Hao Su

Ruoxi Shi, Hansheng Chen, Zhuoyang Zhang et al. · stanford

We report Zero123++, an image-conditioned diffusion model for generating 3D-consistent multi-view images from a single input view. To take full advantage of pretrained 2D generative priors, we develop various conditioning and training schemes to minimize the effort of finetuning from off-the-shelf image diffusion models such as Stable Diffusion. Zero123++ excels in producing high-quality, consistent multi-view images from a single image, overcoming common issues like texture degradation and geometric misalignment. Furthermore, we showcase the feasibility of training a ControlNet on Zero123++ for enhanced control over the generation process. The code is available at https://github.com/SUDO-AI-3D/zero123plus.

Hansheng Chen, Jiatao Gu, Anpei Chen et al. · eth-zurich, meta-ai

3D-aware image synthesis encompasses a variety of tasks, such as scene generation and novel view synthesis from images. Despite numerous task-specific methods, developing a comprehensive model remains challenging. In this paper, we present SSDNeRF, a unified approach that employs an expressive diffusion model to learn a generalizable prior of neural radiance fields (NeRF) from multi-view images of diverse objects. Previous studies have used two-stage approaches that rely on pretrained NeRFs as real data to train diffusion models. In contrast, we propose a new single-stage training paradigm with an end-to-end objective that jointly optimizes a NeRF auto-decoder and a latent diffusion model, enabling simultaneous 3D reconstruction and prior learning, even from sparsely available views. At test time, we can directly sample the diffusion prior for unconditional generation, or combine it with arbitrary observations of unseen objects for NeRF reconstruction. SSDNeRF demonstrates robust results comparable to or better than leading task-specific methods in unconditional generation and single/sparse-view 3D reconstruction.

Yuzhe Qin, Wei Yang, Binghao Huang et al. · nvidia

Vision-based teleoperation offers the possibility to endow robots with human-level intelligence to physically interact with the environment, while only requiring low-cost camera sensors. However, current vision-based teleoperation systems are designed and engineered towards a particular robot model and deploy environment, which scales poorly as the pool of the robot models expands and the variety of the operating environment increases. In this paper, we propose AnyTeleop, a unified and general teleoperation system to support multiple different arms, hands, realities, and camera configurations within a single system. Although being designed to provide great flexibility to the choice of simulators and real hardware, our system can still achieve great performance. For real-world experiments, AnyTeleop can outperform a previous system that was designed for a specific robot hardware with a higher success rate, using the same robot. For teleoperation in simulation, AnyTeleop leads to better imitation learning performance, compared with a previous system that is particularly designed for that simulator. Project page: https://yzqin.github.io/anyteleop/.

Anpei Chen, Zexiang Xu, Xinyue Wei et al. · eth-zurich

We present Factor Fields, a novel framework for modeling and representing signals. Factor Fields decomposes a signal into a product of factors, each represented by a classical or neural field representation which operates on transformed input coordinates. This decomposition results in a unified framework that accommodates several recent signal representations including NeRF, Plenoxels, EG3D, Instant-NGP, and TensoRF. Additionally, our framework allows for the creation of powerful new signal representations, such as the "Dictionary Field" (DiF) which is a second contribution of this paper. Our experiments show that DiF leads to improvements in approximation quality, compactness, and training time when compared to previous fast reconstruction methods. Experimentally, our representation achieves better image approximation quality on 2D image regression tasks, higher geometric quality when reconstructing 3D signed distance fields, and higher compactness for radiance field reconstruction tasks. Furthermore, DiF enables generalization to unseen images/3D scenes by sharing bases across signals during training which greatly benefits use cases such as image regression from sparse observations and few-shot radiance field reconstruction.

Minghua Liu, Ruoxi Shi, Linghao Chen et al. · stanford

Recent advancements in open-world 3D object generation have been remarkable, with image-to-3D methods offering superior fine-grained control over their text-to-3D counterparts. However, most existing models fall short in simultaneously providing rapid generation speeds and high fidelity to input images - two features essential for practical applications. In this paper, we present One-2-3-45++, an innovative method that transforms a single image into a detailed 3D textured mesh in approximately one minute. Our approach aims to fully harness the extensive knowledge embedded in 2D diffusion models and priors from valuable yet limited 3D data. This is achieved by initially finetuning a 2D diffusion model for consistent multi-view image generation, followed by elevating these images to 3D with the aid of multi-view conditioned 3D native diffusion models. Extensive experimental evaluations demonstrate that our method can produce high-quality, diverse 3D assets that closely mirror the original input image. Our project webpage: https://sudo-ai-3d.github.io/One2345plus_page.

Minghua Liu, Yinhao Zhu, Hong Cai et al.

Generalizable 3D part segmentation is important but challenging in vision and robotics. Training deep models via conventional supervised methods requires large-scale 3D datasets with fine-grained part annotations, which are costly to collect. This paper explores an alternative way for low-shot part segmentation of 3D point clouds by leveraging a pretrained image-language model, GLIP, which achieves superior performance on open-vocabulary 2D detection. We transfer the rich knowledge from 2D to 3D through GLIP-based part detection on point cloud rendering and a novel 2D-to-3D label lifting algorithm. We also utilize multi-view 3D priors and few-shot prompt tuning to boost performance significantly. Extensive evaluation on PartNet and PartNet-Mobility datasets shows that our method enables excellent zero-shot 3D part segmentation. Our few-shot version not only outperforms existing few-shot approaches by a large margin but also achieves highly competitive results compared to the fully supervised counterpart. Furthermore, we demonstrate that our method can be directly applied to iPhone-scanned point clouds without significant domain gaps.

Xiaoshuai Zhang, Abhijit Kundu, Thomas Funkhouser et al.

We address efficient and structure-aware 3D scene representation from images. Nerflets are our key contribution -- a set of local neural radiance fields that together represent a scene. Each nerflet maintains its own spatial position, orientation, and extent, within which it contributes to panoptic, density, and radiance reconstructions. By leveraging only photometric and inferred panoptic image supervision, we can directly and jointly optimize the parameters of a set of nerflets so as to form a decomposed representation of the scene, where each object instance is represented by a group of nerflets. During experiments with indoor and outdoor environments, we find that nerflets: (1) fit and approximate the scene more efficiently than traditional global NeRFs, (2) allow the extraction of panoptic and photometric renderings from arbitrary views, and (3) enable tasks rare for NeRFs, such as 3D panoptic segmentation and interactive editing.

