. To address these challenges, we propose a hierarchical modular approach, named ARCH (Adaptive Robotic Composition Hierarchy), which enables long-horizon high-precision assembly in contact-rich settings. ARCH employs a hierarchical planning framework, including a low-level primitive library of continuously parameterized skills and a high-level policy. The low-level primitive library includes essential skills for assembly tasks, such as grasping and inserting. These primitives consist of both RL and model-based controllers. The high-level policy, learned via imitation learning from a handful of demonstrations, selects the appropriate primitive skills and instantiates them with continuous input parameters. We extensively evaluate our approach on a real robot manipulation platform. We show that while trained on a single task, ARCH generalizes well to unseen tasks and outperforms baseline methods in terms of success rate and data efficiency. Videos can be found at https://long-horizon-assembly.github.io.

This paper presents a method to leverage arbitrary neural network architecture for control variates. Control variates are crucial in reducing the variance of Monte Carlo integration, but they hinge on finding a function that both correlates with the integrand and has a known analytical integral. Traditional approaches rely on heuristics to choose this function, which might not be expressive enough to correlate well with the integrand. Recent research alleviates this issue by modeling the integrands with a learnable parametric model, such as a neural network. However, the challenge remains in creating an expressive parametric model with a known analytical integral. This paper proposes a novel approach to construct learnable parametric control variates functions from arbitrary neural network architectures. Instead of using a network to approximate the integrand directly, we employ the network to approximate the anti-derivative of the integrand. This allows us to use automatic differentiation to create a function whose integration can be constructed by the antiderivative network. We apply our method to solve partial differential equations using the Walk-on-sphere algorithm. Our results indicate that this approach is unbiased and uses various network architectures to achieve lower variance than other control variate methods.

Depth sensing is an important problem for 3D vision-based robotics. Yet, a real-world active stereo or ToF depth camera often produces noisy and incomplete depth which bottlenecks robot performances. In this work, we propose D3RoMa, a learning-based depth estimation framework on stereo image pairs that predicts clean and accurate depth in diverse indoor scenes, even in the most challenging scenarios with translucent or specular surfaces where classical depth sensing completely fails. Key to our method is that we unify depth estimation and restoration into an image-to-image translation problem by predicting the disparity map with a denoising diffusion probabilistic model. At inference time, we further incorporated a left-right consistency constraint as classifier guidance to the diffusion process. Our framework combines recently advanced learning-based approaches and geometric constraints from traditional stereo vision. For model training, we create a large scene-level synthetic dataset with diverse transparent and specular objects to compensate for existing tabletop datasets. The trained model can be directly applied to real-world in-the-wild scenes and achieve state-of-the-art performance in multiple public depth estimation benchmarks. Further experiments in real environments show that accurate depth prediction significantly improves robotic manipulation in various scenarios.

Symmetries are ubiquitous across all kinds of objects, whether in nature or in man-made creations. While these symmetries may seem intuitive to the human eye, detecting them with a machine is nontrivial due to the vast search space. Classical geometry-based methods work by aggregating "votes" for each symmetry but struggle with noise. In contrast, learning-based methods may be more robust to noise, but often overlook partial symmetries due to the scarcity of annotated data. In this work, we address this challenge by proposing a novel symmetry detection method that marries classical symmetry detection techniques with recent advances in generative modeling. Specifically, we apply Langevin dynamics to a redefined symmetry space to enhance robustness against noise. We provide empirical results on a variety of shapes that suggest our method is not only robust to noise, but can also identify both partial and global symmetries. Moreover, we demonstrate the utility of our detected symmetries in various downstream tasks, such as compression and symmetrization of noisy shapes., Comment: Project page: https://symmetry-langevin.github.io/

