NeurIPS 2023 论文列表

In this paper, we present Motion-X, a large-scale 3D expressive whole-body motion dataset. Existing motion datasets predominantly contain body-only poses, lacking facial expressions, hand gestures, and fine-grained pose descriptions. Moreover, they are primarily collected from limited laboratory scenes with textual descriptions manually labeled, which greatly limits their scalability. To overcome these limitations, we develop a whole-body motion and text annotation pipeline, which can automatically annotate motion from either single- or multi-view videos and provide comprehensive semantic labels for each video and fine-grained whole-body pose descriptions for each frame. This pipeline is of high precision, cost-effective, and scalable for further research. Based on it, we construct Motion-X, which comprises 15.6M precise 3D whole-body pose annotations (i.e., SMPL-X) covering 81.1K motion sequences from massive scenes. Besides, Motion-X provides 15.6M frame-level whole-body pose descriptions and 81.1K sequence-level semantic labels. Comprehensive experiments demonstrate the accuracy of the annotation pipeline and the significant benefit of Motion-X in enhancing expressive, diverse, and natural motion generation, as well as 3D whole-body human mesh recovery.

Neural Radiance Field (NeRF) has enabled novel view synthesis with high fidelity given images and camera poses. Subsequent works even succeeded in eliminating the necessity of pose priors by jointly optimizing NeRF and camera pose. However, these works are limited to relatively simple settings such as photometrically consistent and occluder-free image collections or a sequence of images from a video. So they have difficulty handling unconstrained images with varying illumination and transient occluders. In this paper, we propose **UP-NeRF** (**U**nconstrained **P**ose-prior-free **Ne**ural **R**adiance **F**ields) to optimize NeRF with unconstrained image collections without camera pose prior. We tackle these challenges with surrogate tasks that optimize color-insensitive feature fields and a separate module for transient occluders to block their influence on pose estimation. In addition, we introduce a candidate head to enable more robust pose estimation and transient-aware depth supervision to minimize the effect of incorrect prior. Our experiments verify the superior performance of our method compared to the baselines including BARF and its variants in a challenging internet photo collection, *Phototourism dataset*. The code of UP-NeRF is available at https://github.com/mlvlab/UP-NeRF.

Monocular 3D scene reconstruction aims to reconstruct the 3D structure of scenes based on posed images. Recent volumetric-based methods directly predict the truncated signed distance function (TSDF) volume and have achieved promising results. The memory cost of volumetric-based methods will grow cubically as the volume size increases, so a coarse-to-fine strategy is necessary for saving memory. Specifically, the coarse-to-fine strategy distinguishes surface voxels from non-surface voxels, and only potential surface voxels are considered in the succeeding procedure. However, the non-surface voxels have various features, and in particular, the voxels on the inner side of the surface are quite different from those on the outer side since there exists an intrinsic gap between them. Therefore, grouping inner-surface and outer-surface voxels into the same class will force the classifier to spend its capacity to bridge the gap. By contrast, it is relatively easy for the classifier to distinguish inner-surface and outer-surface voxels due to the intrinsic gap. Inspired by this, we propose the inner-outer aware reconstruction (IOAR) model. IOAR explores a new coarse-to-fine strategy to classify outer-surface, inner-surface and surface voxels. In addition, IOAR separates occupancy branches from TSDF branches to avoid mutual interference between them. Since our model can better classify the surface, outer-surface and inner-surface voxels, it can predict more precise meshes than existing methods. Experiment results on ScanNet, ICL-NUIM and TUM-RGBD datasets demonstrate the effectiveness and generalization of our model. The code is available at https://github.com/YorkQiu/InnerOuterAwareReconstruction.

We propose Neural 3D Articulated object Prior (NAP), the first 3D deep generative model to synthesize 3D articulated object models. Despite the extensive research on generating 3D static objects, compositions, or scenes, there are hardly any approaches on capturing the distribution of articulated objects, a common object category for human and robot interaction. To generate articulated objects, we first design a novel articulation tree/graph parameterization and then apply a diffusion-denoising probabilistic model over this representation where articulated objects can be generated via denoising from random complete graphs. In order to capture both the geometry and the motion structure whose distribution will affect each other, we design a graph denoising network for learning the reverse diffusion process. We propose a novel distance that adapts widely used 3D generation metrics to our novel task to evaluate generation quality. Experiments demonstrate our high performance in articulated object generation as well as its applications on conditioned generation, including Part2Motion, PartNet-Imagination, Motion2Part, and GAPart2Object.