Haian Jin, Isabella Liu, Peijia Xu et al.

We propose TensoIR, a novel inverse rendering approach based on tensor factorization and neural fields. Unlike previous works that use purely MLP-based neural fields, thus suffering from low capacity and high computation costs, we extend TensoRF, a state-of-the-art approach for radiance field modeling, to estimate scene geometry, surface reflectance, and environment illumination from multi-view images captured under unknown lighting conditions. Our approach jointly achieves radiance field reconstruction and physically-based model estimation, leading to photo-realistic novel view synthesis and relighting results. Benefiting from the efficiency and extensibility of the TensoRF-based representation, our method can accurately model secondary shading effects (like shadows and indirect lighting) and generally support input images captured under single or multiple unknown lighting conditions. The low-rank tensor representation allows us to not only achieve fast and compact reconstruction but also better exploit shared information under an arbitrary number of capturing lighting conditions. We demonstrate the superiority of our method to baseline methods qualitatively and quantitatively on various challenging synthetic and real-world scenes.

Xiaoshuai Zhang, Sai Bi, Kalyan Sunkavalli et al.

While NeRF has shown great success for neural reconstruction and rendering, its limited MLP capacity and long per-scene optimization times make it challenging to model large-scale indoor scenes. In contrast, classical 3D reconstruction methods can handle large-scale scenes but do not produce realistic renderings. We propose NeRFusion, a method that combines the advantages of NeRF and TSDF-based fusion techniques to achieve efficient large-scale reconstruction and photo-realistic rendering. We process the input image sequence to predict per-frame local radiance fields via direct network inference. These are then fused using a novel recurrent neural network that incrementally reconstructs a global, sparse scene representation in real-time at 22 fps. This global volume can be further fine-tuned to boost rendering quality. We demonstrate that NeRFusion achieves state-of-the-art quality on both large-scale indoor and small-scale object scenes, with substantially faster reconstruction than NeRF and other recent methods.

Jiayuan Gu, Sean Kirmani, Paul Wohlhart et al.

Generalization remains one of the most important desiderata for robust robot learning systems. While recently proposed approaches show promise in generalization to novel objects, semantic concepts, or visual distribution shifts, generalization to new tasks remains challenging. For example, a language-conditioned policy trained on pick-and-place tasks will not be able to generalize to a folding task, even if the arm trajectory of folding is similar to pick-and-place. Our key insight is that this kind of generalization becomes feasible if we represent the task through rough trajectory sketches. We propose a policy conditioning method using such rough trajectory sketches, which we call RT-Trajectory, that is practical, easy to specify, and allows the policy to effectively perform new tasks that would otherwise be challenging to perform. We find that trajectory sketches strike a balance between being detailed enough to express low-level motion-centric guidance while being coarse enough to allow the learned policy to interpret the trajectory sketch in the context of situational visual observations. In addition, we show how trajectory sketches can provide a useful interface to communicate with robotic policies: they can be specified through simple human inputs like drawings or videos, or through automated methods such as modern image-generating or waypoint-generating methods. We evaluate RT-Trajectory at scale on a variety of real-world robotic tasks, and find that RT-Trajectory is able to perform a wider range of tasks compared to language-conditioned and goal-conditioned policies, when provided the same training data.

Yuzhe Qin, Hao Su, Xiaolong Wang

We propose to perform imitation learning for dexterous manipulation with multi-finger robot hand from human demonstrations, and transfer the policy to the real robot hand. We introduce a novel single-camera teleoperation system to collect the 3D demonstrations efficiently with only an iPad and a computer. One key contribution of our system is that we construct a customized robot hand for each user in the physical simulator, which is a manipulator resembling the same kinematics structure and shape of the operator's hand. This provides an intuitive interface and avoid unstable human-robot hand retargeting for data collection, leading to large-scale and high quality data. Once the data is collected, the customized robot hand trajectories can be converted to different specified robot hands (models that are manufactured) to generate training demonstrations. With imitation learning using our data, we show large improvement over baselines with multiple complex manipulation tasks. Importantly, we show our learned policy is significantly more robust when transferring to the real robot. More videos can be found in the https://yzqin.github.io/dex-teleop-imitation .

Nicklas Hansen, Zhecheng Yuan, Yanjie Ze et al.

In this paper, we examine the effectiveness of pre-training for visuo-motor control tasks. We revisit a simple Learning-from-Scratch (LfS) baseline that incorporates data augmentation and a shallow ConvNet, and find that this baseline is surprisingly competitive with recent approaches (PVR, MVP, R3M) that leverage frozen visual representations trained on large-scale vision datasets -- across a variety of algorithms, task domains, and metrics in simulation and on a real robot. Our results demonstrate that these methods are hindered by a significant domain gap between the pre-training datasets and current benchmarks for visuo-motor control, which is alleviated by finetuning. Based on our findings, we provide recommendations for future research in pre-training for control and hope that our simple yet strong baseline will aid in accurately benchmarking progress in this area.

Isabella Liu, Linghao Chen, Ziyang Fu et al.

We introduce OpenIllumination, a real-world dataset containing over 108K images of 64 objects with diverse materials, captured under 72 camera views and a large number of different illuminations. For each image in the dataset, we provide accurate camera parameters, illumination ground truth, and foreground segmentation masks. Our dataset enables the quantitative evaluation of most inverse rendering and material decomposition methods for real objects. We examine several state-of-the-art inverse rendering methods on our dataset and compare their performances. The dataset and code can be found on the project page: https://oppo-us-research.github.io/OpenIllumination.

Zhiao Huang, Litian Liang, Zhan Ling et al.