To advance the state of the art in the creation of 3D foundation models, this paper introduces the ConDense framework for 3D pre-training utilizing existing pre-trained 2D networks and large-scale multi-view datasets. We propose a novel 2D-3D joint training scheme to extract co-embedded 2D and 3D features in an end-to-end pipeline, where 2D-3D feature consistency is enforced through a volume rendering NeRF-like ray marching process. Using dense per pixel features we are able to 1) directly distill the learned priors from 2D models to 3D models and create useful 3D backbones, 2) extract more consistent and less noisy 2D features, 3) formulate a consistent embedding space where 2D, 3D, and other modalities of data (e.g., natural language prompts) can be jointly queried. Furthermore, besides dense features, ConDense can be trained to extract sparse features (e.g., key points), also with 2D-3D consistency -- condensing 3D NeRF representations into compact sets of decorated key points. We demonstrate that our pre-trained model provides good initialization for various 3D tasks including 3D classification and segmentation, outperforming other 3D pre-training methods by a significant margin. It also enables, by exploiting our sparse features, additional useful downstream tasks, such as matching 2D images to 3D scenes, detecting duplicate 3D scenes, and querying a repository of 3D scenes through natural language -- all quite efficiently and without any per-scene fine-tuning., Comment: ECCV 2024

Interactable objects are ubiquitous in our daily lives. Recent advances in 3D generative models make it possible to automate the modeling of these objects, benefiting a range of applications from 3D printing to the creation of robot simulation environments. However, while significant progress has been made in modeling 3D shapes and appearances, modeling object physics, particularly for interactable objects, remains challenging due to the physical constraints imposed by inter-part motions. In this paper, we tackle the problem of physically plausible part completion for interactable objects, aiming to generate 3D parts that not only fit precisely into the object but also allow smooth part motions. To this end, we propose a diffusion-based part generation model that utilizes geometric conditioning through classifier-free guidance and formulates physical constraints as a set of stability and mobility losses to guide the sampling process. Additionally, we demonstrate the generation of dependent parts, paving the way toward sequential part generation for objects with complex part-whole hierarchies. Experimentally, we introduce a new metric for measuring physical plausibility based on motion success rates. Our model outperforms existing baselines over shape and physical metrics, especially those that do not adequately model physical constraints. We also demonstrate our applications in 3D printing, robot manipulation, and sequential part generation, showing our strength in realistic tasks with the demand for high physical plausibility., Comment: ICRA 2025

Recognizing and disentangling visual attributes from objects is a foundation to many computer vision applications. While large vision language representations like CLIP had largely resolved the task of zero-shot object recognition, zero-shot visual attribute recognition remains a challenge because CLIP's contrastively-learned vision-language representation cannot effectively capture object-attribute dependencies. In this paper, we target this weakness and propose a sentence generation-based retrieval formulation for attribute recognition that is novel in 1) explicitly modeling a to-be-measured and retrieved object-attribute relation as a conditional probability graph, which converts the recognition problem into a dependency-sensitive language-modeling problem, and 2) applying a large pretrained Vision-Language Model (VLM) on this reformulation and naturally distilling its knowledge of image-object-attribute relations to use towards attribute recognition. Specifically, for each attribute to be recognized on an image, we measure the visual-conditioned probability of generating a short sentence encoding the attribute's relation to objects on the image. Unlike contrastive retrieval, which measures likelihood by globally aligning elements of the sentence to the image, generative retrieval is sensitive to the order and dependency of objects and attributes in the sentence. We demonstrate through experiments that generative retrieval consistently outperforms contrastive retrieval on two visual reasoning datasets, Visual Attribute in the Wild (VAW), and our newly-proposed Visual Genome Attribute Ranking (VGARank)., Comment: Accepted at ECCV 2024. Contact: zhuwilliam[at]google[dot]com. GitHub: https://github.com/google-research/google-research/tree/master/attribute_with_prefixlm

Reverse engineering 3D computer-aided design (CAD) models from images is an important task for many downstream applications including interactive editing, manufacturing, architecture, robotics, etc. The difficulty of the task lies in vast representational disparities between the CAD output and the image input. CAD models are precise, programmatic constructs that involves sequential operations combining discrete command structure with continuous attributes -- making it challenging to learn and optimize in an end-to-end fashion. Concurrently, input images introduce inherent challenges such as photo-metric variability and sensor noise, complicating the reverse engineering process. In this work, we introduce a novel approach that conditionally factorizes the task into two sub-problems. First, we leverage large foundation models, particularly GPT-4V, to predict the global discrete base structure with semantic information. Second, we propose TrAssembler that conditioned on the discrete structure with semantics predicts the continuous attribute values. To support the training of our TrAssembler, we further constructed an annotated CAD dataset of common objects from ShapeNet. Putting all together, our approach and data demonstrate significant first steps towards CAD-ifying images in the wild. Our project page: https://anonymous123342.github.io/