The kinematics of human movements and locomotion are closely linked to the activation and contractions of muscles. To investigate this, we present a multimodal dataset with benchmarks collected using a novel pair of Intelligent Knee Sleeves (Texavie MarsWear Knee Sleeves) for human pose estimation. Our system utilizes synchronized datasets that comprise time-series data from the Knee Sleeves and the corresponding ground truth labels from visualized motion capture camera system. We employ these to generate 3D human models solely based on the wearable data of individuals performing different activities. We demonstrate the effectiveness of this camera-free system and machine learning algorithms in the assessment of various movements and exercises, including extension to unseen exercises and individuals. The results show an average error of 7.21 degrees across all eight lower body joints when compared to the ground truth, indicating the effectiveness and reliability of the Knee Sleeve system for the prediction of different lower body joints beyond knees. The results enable human pose estimation in a seamless manner without being limited by visual occlusion or the field of view of cameras. Our results show the potential of multimodal wearable sensing in a variety of applications from home fitness to sports, healthcare, and physical rehabilitation focusing on pose and movement estimation.

Natural language processing and 2D vision models have attained remarkable proficiency on many tasks primarily by escalating the scale of training data. However, 3D vision tasks have not seen the same progress, in part due to the challenges of acquiring high-quality 3D data. In this work, we present Objaverse-XL, a dataset of over 10 million 3D objects. Our compilation comprises deduplicated 3D objects from a diverse set of sources, including manually designed objects, photogrammetry scans of landmarks and everyday items, and professional scans of historic and antique artifacts. Representing the largest scale and diversity in the realm of 3D datasets, Objaverse-XL enables significant new possibilities for 3D vision. Our experiments demonstrate the vast improvements enabled with the scale provided by Objaverse-XL. We show that by training Zero123 on novel view synthesis, utilizing over 100 million multi-view rendered images, we achieve strong zero-shot generalization abilities. We hope that releasing Objaverse-XL will enable further innovations in the field of 3D vision at scale.

Recent Diffusion Transformers (i.e., DiT) have demonstrated their powerful effectiveness in generating high-quality 2D images. However, it is unclear how the Transformer architecture performs equally well in 3D shape generation, as previous 3D diffusion methods mostly adopted the U-Net architecture. To bridge this gap, we propose a novel Diffusion Transformer for 3D shape generation, named DiT-3D, which can directly operate the denoising process on voxelized point clouds using plain Transformers. Compared to existing U-Net approaches, our DiT-3D is more scalable in model size and produces much higher quality generations. Specifically, the DiT-3D adopts the design philosophy of DiT but modifies it by incorporating 3D positional and patch embeddings to aggregate input from voxelized point clouds. To reduce the computational cost of self-attention in 3D shape generation, we incorporate 3D window attention into Transformer blocks, as the increased 3D token length resulting from the additional dimension of voxels can lead to high computation. Finally, linear and devoxelization layers are used to predict the denoised point clouds. In addition, we empirically observe that the pre-trained DiT-2D checkpoint on ImageNet can significantly improve DiT-3D on ShapeNet. Experimental results on the ShapeNet dataset demonstrate that the proposed DiT-3D achieves state-of-the-art performance in high-fidelity and diverse 3D point cloud generation.

The aim of this paper is the reconstruction of a smooth surface from an unorganized point cloud sampled by a closed surface, with the preservation of geometric shapes, without any further information other than the point cloud. Implicit neural representations (INRs) have recently emerged as a promising approach to surface reconstruction. However, the reconstruction quality of existing methods relies on ground truth implicit function values or surface normal vectors. In this paper, we show that proper supervision of partial differential equations and fundamental properties of differential vector fields are sufficient to robustly reconstruct high-quality surfaces. We cast the $p$-Poisson equation to learn a signed distance function (SDF) and the reconstructed surface is implicitly represented by the zero-level set of the SDF. For efficient training, we develop a variable splitting structure by introducing a gradient of the SDF as an auxiliary variable and impose the $p$-Poisson equation directly on the auxiliary variable as a hard constraint. Based on the curl-free property of the gradient field, we impose a curl-free constraint on the auxiliary variable, which leads to a more faithful reconstruction. Experiments on standard benchmark datasets show that the proposed INR provides a superior and robust reconstruction. The code is available at https://github.com/Yebbi/PINC.