We investigate the challenge of parametrizing policies for reinforcement learning (RL) in high-dimensional continuous action spaces. Our objective is to develop a multimodal policy that overcomes limitations inherent in the commonly-used Gaussian parameterization. To achieve this, we propose a principled framework that models the continuous RL policy as a generative model of optimal trajectories. By conditioning the policy on a latent variable, we derive a novel variational bound as the optimization objective, which promotes exploration of the environment. We then present a practical model-based RL method, called Reparameterized Policy Gradient (RPG), which leverages the multimodal policy parameterization and learned world model to achieve strong exploration capabilities and high data efficiency. Empirical results demonstrate that our method can help agents evade local optima in tasks with dense rewards and solve challenging sparse-reward environments by incorporating an object-centric intrinsic reward. Our method consistently outperforms previous approaches across a range of tasks. Code and supplementary materials are available on the project page https://haosulab.github.io/RPG/

Minghua Liu, Xuanlin Li, Zhan Ling et al.

We study how choices of input point cloud coordinate frames impact learning of manipulation skills from 3D point clouds. There exist a variety of coordinate frame choices to normalize captured robot-object-interaction point clouds. We find that different frames have a profound effect on agent learning performance, and the trend is similar across 3D backbone networks. In particular, the end-effector frame and the target-part frame achieve higher training efficiency than the commonly used world frame and robot-base frame in many tasks, intuitively because they provide helpful alignments among point clouds across time steps and thus can simplify visual module learning. Moreover, the well-performing frames vary across tasks, and some tasks may benefit from multiple frame candidates. We thus propose FrameMiners to adaptively select candidate frames and fuse their merits in a task-agnostic manner. Experimentally, FrameMiners achieves on-par or significantly higher performance than the best single-frame version on five fully physical manipulation tasks adapted from ManiSkill and OCRTOC. Without changing existing camera placements or adding extra cameras, point cloud frame mining can serve as a free lunch to improve 3D manipulation learning.

Hao Su, Xuefeng Liu, Jianwei Niu et al.

We propose 3Deformer, a general-purpose framework for interactive 3D shape editing. Given a source 3D mesh with semantic materials, and a user-specified semantic image, 3Deformer can accurately edit the source mesh following the shape guidance of the semantic image, while preserving the source topology as rigid as possible. Recent studies of 3D shape editing mostly focus on learning neural networks to predict 3D shapes, which requires high-cost 3D training datasets and is limited to handling objects involved in the datasets. Unlike these studies, our 3Deformer is a non-training and common framework, which only requires supervision of readily-available semantic images, and is compatible with editing various objects unlimited by datasets. In 3Deformer, the source mesh is deformed utilizing the differentiable renderer technique, according to the correspondences between semantic images and mesh materials. However, guiding complex 3D shapes with a simple 2D image incurs extra challenges, that is, the deform accuracy, surface smoothness, geometric rigidity, and global synchronization of the edited mesh should be guaranteed. To address these challenges, we propose a hierarchical optimization architecture to balance the global and local shape features, and propose further various strategies and losses to improve properties of accuracy, smoothness, rigidity, and so on. Extensive experiments show that our 3Deformer is able to produce impressive results and reaches the state-of-the-art level.

Songlin Wei, Jiazhao Zhang, Yang Wang et al.

Neural radiance fields with stochasticity have garnered significant interest by enabling the sampling of plausible radiance fields and quantifying uncertainty for downstream tasks. Existing works rely on the independence assumption of points in the radiance field or the pixels in input views to obtain tractable forms of the probability density function. However, this assumption inadvertently impacts performance when dealing with intricate geometry and texture. In this work, we propose an independence-assumption-free probabilistic neural radiance field based on Flow-GAN. By combining the generative capability of adversarial learning and the powerful expressivity of normalizing flow, our method explicitly models the density-radiance distribution of the whole scene. We represent our probabilistic NeRF as a mean-shifted probabilistic residual neural model. Our model is trained without an explicit likelihood function, thereby avoiding the independence assumption. Specifically, We downsample the training images with different strides and centers to form fixed-size patches which are used to train the generator with patch-based adversarial learning. Through extensive experiments, our method demonstrates state-of-the-art performance by predicting lower rendering errors and more reliable uncertainty on both synthetic and real-world datasets.

Qiyue Gao, Xinyu Pi, Kevin Liu et al. · cmu

Internal world models (WMs) enable agents to understand the world's state and predict transitions, serving as the basis for advanced deliberative reasoning. Recent large Vision-Language Models (VLMs), such as OpenAI o3, GPT-4o and Gemini, exhibit potential as general-purpose WMs. While the latest studies have evaluated and shown limitations in specific capabilities such as visual understanding, a systematic evaluation of VLMs' fundamental WM abilities remains absent. Drawing on comparative psychology and cognitive science, we propose a two-stage framework that assesses Perception (visual, spatial, temporal, quantitative, and motion) and Prediction (mechanistic simulation, transitive inference, compositional inference) to provide an atomic evaluation of VLMs as WMs. Guided by this framework, we introduce WM-ABench, a large-scale benchmark comprising 23 fine-grained evaluation dimensions across 6 diverse simulated environments with controlled counterfactual simulations. Through 660 experiments on 15 latest commercial and open-source VLMs, we find that these models exhibit striking limitations in basic world modeling abilities. For instance, almost all models perform at near-random accuracy when distinguishing motion trajectories. Additionally, they lack disentangled understanding -- e.g., some models tend to believe blue objects move faster than green ones. More rich results and analyses reveal significant gaps between VLMs and human-level world modeling.

Yuchen Zhou, Jiayuan Gu, Tung Yen Chiang et al.

The development of 2D foundation models for image segmentation has been significantly advanced by the Segment Anything Model (SAM). However, achieving similar success in 3D models remains a challenge due to issues such as non-unified data formats, poor model scalability, and the scarcity of labeled data with diverse masks. To this end, we propose a 3D promptable segmentation model Point-SAM, focusing on point clouds. We employ an efficient transformer-based architecture tailored for point clouds, extending SAM to the 3D domain. We then distill the rich knowledge from 2D SAM for Point-SAM training by introducing a data engine to generate part-level and object-level pseudo-labels at scale from 2D SAM. Our model outperforms state-of-the-art 3D segmentation models on several indoor and outdoor benchmarks and demonstrates a variety of applications, such as interactive 3D annotation and zero-shot 3D instance proposal. Codes and demo can be found at https://github.com/zyc00/Point-SAM.

Linghao Chen, Yuzhe Qin, Xiaowei Zhou et al.