We present a method for generating Streetscapes-long sequences of views through an on-the-fly synthesized city-scale scene. Our generation is conditioned by language input (e.g., city name, weather), as well as an underlying map/layout hosting the desired trajectory. Compared to recent models for video generation or 3D view synthesis, our method can scale to much longer-range camera trajectories, spanning several city blocks, while maintaining visual quality and consistency. To achieve this goal, we build on recent work on video diffusion, used within an autoregressive framework that can easily scale to long sequences. In particular, we introduce a new temporal imputation method that prevents our autoregressive approach from drifting from the distribution of realistic city imagery. We train our Streetscapes system on a compelling source of data-posed imagery from Google Street View, along with contextual map data-which allows users to generate city views conditioned on any desired city layout, with controllable camera poses. Please see more results at our project page at https://boyangdeng.com/streetscapes., Comment: *Equal Contributions; Fixed few duplicated references from 1st upload; Project Page: https://boyangdeng.com/streetscapes

The increased demand for tools that automate the 3D content creation process led to tremendous progress in deep generative models that can generate diverse 3D objects of high fidelity. In this paper, we present PASTA, an autoregressive transformer architecture for generating high quality 3D shapes. PASTA comprises two main components: An autoregressive transformer that generates objects as a sequence of cuboidal primitives and a blending network, implemented with a transformer decoder that composes the sequences of cuboids and synthesizes high quality meshes for each object. Our model is trained in two stages: First we train our autoregressive generative model using only annotated cuboidal parts as supervision and next, we train our blending network using explicit 3D supervision, in the form of watertight meshes. Evaluations on various ShapeNet objects showcase the ability of our model to perform shape generation from diverse inputs \eg from scratch, from a partial object, from text and images, as well size-guided generation, by explicitly conditioning on a bounding box that defines the object's boundaries. Moreover, as our model considers the underlying part-based structure of a 3D object, we are able to select a specific part and produce shapes with meaningful variations of this part. As evidenced by our experiments, our model generates 3D shapes that are both more realistic and diverse than existing part-based and non part-based methods, while at the same time is simpler to implement and train.

Estimating the difficulty of multiple-choice questions would be great help for educators who must spend substantial time creating and piloting stimuli for their tests, and for learners who want to practice. Supervised approaches to difficulty estimation have yielded to date mixed results. In this contribution we leverage an aspect of generative large models which might be seen as a weakness when answering questions, namely their uncertainty, and exploit it towards exploring correlations between two different metrics of uncertainty, and the actual student response distribution. While we observe some present but weak correlations, we also discover that the models' behaviour is different in the case of correct vs wrong answers, and that correlations differ substantially according to the different question types which are included in our fine-grained, previously unused dataset of 451 questions from a Biopsychology course. In discussing our findings, we also suggest potential avenues to further leverage model uncertainty as an additional proxy for item difficulty., Comment: 13 pages, 7 figures, Published in The 31st International Conference on Computational Linguistics, available in the ACL Anthology: https://aclanthology.org/2025.coling-main.749/

This work proposes a retrieve-and-transfer framework for zero-shot robotic manipulation, dubbed RAM, featuring generalizability across various objects, environments, and embodiments. Unlike existing approaches that learn manipulation from expensive in-domain demonstrations, RAM capitalizes on a retrieval-based affordance transfer paradigm to acquire versatile manipulation capabilities from abundant out-of-domain data. First, RAM extracts unified affordance at scale from diverse sources of demonstrations including robotic data, human-object interaction (HOI) data, and custom data to construct a comprehensive affordance memory. Then given a language instruction, RAM hierarchically retrieves the most similar demonstration from the affordance memory and transfers such out-of-domain 2D affordance to in-domain 3D executable affordance in a zero-shot and embodiment-agnostic manner. Extensive simulation and real-world evaluations demonstrate that our RAM consistently outperforms existing works in diverse daily tasks. Additionally, RAM shows significant potential for downstream applications such as automatic and efficient data collection, one-shot visual imitation, and LLM/VLM-integrated long-horizon manipulation. For more details, please check our website at https://yxkryptonite.github.io/RAM/.