Combining the signed distance function (SDF) and differentiable volume rendering has emerged as a powerful paradigm for surface reconstruction from multi-view images without 3D supervision. However, current methods are impeded by requiring long-time per-scene optimizations and cannot generalize to new scenes. In this paper, we present GenS, an end-to-end generalizable neural surface reconstruction model. Unlike coordinate-based methods that train a separate network for each scene, we construct a generalized multi-scale volume to directly encode all scenes. Compared with existing solutions, our representation is more powerful, which can recover high-frequency details while maintaining global smoothness. Meanwhile, we introduce a multi-scale feature-metric consistency to impose the multi-view consistency in a more discriminative multi-scale feature space, which is robust to the failures of the photometric consistency. And the learnable feature can be self-enhanced to continuously improve the matching accuracy and mitigate aggregation ambiguity. Furthermore, we design a view contrast loss to force the model to be robust to those regions covered by few viewpoints through distilling the geometric prior from dense input to sparse input. Extensive experiments on popular benchmarks show that our model can generalize well to new scenes and outperform existing state-of-the-art methods even those employing ground-truth depth supervision. Code will be available at https://github.com/prstrive/GenS.

Open-vocabulary 3D Object Detection (OV-3DDet) aims to detect objects from an arbitrary list of categories within a 3D scene, which remains seldom explored in the literature. There are primarily two fundamental problems in OV-3DDet, *i.e.*, localizing and classifying novel objects. This paper aims at addressing the two problems simultaneously via a unified framework, under the condition of limited base categories. To localize novel 3D objects, we propose an effective 3D Novel Object Discovery strategy, which utilizes both the 3D box geometry priors and 2D semantic open-vocabulary priors to generate pseudo box labels of the novel objects. To classify novel object boxes, we further develop a cross-modal alignment module based on discovered novel boxes, to align feature spaces between 3D point cloud and image/text modalities. Specifically, the alignment process contains a class-agnostic and a class-discriminative alignment, incorporating not only the base objects with annotations but also the increasingly discovered novel objects, resulting in an iteratively enhanced alignment. The novel box discovery and crossmodal alignment are jointly learned to collaboratively benefit each other. The novel object discovery can directly impact the cross-modal alignment, while a better feature alignment can, in turn, boost the localization capability, leading to a unified OV-3DDet framework, named **CoDA**, for simultaneous novel object localization and classification. Extensive experiments on two challenging datasets (*i.e.*, SUN-RGBD and ScanNet) demonstrate the effectiveness of our method and also show a significant mAP improvement upon the best-performing alternative method by 80%. Codes and pre-trained models are released on [the project page](https://yangcaoai.github.io/publications/CoDA.html).

The ability to create high-quality 3D faces from a single image has become increasingly important with wide applications in video conferencing, AR/VR, and advanced video editing in movie industries. In this paper, we propose Face Diffusion NeRF (FaceDNeRF), a new generative method to reconstruct high-quality Face NeRFs from single images, complete with semantic editing and relighting capabilities. FaceDNeRF utilizes high-resolution 3D GAN inversion and expertly trained 2D latent-diffusion model, allowing users to manipulate and construct Face NeRFs in zero-shot learning without the need for explicit 3D data. With carefully designed illumination and identity preserving loss, as well as multi-modal pre-training, FaceDNeRF offers users unparalleled control over the editing process enabling them to create and edit face NeRFs using just single-view images, text prompts, and explicit target lighting. The advanced features of FaceDNeRF have been designed to produce more impressive results than existing 2D editing approaches that rely on 2D segmentation maps for editable attributes. Experiments show that our FaceDNeRF achieves exceptionally realistic results and unprecedented flexibility in editing compared with state-of-the-art 3D face reconstruction and editing methods. Our code will be available at https://github.com/BillyXYB/FaceDNeRF.