Hand-eye calibration is a critical task in robotics, as it directly affects the efficacy of critical operations such as manipulation and grasping. Traditional methods for achieving this objective necessitate the careful design of joint poses and the use of specialized calibration markers, while most recent learning-based approaches using solely pose regression are limited in their abilities to diagnose inaccuracies. In this work, we introduce a new approach to hand-eye calibration called EasyHeC, which is markerless, white-box, and delivers superior accuracy and robustness. We propose to use two key technologies: differentiable rendering-based camera pose optimization and consistency-based joint space exploration, which enables accurate end-to-end optimization of the calibration process and eliminates the need for the laborious manual design of robot joint poses. Our evaluation demonstrates superior performance in synthetic and real-world datasets, enhancing downstream manipulation tasks by providing precise camera poses for locating and interacting with objects. The code is available at the project page: https://ootts.github.io/easyhec.

Xiaoshuai Zhang, Rui Chen, Ang Li et al.

In this paper, we focus on the simulation of active stereovision depth sensors, which are popular in both academic and industry communities. Inspired by the underlying mechanism of the sensors, we designed a fully physics-grounded simulation pipeline that includes material acquisition, ray-tracing-based infrared (IR) image rendering, IR noise simulation, and depth estimation. The pipeline is able to generate depth maps with material-dependent error patterns similar to a real depth sensor in real time. We conduct real experiments to show that perception algorithms and reinforcement learning policies trained in our simulation platform could transfer well to the real-world test cases without any fine-tuning. Furthermore, due to the high degree of realism of this simulation, our depth sensor simulator can be used as a convenient testbed to evaluate the algorithm performance in the real world, which will largely reduce the human effort in developing robotic algorithms. The entire pipeline has been integrated into the SAPIEN simulator and is open-sourced to promote the research of vision and robotics communities.

Xingchao Liu, Chengyue Gong, Lemeng Wu et al.

Generating images from natural language instructions is an intriguing yet highly challenging task. We approach text-to-image generation by combining the power of the retrained CLIP representation with an off-the-shelf image generator (GANs), optimizing in the latent space of GAN to find images that achieve maximum CLIP score with the given input text. Compared to traditional methods that train generative models from text to image starting from scratch, the CLIP+GAN approach is training-free, zero shot and can be easily customized with different generators. However, optimizing CLIP score in the GAN space casts a highly challenging optimization problem and off-the-shelf optimizers such as Adam fail to yield satisfying results. In this work, we propose a FuseDream pipeline, which improves the CLIP+GAN approach with three key techniques: 1) an AugCLIP score which robustifies the CLIP objective by introducing random augmentation on image. 2) a novel initialization and over-parameterization strategy for optimization which allows us to efficiently navigate the non-convex landscape in GAN space. 3) a composed generation technique which, by leveraging a novel bi-level optimization formulation, can compose multiple images to extend the GAN space and overcome the data-bias. When promoted by different input text, FuseDream can generate high-quality images with varying objects, backgrounds, artistic styles, even novel counterfactual concepts that do not appear in the training data of the GAN we use. Quantitatively, the images generated by FuseDream yield top-level Inception score and FID score on MS COCO dataset, without additional architecture design or training. Our code is publicly available at \url{https://github.com/gnobitab/FuseDream}.

Hao Su, Jianwei Niu, Xuefeng Liu et al.

Manga is a fashionable Japanese-style comic form that is composed of black-and-white strokes and is generally displayed as raster images on digital devices. Typical mangas have simple textures, wide lines, and few color gradients, which are vectorizable natures to enjoy the merits of vector graphics, e.g., adaptive resolutions and small file sizes. In this paper, we propose MARVEL (MAnga's Raster to VEctor Learning), a primitive-wise approach for vectorizing raster mangas by Deep Reinforcement Learning (DRL). Unlike previous learning-based methods which predict vector parameters for an entire image, MARVEL introduces a new perspective that regards an entire manga as a collection of basic primitives\textemdash stroke lines, and designs a DRL model to decompose the target image into a primitive sequence for achieving accurate vectorization. To improve vectorization accuracies and decrease file sizes, we further propose a stroke accuracy reward to predict accurate stroke lines, and a pruning mechanism to avoid generating erroneous and repeated strokes. Extensive subjective and objective experiments show that our MARVEL can generate impressive results and reaches the state-of-the-art level. Our code is open-source at: https://github.com/SwordHolderSH/Mang2Vec.

Raghav Kansal, Javier Duarte, Hao Su et al.

In high energy physics (HEP), jets are collections of correlated particles produced ubiquitously in particle collisions such as those at the CERN Large Hadron Collider (LHC). Machine learning (ML)-based generative models, such as generative adversarial networks (GANs), have the potential to significantly accelerate LHC jet simulations. However, despite jets having a natural representation as a set of particles in momentum-space, a.k.a. a particle cloud, there exist no generative models applied to such a dataset. In this work, we introduce a new particle cloud dataset (JetNet), and apply to it existing point cloud GANs. Results are evaluated using (1) 1-Wasserstein distances between high- and low-level feature distributions, (2) a newly developed Fréchet ParticleNet Distance, and (3) the coverage and (4) minimum matching distance metrics. Existing GANs are found to be inadequate for physics applications, hence we develop a new message passing GAN (MPGAN), which outperforms existing point cloud GANs on virtually every metric and shows promise for use in HEP. We propose JetNet as a novel point-cloud-style dataset for the ML community to experiment with, and set MPGAN as a benchmark to improve upon for future generative models. Additionally, to facilitate research and improve accessibility and reproducibility in this area, we release the open-source JetNet Python package with interfaces for particle cloud datasets, implementations for evaluation and loss metrics, and more tools for ML in HEP development.

Rui Chen, Songfang Han, Jing Xu et al.

We introduce Point-MVSNet, a novel point-based deep framework for multi-view stereo (MVS). Distinct from existing cost volume approaches, our method directly processes the target scene as point clouds. More specifically, our method predicts the depth in a coarse-to-fine manner. We first generate a coarse depth map, convert it into a point cloud and refine the point cloud iteratively by estimating the residual between the depth of the current iteration and that of the ground truth. Our network leverages 3D geometry priors and 2D texture information jointly and effectively by fusing them into a feature-augmented point cloud, and processes the point cloud to estimate the 3D flow for each point. This point-based architecture allows higher accuracy, more computational efficiency and more flexibility than cost-volume-based counterparts. Experimental results show that our approach achieves a significant improvement in reconstruction quality compared with state-of-the-art methods on the DTU and the Tanks and Temples dataset. Our source code and trained models are available at https://github.com/callmeray/PointMVSNet .