3D Gaussian Splatting (3DGS) is a promising technique for 3D reconstruction, offering efficient training and rendering speeds, making it suitable for real-time applications.However, current methods require highly controlled environments (no moving people or wind-blown elements, and consistent lighting) to meet the inter-view consistency assumption of 3DGS. This makes reconstruction of real-world captures problematic. We present SpotLessSplats, an approach that leverages pre-trained and general-purpose features coupled with robust optimization to effectively ignore transient distractors. Our method achieves state-of-the-art reconstruction quality both visually and quantitatively, on casual captures. Additional results available at: https://spotlesssplats.github.io

Gaussian splatting has become a popular representation for novel-view synthesis, exhibiting clear strengths in efficiency, photometric quality, and compositional edibility. Following its success, many works have extended Gaussians to 4D, showing that dynamic Gaussians maintain these benefits while also tracking scene geometry far better than alternative representations. Yet, these methods assume dense multi-view videos as supervision. In this work, we are interested in extending the capability of Gaussian scene representations to casually captured monocular videos. We show that existing 4D Gaussian methods dramatically fail in this setup because the monocular setting is underconstrained. Building off this finding, we propose a method we call Dynamic Gaussian Marbles, which consist of three core modifications that target the difficulties of the monocular setting. First, we use isotropic Gaussian "marbles'', reducing the degrees of freedom of each Gaussian. Second, we employ a hierarchical divide and-conquer learning strategy to efficiently guide the optimization towards solutions with globally coherent motion. Finally, we add image-level and geometry-level priors into the optimization, including a tracking loss that takes advantage of recent progress in point tracking. By constraining the optimization, Dynamic Gaussian Marbles learns Gaussian trajectories that enable novel-view rendering and accurately capture the 3D motion of the scene elements. We evaluate on the Nvidia Dynamic Scenes dataset and the DyCheck iPhone dataset, and show that Gaussian Marbles significantly outperforms other Gaussian baselines in quality, and is on-par with non-Gaussian representations, all while maintaining the efficiency, compositionality, editability, and tracking benefits of Gaussians. Our project page can be found here https://geometry.stanford.edu/projects/dynamic-gaussian-marbles.github.io/.

aeon is a unified Python 3 library for all machine learning tasks involving time series. The package contains modules for time series forecasting, classification, extrinsic regression and clustering, as well as a variety of utilities, transformations and distance measures designed for time series data. aeon also has a number of experimental modules for tasks such as anomaly detection, similarity search and segmentation. aeon follows the scikit-learn API as much as possible to help new users and enable easy integration of aeon estimators with useful tools such as model selection and pipelines. It provides a broad library of time series algorithms, including efficient implementations of the very latest advances in research. Using a system of optional dependencies, aeon integrates a wide variety of packages into a single interface while keeping the core framework with minimal dependencies. The package is distributed under the 3-Clause BSD license and is available at https://github.com/ aeon-toolkit/aeon. This version was submitted to the JMLR journal on 02 Nov 2023 for v0.5.0 of aeon. At the time of this preprint aeon has released v0.9.0, and has had substantial changes., Comment: 10 pages

Code Language Models have been trained to generate accurate solutions, typically with no regard for runtime. On the other hand, previous works that explored execution optimisation have observed corresponding drops in functional correctness. To that end, we introduce Code-Optimise, a framework that incorporates both correctness (passed, failed) and runtime (quick, slow) as learning signals via self-generated preference data. Our framework is both lightweight and robust as it dynamically selects solutions to reduce overfitting while avoiding a reliance on larger models for learning signals. Code-Optimise achieves significant improvements in pass@k while decreasing the competitive baseline runtimes by an additional 6% for in-domain data and up to 3% for out-of-domain data. As a by-product, the average length of the generated solutions is reduced by up to 48% on MBPP and 23% on HumanEval, resulting in faster and cheaper inference. The generated data and codebase is open-sourced at https://github.com/huawei-noah/HEBO/tree/Code_Optimise., Comment: NAACL 2025 (Findings)