Hansheng Chen, Ruoxi Shi, Yulin Liu et al.

Open-domain 3D object synthesis has been lagging behind image synthesis due to limited data and higher computational complexity. To bridge this gap, recent works have investigated multi-view diffusion but often fall short in either 3D consistency, visual quality, or efficiency. This paper proposes MVEdit, which functions as a 3D counterpart of SDEdit, employing ancestral sampling to jointly denoise multi-view images and output high-quality textured meshes. Built on off-the-shelf 2D diffusion models, MVEdit achieves 3D consistency through a training-free 3D Adapter, which lifts the 2D views of the last timestep into a coherent 3D representation, then conditions the 2D views of the next timestep using rendered views, without uncompromising visual quality. With an inference time of only 2-5 minutes, this framework achieves better trade-off between quality and speed than score distillation. MVEdit is highly versatile and extendable, with a wide range of applications including text/image-to-3D generation, 3D-to-3D editing, and high-quality texture synthesis. In particular, evaluations demonstrate state-of-the-art performance in both image-to-3D and text-guided texture generation tasks. Additionally, we introduce a method for fine-tuning 2D latent diffusion models on small 3D datasets with limited resources, enabling fast low-resolution text-to-3D initialization.

Ruoxi Shi, Xinyue Wei, Cheng Wang et al.

We present ZeroRF, a novel per-scene optimization method addressing the challenge of sparse view 360° reconstruction in neural field representations. Current breakthroughs like Neural Radiance Fields (NeRF) have demonstrated high-fidelity image synthesis but struggle with sparse input views. Existing methods, such as Generalizable NeRFs and per-scene optimization approaches, face limitations in data dependency, computational cost, and generalization across diverse scenarios. To overcome these challenges, we propose ZeroRF, whose key idea is to integrate a tailored Deep Image Prior into a factorized NeRF representation. Unlike traditional methods, ZeroRF parametrizes feature grids with a neural network generator, enabling efficient sparse view 360° reconstruction without any pretraining or additional regularization. Extensive experiments showcase ZeroRF's versatility and superiority in terms of both quality and speed, achieving state-of-the-art results on benchmark datasets. ZeroRF's significance extends to applications in 3D content generation and editing. Project page: https://sarahweiii.github.io/zerorf/

Minghua Liu, Mikaela Angelina Uy, Donglai Xiang et al.

We propose PartField, a feedforward approach for learning part-based 3D features, which captures the general concept of parts and their hierarchy without relying on predefined templates or text-based names, and can be applied to open-world 3D shapes across various modalities. PartField requires only a 3D feedforward pass at inference time, significantly improving runtime and robustness compared to prior approaches. Our model is trained by distilling 2D and 3D part proposals from a mix of labeled datasets and image segmentations on large unsupervised datasets, via a contrastive learning formulation. It produces a continuous feature field which can be clustered to yield a hierarchical part decomposition. Comparisons show that PartField is up to 20% more accurate and often orders of magnitude faster than other recent class-agnostic part-segmentation methods. Beyond single-shape part decomposition, consistency in the learned field emerges across shapes, enabling tasks such as co-segmentation and correspondence, which we demonstrate in several applications of these general-purpose, hierarchical, and consistent 3D feature fields. Check our Webpage! https://research.nvidia.com/labs/toronto-ai/partfield-release/

Mukund Varma T, Peihao Wang, Zhiwen Fan et al.

In recent years, there has been an explosion of 2D vision models for numerous tasks such as semantic segmentation, style transfer or scene editing, enabled by large-scale 2D image datasets. At the same time, there has been renewed interest in 3D scene representations such as neural radiance fields from multi-view images. However, the availability of 3D or multiview data is still substantially limited compared to 2D image datasets, making extending 2D vision models to 3D data highly desirable but also very challenging. Indeed, extending a single 2D vision operator like scene editing to 3D typically requires a highly creative method specialized to that task and often requires per-scene optimization. In this paper, we ask the question of whether any 2D vision model can be lifted to make 3D consistent predictions. We answer this question in the affirmative; our new Lift3D method trains to predict unseen views on feature spaces generated by a few visual models (i.e. DINO and CLIP), but then generalizes to novel vision operators and tasks, such as style transfer, super-resolution, open vocabulary segmentation and image colorization; for some of these tasks, there is no comparable previous 3D method. In many cases, we even outperform state-of-the-art methods specialized for the task in question. Moreover, Lift3D is a zero-shot method, in the sense that it requires no task-specific training, nor scene-specific optimization.

Hansheng Chen, Bokui Shen, Yulin Liu et al.

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.

Isabella Liu, Hao Su, Xiaolong Wang

Modern 3D engines and graphics pipelines require mesh as a memory-efficient representation, which allows efficient rendering, geometry processing, texture editing, and many other downstream operations. However, it is still highly difficult to obtain high-quality mesh in terms of detailed structure and time consistency from dynamic observations. To this end, we introduce Dynamic Gaussians Mesh (DG-Mesh), a framework to reconstruct a high-fidelity and time-consistent mesh from dynamic input. Our work leverages the recent advancement in 3D Gaussian Splatting to construct the mesh sequence with temporal consistency from dynamic observations. Building on top of this representation, DG-Mesh recovers high-quality meshes from the Gaussian points and can track the mesh vertices over time, which enables applications such as texture editing on dynamic objects. We introduce the Gaussian-Mesh Anchoring, which encourages evenly distributed Gaussians, resulting better mesh reconstruction through mesh-guided densification and pruning on the deformed Gaussians. By applying cycle-consistent deformation between the canonical and the deformed space, we can project the anchored Gaussian back to the canonical space and optimize Gaussians across all time frames. During the evaluation on different datasets, DG-Mesh provides significantly better mesh reconstruction and rendering than baselines. Project page: https://www.liuisabella.com/DG-Mesh

Zhiao Huang, Feng Chen, Yewen Pu et al.

Combining gradient-based trajectory optimization with differentiable physics simulation is an efficient technique for solving soft-body manipulation problems. Using a well-crafted optimization objective, the solver can quickly converge onto a valid trajectory. However, writing the appropriate objective functions requires expert knowledge, making it difficult to collect a large set of naturalistic problems from non-expert users. We introduce DiffVL, a method that enables non-expert users to communicate soft-body manipulation tasks -- a combination of vision and natural language, given in multiple stages -- that can be readily leveraged by a differential physics solver. We have developed GUI tools that enable non-expert users to specify 100 tasks inspired by real-life soft-body manipulations from online videos, which we'll make public. We leverage large language models to translate task descriptions into machine-interpretable optimization objectives. The optimization objectives can help differentiable physics solvers to solve these long-horizon multistage tasks that are challenging for previous baselines.

Tongzhou Mu, Zhaoyang Li, Stanisław Wiktor Strzelecki et al.

Learning policies from high-dimensional visual inputs, such as pixels and point clouds, is crucial in various applications. Visual reinforcement learning is a promising approach that directly trains policies from visual observations, although it faces challenges in sample efficiency and computational costs. This study conducts an empirical comparison of State-to-Visual DAgger, a two-stage framework that initially trains a state policy before adopting online imitation to learn a visual policy, and Visual RL across a diverse set of tasks. We evaluate both methods across 16 tasks from three benchmarks, focusing on their asymptotic performance, sample efficiency, and computational costs. Surprisingly, our findings reveal that State-to-Visual DAgger does not universally outperform Visual RL but shows significant advantages in challenging tasks, offering more consistent performance. In contrast, its benefits in sample efficiency are less pronounced, although it often reduces the overall wall-clock time required for training. Based on our findings, we provide recommendations for practitioners and hope that our results contribute valuable perspectives for future research in visual policy learning.

Zhiwei Jia, Pradyumna Narayana, Arjun R. Akula et al.

Image ad understanding is a crucial task with wide real-world applications. Although highly challenging with the involvement of diverse atypical scenes, real-world entities, and reasoning over scene-texts, how to interpret image ads is relatively under-explored, especially in the era of foundational vision-language models (VLMs) featuring impressive generalizability and adaptability. In this paper, we perform the first empirical study of image ad understanding through the lens of pre-trained VLMs. We benchmark and reveal practical challenges in adapting these VLMs to image ad understanding. We propose a simple feature adaptation strategy to effectively fuse multimodal information for image ads and further empower it with knowledge of real-world entities. We hope our study draws more attention to image ad understanding which is broadly relevant to the advertising industry.

Minghua Liu, Ruoxi Shi, Kaiming Kuang et al.

We introduce OpenShape, a method for learning multi-modal joint representations of text, image, and point clouds. We adopt the commonly used multi-modal contrastive learning framework for representation alignment, but with a specific focus on scaling up 3D representations to enable open-world 3D shape understanding. To achieve this, we scale up training data by ensembling multiple 3D datasets and propose several strategies to automatically filter and enrich noisy text descriptions. We also explore and compare strategies for scaling 3D backbone networks and introduce a novel hard negative mining module for more efficient training. We evaluate OpenShape on zero-shot 3D classification benchmarks and demonstrate its superior capabilities for open-world recognition. Specifically, OpenShape achieves a zero-shot accuracy of 46.8% on the 1,156-category Objaverse-LVIS benchmark, compared to less than 10% for existing methods. OpenShape also achieves an accuracy of 85.3% on ModelNet40, outperforming previous zero-shot baseline methods by 20% and performing on par with some fully-supervised methods. Furthermore, we show that our learned embeddings encode a wide range of visual and semantic concepts (e.g., subcategories, color, shape, style) and facilitate fine-grained text-3D and image-3D interactions. Due to their alignment with CLIP embeddings, our learned shape representations can also be integrated with off-the-shelf CLIP-based models for various applications, such as point cloud captioning and point cloud-conditioned image generation.

Isabella Liu, Edward Yang, Jianyu Tao et al.

Traditional depth sensors generate accurate real world depth estimates that surpass even the most advanced learning approaches trained only on simulation domains. Since ground truth depth is readily available in the simulation domain but quite difficult to obtain in the real domain, we propose a method that leverages the best of both worlds. In this paper we present a new framework, ActiveZero, which is a mixed domain learning solution for active stereovision systems that requires no real world depth annotation. First, we demonstrate the transferability of our method to out-of-distribution real data by using a mixed domain learning strategy. In the simulation domain, we use a combination of supervised disparity loss and self-supervised losses on a shape primitives dataset. By contrast, in the real domain, we only use self-supervised losses on a dataset that is out-of-distribution from either training simulation data or test real data. Second, our method introduces a novel self-supervised loss called temporal IR reprojection to increase the robustness and accuracy of our reprojections in hard-to-perceive regions. Finally, we show how the method can be trained end-to-end and that each module is important for attaining the end result. Extensive qualitative and quantitative evaluations on real data demonstrate state of the art results that can even beat a commercial depth sensor.

Yijia Weng, He Wang, Qiang Zhou et al.

In this work, we tackle the problem of category-level online pose tracking of objects from point cloud sequences. For the first time, we propose a unified framework that can handle 9DoF pose tracking for novel rigid object instances as well as per-part pose tracking for articulated objects from known categories. Here the 9DoF pose, comprising 6D pose and 3D size, is equivalent to a 3D amodal bounding box representation with free 6D pose. Given the depth point cloud at the current frame and the estimated pose from the last frame, our novel end-to-end pipeline learns to accurately update the pose. Our pipeline is composed of three modules: 1) a pose canonicalization module that normalizes the pose of the input depth point cloud; 2) RotationNet, a module that directly regresses small interframe delta rotations; and 3) CoordinateNet, a module that predicts the normalized coordinates and segmentation, enabling analytical computation of the 3D size and translation. Leveraging the small pose regime in the pose-canonicalized point clouds, our method integrates the best of both worlds by combining dense coordinate prediction and direct rotation regression, thus yielding an end-to-end differentiable pipeline optimized for 9DoF pose accuracy (without using non-differentiable RANSAC). Our extensive experiments demonstrate that our method achieves new state-of-the-art performance on category-level rigid object pose (NOCS-REAL275) and articulated object pose benchmarks (SAPIEN, BMVC) at the fastest FPS ~12.