3D Gaussians, as a low-level scene representation, typically involve thousands to millions of Gaussians. This makes it difficult to control the scene in ways that reflect the underlying dynamic structure, where the number of independent entities is typically much smaller. In particular, it can be challenging to animate and move objects in the scene, which requires coordination among many Gaussians. To address this issue, we develop a mutual information shaping technique that enforces movement resonance between correlated Gaussians in a motion network. Such correlations can be learned from putative 2D object masks in different views. By approximating the mutual information with the Jacobians of the motions, our method ensures consistent movements of the Gaussians composing different objects under various perturbations. In particular, we develop an efficient contrastive training pipeline with lightweight optimization to shape the motion network, avoiding the need for re-shaping throughout the motion sequence. Notably, our training only touches a small fraction of all Gaussians in the scene yet attains the desired compositional behavior according to the underlying dynamic structure. The proposed technique is evaluated on challenging scenes and demonstrates significant performance improvement in promoting consistent movements and 3D object segmentation while inducing low computation and memory requirements.

Large-scale vision foundation models such as Segment Anything (SAM) demonstrate impressive performance in zero-shot image segmentation at multiple levels of granularity. However, these zero-shot predictions are rarely 3D-consistent. As the camera viewpoint changes in a scene, so do the segmentation predictions, as well as the characterizations of "coarse" or "fine" granularity. In this work, we address the challenging task of lifting multi-granular and view-inconsistent image segmentations into a hierarchical and 3D-consistent representation. We learn a novel feature field within a Neural Radiance Field (NeRF) representing a 3D scene, whose segmentation structure can be revealed at different scales by simply using different thresholds on feature distance. Our key idea is to learn an ultrametric feature space, which unlike a Euclidean space, exhibits transitivity in distance-based grouping, naturally leading to a hierarchical clustering. Put together, our method takes view-inconsistent multi-granularity 2D segmentations as input and produces a hierarchy of 3D-consistent segmentations as output. We evaluate our method and several baselines on synthetic datasets with multi-view images and multi-granular segmentation, showcasing improved accuracy and viewpoint-consistency. We additionally provide qualitative examples of our model's 3D hierarchical segmentations in real world scenes. The code and dataset are available at https://github.com/hardyho/ultrametric_feature_fields

We introduce 4D Motion Scaffolds (MoSca), a modern 4D reconstruction system designed to reconstruct and synthesize novel views of dynamic scenes from monocular videos captured casually in the wild. To address such a challenging and ill-posed inverse problem, we leverage prior knowledge from foundational vision models and lift the video data to a novel Motion Scaffold (MoSca) representation, which compactly and smoothly encodes the underlying motions/deformations. The scene geometry and appearance are then disentangled from the deformation field and are encoded by globally fusing the Gaussians anchored onto the MoSca and optimized via Gaussian Splatting. Additionally, camera focal length and poses can be solved using bundle adjustment without the need of any other pose estimation tools. Experiments demonstrate state-of-the-art performance on dynamic rendering benchmarks and its effectiveness on real videos., Comment: project page: https://www.cis.upenn.edu/~leijh/projects/mosca code release: https://github.com/JiahuiLei/MoSca

Research on video generation has recently made tremendous progress, enabling high-quality videos to be generated from text prompts or images. Adding control to the video generation process is an important goal moving forward and recent approaches that condition video generation models on camera trajectories make strides towards it. Yet, it remains challenging to generate a video of the same scene from multiple different camera trajectories. Solutions to this multi-video generation problem could enable large-scale 3D scene generation with editable camera trajectories, among other applications. We introduce collaborative video diffusion (CVD) as an important step towards this vision. The CVD framework includes a novel cross-video synchronization module that promotes consistency between corresponding frames of the same video rendered from different camera poses using an epipolar attention mechanism. Trained on top of a state-of-the-art camera-control module for video generation, CVD generates multiple videos rendered from different camera trajectories with significantly better consistency than baselines, as shown in extensive experiments. Project page: https://collaborativevideodiffusion.github.io/.