Quan Meng, Anpei Chen, Haimin Luo et al.

We introduce GNeRF, a framework to marry Generative Adversarial Networks (GAN) with Neural Radiance Field (NeRF) reconstruction for the complex scenarios with unknown and even randomly initialized camera poses. Recent NeRF-based advances have gained popularity for remarkable realistic novel view synthesis. However, most of them heavily rely on accurate camera poses estimation, while few recent methods can only optimize the unknown camera poses in roughly forward-facing scenes with relatively short camera trajectories and require rough camera poses initialization. Differently, our GNeRF only utilizes randomly initialized poses for complex outside-in scenarios. We propose a novel two-phases end-to-end framework. The first phase takes the use of GANs into the new realm for optimizing coarse camera poses and radiance fields jointly, while the second phase refines them with additional photometric loss. We overcome local minima using a hybrid and iterative optimization scheme. Extensive experiments on a variety of synthetic and natural scenes demonstrate the effectiveness of GNeRF. More impressively, our approach outperforms the baselines favorably in those scenes with repeated patterns or even low textures that are regarded as extremely challenging before.

Anpei Chen, Zexiang Xu, Fuqiang Zhao et al.

We present MVSNeRF, a novel neural rendering approach that can efficiently reconstruct neural radiance fields for view synthesis. Unlike prior works on neural radiance fields that consider per-scene optimization on densely captured images, we propose a generic deep neural network that can reconstruct radiance fields from only three nearby input views via fast network inference. Our approach leverages plane-swept cost volumes (widely used in multi-view stereo) for geometry-aware scene reasoning, and combines this with physically based volume rendering for neural radiance field reconstruction. We train our network on real objects in the DTU dataset, and test it on three different datasets to evaluate its effectiveness and generalizability. Our approach can generalize across scenes (even indoor scenes, completely different from our training scenes of objects) and generate realistic view synthesis results using only three input images, significantly outperforming concurrent works on generalizable radiance field reconstruction. Moreover, if dense images are captured, our estimated radiance field representation can be easily fine-tuned; this leads to fast per-scene reconstruction with higher rendering quality and substantially less optimization time than NeRF.

Zhiwei Jia, Bodi Yuan, Kangkang Wang et al.

Many applications of unpaired image-to-image translation require the input contents to be preserved semantically during translations. Unaware of the inherently unmatched semantics distributions between source and target domains, existing distribution matching methods (i.e., GAN-based) can give undesired solutions. In particular, although producing visually reasonable outputs, the learned models usually flip the semantics of the inputs. To tackle this without using extra supervision, we propose to enforce the translated outputs to be semantically invariant w.r.t. small perceptual variations of the inputs, a property we call "semantic robustness". By optimizing a robustness loss w.r.t. multi-scale feature space perturbations of the inputs, our method effectively reduces semantics flipping and produces translations that outperform existing methods both quantitatively and qualitatively.

Jiayuan Gu, Wei-Chiu Ma, Sivabalan Manivasagam et al.

3D shape completion for real data is important but challenging, since partial point clouds acquired by real-world sensors are usually sparse, noisy and unaligned. Different from previous methods, we address the problem of learning 3D complete shape from unaligned and real-world partial point clouds. To this end, we propose a weakly-supervised method to estimate both 3D canonical shape and 6-DoF pose for alignment, given multiple partial observations associated with the same instance. The network jointly optimizes canonical shapes and poses with multi-view geometry constraints during training, and can infer the complete shape given a single partial point cloud. Moreover, learned pose estimation can facilitate partial point cloud registration. Experiments on both synthetic and real data show that it is feasible and promising to learn 3D shape completion through large-scale data without shape and pose supervision.

Hao Su, Jianwei Niu, Xuefeng Liu et al.

Manga is a world popular comic form originated in Japan, which typically employs black-and-white stroke lines and geometric exaggeration to describe humans' appearances, poses, and actions. In this paper, we propose MangaGAN, the first method based on Generative Adversarial Network (GAN) for unpaired photo-to-manga translation. Inspired by how experienced manga artists draw manga, MangaGAN generates the geometric features of manga face by a designed GAN model and delicately translates each facial region into the manga domain by a tailored multi-GANs architecture. For training MangaGAN, we construct a new dataset collected from a popular manga work, containing manga facial features, landmarks, bodies, and so on. Moreover, to produce high-quality manga faces, we further propose a structural smoothing loss to smooth stroke-lines and avoid noisy pixels, and a similarity preserving module to improve the similarity between domains of photo and manga. Extensive experiments show that MangaGAN can produce high-quality manga faces which preserve both the facial similarity and a popular manga style, and outperforms other related state-of-the-art methods.

Fanbo Xiang, Yuzhe Qin, Kaichun Mo et al.

Building home assistant robots has long been a pursuit for vision and robotics researchers. To achieve this task, a simulated environment with physically realistic simulation, sufficient articulated objects, and transferability to the real robot is indispensable. Existing environments achieve these requirements for robotics simulation with different levels of simplification and focus. We take one step further in constructing an environment that supports household tasks for training robot learning algorithm. Our work, SAPIEN, is a realistic and physics-rich simulated environment that hosts a large-scale set for articulated objects. Our SAPIEN enables various robotic vision and interaction tasks that require detailed part-level understanding.We evaluate state-of-the-art vision algorithms for part detection and motion attribute recognition as well as demonstrate robotic interaction tasks using heuristic approaches and reinforcement learning algorithms. We hope that our SAPIEN can open a lot of research directions yet to be explored, including learning cognition through interaction, part motion discovery, and construction of robotics-ready simulated game environment.

Shuo Cheng, Zexiang Xu, Shilin Zhu et al.