The in-plane anisotropic properties of two-dimensional (2D) group IV monochalcogenides provide an additional degree of freedom which can be used in future optoelectronic devices. Here, it is shown that the third harmonic generation (THG) signal produced by ultrathin tin (II) sulfide (SnS) is in-plane anisotropic with respect to the incident linear polarization of the laser field. We fit the experimental polarization-resolved THG (P-THG) measurements with a nonlinear optics model, which accounts for the orthorhombic crystal structure of 2D SnS. We calculate the relative magnitudes of the \{chi}^(3) tensor components by recording and simultaneously fitting both orthogonal components of the P-THG intensity. Furthermore, we introduce a THG anisotropy ratio, whose calculated values compare the total THG intensity when the excitation linear polarization is along the armchair crystallographic direction with the case when it is along the zigzag direction. Our results provide quantitative information on the anisotropic nature of the THG process in SnS, paving the way to a better understanding of anisotropic nonlinear light-matter interactions, and the development of polarization-sensitive nonlinear optical devices.

Graphics design is important for various applications, including movie production and game design. To create a high-quality scene, designers usually need to spend hours in software like Blender, in which they might need to interleave and repeat operations, such as connecting material nodes, hundreds of times. Moreover, slightly different design goals may require completely different sequences, making automation difficult. In this paper, we propose a system that leverages Vision-Language Models (VLMs), like GPT-4V, to intelligently search the design action space to arrive at an answer that can satisfy a user's intent. Specifically, we design a vision-based edit generator and state evaluator to work together to find the correct sequence of actions to achieve the goal. Inspired by the role of visual imagination in the human design process, we supplement the visual reasoning capabilities of VLMs with "imagined" reference images from image-generation models, providing visual grounding of abstract language descriptions. In this paper, we provide empirical evidence suggesting our system can produce simple but tedious Blender editing sequences for tasks such as editing procedural materials and geometry from text and/or reference images, as well as adjusting lighting configurations for product renderings in complex scenes.

We introduce MultiPhys, a method designed for recovering multi-person motion from monocular videos. Our focus lies in capturing coherent spatial placement between pairs of individuals across varying degrees of engagement. MultiPhys, being physically aware, exhibits robustness to jittering and occlusions, and effectively eliminates penetration issues between the two individuals. We devise a pipeline in which the motion estimated by a kinematic-based method is fed into a physics simulator in an autoregressive manner. We introduce distinct components that enable our model to harness the simulator's properties without compromising the accuracy of the kinematic estimates. This results in final motion estimates that are both kinematically coherent and physically compliant. Extensive evaluations on three challenging datasets characterized by substantial inter-person interaction show that our method significantly reduces errors associated with penetration and foot skating, while performing competitively with the state-of-the-art on motion accuracy and smoothness. Results and code can be found on our project page (http://www.iri.upc.edu/people/nugrinovic/multiphys/).

Recent advances in large-scale pretraining have yielded visual foundation models with strong capabilities. Not only can recent models generalize to arbitrary images for their training task, their intermediate representations are useful for other visual tasks such as detection and segmentation. Given that such models can classify, delineate, and localize objects in 2D, we ask whether they also represent their 3D structure? In this work, we analyze the 3D awareness of visual foundation models. We posit that 3D awareness implies that representations (1) encode the 3D structure of the scene and (2) consistently represent the surface across views. We conduct a series of experiments using task-specific probes and zero-shot inference procedures on frozen features. Our experiments reveal several limitations of the current models. Our code and analysis can be found at https://github.com/mbanani/probe3d., Comment: Accepted to CVPR 2024. Project page: https://github.com/mbanani/probe3d

Modeling and rendering photorealistic avatars is of crucial importance in many applications. Existing methods that build a 3D avatar from visual observations, however, struggle to reconstruct clothed humans. We introduce PhysAvatar, a novel framework that combines inverse rendering with inverse physics to automatically estimate the shape and appearance of a human from multi-view video data along with the physical parameters of the fabric of their clothes. For this purpose, we adopt a mesh-aligned 4D Gaussian technique for spatio-temporal mesh tracking as well as a physically based inverse renderer to estimate the intrinsic material properties. PhysAvatar integrates a physics simulator to estimate the physical parameters of the garments using gradient-based optimization in a principled manner. These novel capabilities enable PhysAvatar to create high-quality novel-view renderings of avatars dressed in loose-fitting clothes under motions and lighting conditions not seen in the training data. This marks a significant advancement towards modeling photorealistic digital humans using physically based inverse rendering with physics in the loop. Our project website is at: https://qingqing-zhao.github.io/PhysAvatar, Comment: Project Page: https://qingqing-zhao.github.io/PhysAvatar

We address the problem of building digital twins of unknown articulated objects from two RGBD scans of the object at different articulation states. We decompose the problem into two stages, each addressing distinct aspects. Our method first reconstructs object-level shape at each state, then recovers the underlying articulation model including part segmentation and joint articulations that associate the two states. By explicitly modeling point-level correspondences and exploiting cues from images, 3D reconstructions, and kinematics, our method yields more accurate and stable results compared to prior work. It also handles more than one movable part and does not rely on any object shape or structure priors. Project page: https://github.com/NVlabs/DigitalTwinArt, Comment: CVPR 2024

Two-dimensional (2D) graphene and graphene-related materials (GRMs) show great promise for future electronic devices. Nevertheless, GRMs result distinct properties under the influence of the substrate that serves as support through uneven compression/ elongation of GRMs surface atoms. Strain in GRM monolayers is the most common feature that alters the interatomic distances, band structure, providing a new degree of freedom that allows regulation of their electronic properties and introducing the field of straintronics. Having an all-optical detection, a minimally invasive tool that rapidly probes strain in large areas of GRM monolayers, would be of great importance in the research and development of novel 2D devices. Here, we use Polarization-resolved Second Harmonic Generation (P-SHG) optical imaging to identify strain distribution, induced in a single layer of WS2 placed on a pre-patterned Si/SiO2 substrate with cylindrical wells. By fitting the P-SHG data pixel-by-pixel, we produce spatially resolved images of the crystal armchair direction. In regions where the WS2 monolayer conforms to the pattern topography, a distinct cross-shaped pattern is evident in the armchair image owing to strain. The presence of strain in these regions is independently confirmed using a combination of atomic force microscopy and Raman mapping.

To reduce the inference cost of large language models, model compression is increasingly used to create smaller scalable models. However, little is known about their robustness to minority subgroups defined by the labels and attributes of a dataset. In this paper, we investigate the effects of 18 different compression methods and settings on the subgroup robustness of BERT language models. We show that worst-group performance does not depend on model size alone, but also on the compression method used. Additionally, we find that model compression does not always worsen the performance on minority subgroups. Altogether, our analysis serves to further research into the subgroup robustness of model compression., Comment: The 2023 Conference on Empirical Methods in Natural Language Processing (EMNLP 2023)

Correspondences emerge from large-scale vision models trained for generative and discriminative tasks. This has been revealed and benchmarked by computing correspondence maps between pairs of images, using nearest neighbors on the feature grids. Existing work has attempted to improve the quality of these correspondence maps by carefully mixing features from different sources, such as by combining the features of different layers or networks. We point out that a better correspondence strategy is available, which directly imposes structure on the correspondence field: the functional map. Wielding this simple mathematical tool, we lift the correspondence problem from the pixel space to the function space and directly optimize for mappings that are globally coherent. We demonstrate that our technique yields correspondences that are not only smoother but also more accurate, with the possibility of better reflecting the knowledge embedded in the large-scale vision models that we are studying. Our approach sets a new state-of-the-art on various dense correspondence tasks. We also demonstrate our effectiveness in keypoint correspondence and affordance map transfer.