We present Uncertainty-aware Cascaded Stereo Network (UCS-Net) for 3D reconstruction from multiple RGB images. Multi-view stereo (MVS) aims to reconstruct fine-grained scene geometry from multi-view images. Previous learning-based MVS methods estimate per-view depth using plane sweep volumes with a fixed depth hypothesis at each plane; this generally requires densely sampled planes for desired accuracy, and it is very hard to achieve high-resolution depth. In contrast, we propose adaptive thin volumes (ATVs); in an ATV, the depth hypothesis of each plane is spatially varying, which adapts to the uncertainties of previous per-pixel depth predictions. Our UCS-Net has three stages: the first stage processes a small standard plane sweep volume to predict low-resolution depth; two ATVs are then used in the following stages to refine the depth with higher resolution and higher accuracy. Our ATV consists of only a small number of planes; yet, it efficiently partitions local depth ranges within learned small intervals. In particular, we propose to use variance-based uncertainty estimates to adaptively construct ATVs; this differentiable process introduces reasonable and fine-grained spatial partitioning. Our multi-stage framework progressively subdivides the vast scene space with increasing depth resolution and precision, which enables scene reconstruction with high completeness and accuracy in a coarse-to-fine fashion. We demonstrate that our method achieves superior performance compared with state-of-the-art benchmarks on various challenging datasets.

Hao Zhu, Hao Su, Peng Wang et al.

We study how to synthesize novel views of human body from a single image. Though recent deep learning based methods work well for rigid objects, they often fail on objects with large articulation, like human bodies. The core step of existing methods is to fit a map from the observable views to novel views by CNNs; however, the rich articulation modes of human body make it rather challenging for CNNs to memorize and interpolate the data well. To address the problem, we propose a novel deep learning based pipeline that explicitly estimates and leverages the geometry of the underlying human body. Our new pipeline is a composition of a shape estimation network and an image generation network, and at the interface a perspective transformation is applied to generate a forward flow for pixel value transportation. Our design is able to factor out the space of data variation and makes learning at each step much easier. Empirically, we show that the performance for pose-varying objects can be improved dramatically. Our method can also be applied on real data captured by 3D sensors, and the flow generated by our methods can be used for generating high quality results in higher resolution.

Cewu Lu, Hao Su, Yongyi Lu et al.

Important high-level vision tasks such as human-object interaction, image captioning and robotic manipulation require rich semantic descriptions of objects at part level. Based upon previous work on part localization, in this paper, we address the problem of inferring rich semantics imparted by an object part in still images. We propose to tokenize the semantic space as a discrete set of part states. Our modeling of part state is spatially localized, therefore, we formulate the part state inference problem as a pixel-wise annotation problem. An iterative part-state inference neural network is specifically designed for this task, which is efficient in time and accurate in performance. Extensive experiments demonstrate that the proposed method can effectively predict the semantic states of parts and simultaneously correct localization errors, thus benefiting a few visual understanding applications. The other contribution of this paper is our part state dataset which contains rich part-level semantic annotations.

Wenzheng Chen, Huan Wang, Yangyan Li et al.

Human 3D pose estimation from a single image is a challenging task with numerous applications. Convolutional Neural Networks (CNNs) have recently achieved superior performance on the task of 2D pose estimation from a single image, by training on images with 2D annotations collected by crowd sourcing. This suggests that similar success could be achieved for direct estimation of 3D poses. However, 3D poses are much harder to annotate, and the lack of suitable annotated training images hinders attempts towards end-to-end solutions. To address this issue, we opt to automatically synthesize training images with ground truth pose annotations. Our work is a systematic study along this road. We find that pose space coverage and texture diversity are the key ingredients for the effectiveness of synthetic training data. We present a fully automatic, scalable approach that samples the human pose space for guiding the synthesis procedure and extracts clothing textures from real images. Furthermore, we explore domain adaptation for bridging the gap between our synthetic training images and real testing photos. We demonstrate that CNNs trained with our synthetic images out-perform those trained with real photos on 3D pose estimation tasks.

Angel X. Chang, Thomas Funkhouser, Leonidas Guibas et al.

We present ShapeNet: a richly-annotated, large-scale repository of shapes represented by 3D CAD models of objects. ShapeNet contains 3D models from a multitude of semantic categories and organizes them under the WordNet taxonomy. It is a collection of datasets providing many semantic annotations for each 3D model such as consistent rigid alignments, parts and bilateral symmetry planes, physical sizes, keywords, as well as other planned annotations. Annotations are made available through a public web-based interface to enable data visualization of object attributes, promote data-driven geometric analysis, and provide a large-scale quantitative benchmark for research in computer graphics and vision. At the time of this technical report, ShapeNet has indexed more than 3,000,000 models, 220,000 models out of which are classified into 3,135 categories (WordNet synsets). In this report we describe the ShapeNet effort as a whole, provide details for all currently available datasets, and summarize future plans.

Hao Su, Charles R. Qi, Yangyan Li et al.

Object viewpoint estimation from 2D images is an essential task in computer vision. However, two issues hinder its progress: scarcity of training data with viewpoint annotations, and a lack of powerful features. Inspired by the growing availability of 3D models, we propose a framework to address both issues by combining render-based image synthesis and CNNs. We believe that 3D models have the potential in generating a large number of images of high variation, which can be well exploited by deep CNN with a high learning capacity. Towards this goal, we propose a scalable and overfit-resistant image synthesis pipeline, together with a novel CNN specifically tailored for the viewpoint estimation task. Experimentally, we show that the viewpoint estimation from our pipeline can significantly outperform state-of-the-art methods on PASCAL 3D+ benchmark.

Hao Su, Fan Wang, Li Yi et al.

Comparing two images in a view-invariant way has been a challenging problem in computer vision for a long time, as visual features are not stable under large view point changes. In this paper, given a single input image of an object, we synthesize new features for other views of the same object. To accomplish this, we introduce an aligned set of 3D models in the same class as the input object image. Each 3D model is represented by a set of views, and we study the correlation of image patches between different views, seeking what we call surrogates --- patches in one view whose feature content predicts well the features of a patch in another view. In particular, for each patch in the novel desired view, we seek surrogates from the observed view of the given image. For a given surrogate, we predict that surrogate using linear combination of the corresponding patches of the 3D model views, learn the coefficients, and then transfer these coefficients on a per patch basis to synthesize the features of the patch in the novel view. In this way we can create feature sets for all views of the latent object, providing us a multi-view representation of the object. View-invariant object comparisons are achieved simply by computing the $L^2$ distances between the features of corresponding views. We provide theoretical and empirical analysis of the feature synthesis process, and evaluate the proposed view-agnostic distance (VAD) in fine-grained image retrieval (100 object classes) and classification tasks. Experimental results show that our synthesized features do enable view-independent comparison between images and perform significantly better than traditional image features in this respect.