The task of point cloud segmentation, comprising semantic, instance, and panoptic segmentation, has been mainly tackled by designing task-specific network architectures, which often lack the flexibility to generalize across tasks, thus resulting in a fragmented research landscape. In this paper, we introduce ProtoSEG, a prototype-based model that unifies semantic, instance, and panoptic segmentation tasks. Our approach treats these three homogeneous tasks as a classification problem with different levels of granularity. By leveraging a Transformer architecture, we extract point embeddings to optimize prototype-class distances and dynamically learn class prototypes to accommodate the end tasks. Our prototypical design enjoys simplicity and transparency, powerful representational learning, and ad-hoc explainability. Empirical results demonstrate that ProtoSEG outperforms concurrent well-known specialized architectures on 3D point cloud benchmarks, achieving 72.3%, 76.4% and 74.2% mIoU for semantic segmentation on S3DIS, ScanNet V2 and SemanticKITTI, 66.8% mCov and 51.2% mAP for instance segmentation on S3DIS and ScanNet V2, 62.4% PQ for panoptic segmentation on SemanticKITTI, validating the strength of our concept and the effectiveness of our algorithm. The code and models are available at https://github.com/zyqin19/PROTOSEG.

We introduce ViCA-NeRF, the *first* view-consistency-aware method for 3D editing with text instructions. In addition to the implicit neural radiance field (NeRF) modeling, our key insight is to exploit two sources of regularization that *explicitly* propagate the editing information across different views, thus ensuring multi-view consistency. For *geometric regularization*, we leverage the depth information derived from NeRF to establish image correspondences between different views. For *learned regularization*, we align the latent codes in the 2D diffusion model between edited and unedited images, enabling us to edit key views and propagate the update throughout the entire scene. Incorporating these two strategies, our ViCA-NeRF operates in two stages. In the initial stage, we blend edits from different views to create a preliminary 3D edit. This is followed by a second stage of NeRF training, dedicated to further refining the scene's appearance. Experimental results demonstrate that ViCA-NeRF provides more flexible, efficient (3 times faster) editing with higher levels of consistency and details, compared with the state of the art. Our code is available at: https://github.com/Dongjiahua/VICA-NeRF

Open-vocabulary segmentation of 3D scenes is a fundamental function of human perception and thus a crucial objective in computer vision research. However, this task is heavily impeded by the lack of large-scale and diverse 3D open-vocabulary segmentation datasets for training robust and generalizable models. Distilling knowledge from pre-trained 2D open-vocabulary segmentation models helps but it compromises the open-vocabulary feature as the 2D models are mostly finetuned with close-vocabulary datasets. We tackle the challenges in 3D open-vocabulary segmentation by exploiting pre-trained foundation models CLIP and DINO in a weakly supervised manner. Specifically, given only the open-vocabulary text descriptions of the objects in a scene, we distill the open-vocabulary multimodal knowledge and object reasoning capability of CLIP and DINO into a neural radiance field (NeRF), which effectively lifts 2D features into view-consistent 3D segmentation. A notable aspect of our approach is that it does not require any manual segmentation annotations for either the foundation models or the distillation process. Extensive experiments show that our method even outperforms fully supervised models trained with segmentation annotations in certain scenes, suggesting that 3D open-vocabulary segmentation can be effectively learned from 2D images and text-image pairs. Code is available at https://github.com/Kunhao-Liu/3D-OVS.

Accurately depicting the complex traffic scene is a vital component for autonomous vehicles to execute correct judgments. However, existing benchmarks tend to oversimplify the scene by solely focusing on lane perception tasks. Observing that human drivers rely on both lanes and traffic signals to operate their vehicles safely, we present OpenLane-V2, the first dataset on topology reasoning for traffic scene structure. The objective of the presented dataset is to advance research in understanding the structure of road scenes by examining the relationship between perceived entities, such as traffic elements and lanes. Leveraging existing datasets, OpenLane-V2 consists of 2,000 annotated road scenes that describe traffic elements and their correlation to the lanes. It comprises three primary sub-tasks, including the 3D lane detection inherited from OpenLane, accompanied by corresponding metrics to evaluate the model’s performance. We evaluate various state-of-the-art methods, and present their quantitative and qualitative results on OpenLane-V2 to indicate future avenues for investigating topology reasoning in traffic scenes.

Various deep learning models have been proposed for 3D bone shape reconstruction from two orthogonal (biplanar) X-ray images. However, it is unclear how these models compare against each other since they are evaluated on different anatomy, cohort and (often privately held) datasets. Moreover, the impact of the commonly optimized image-based segmentation metrics such as dice score on the estimation of clinical parameters relevant in 2D-3D bone shape reconstruction is not well known. To move closer toward clinical translation, we propose a benchmarking framework that evaluates tasks relevant to real-world clinical scenarios, including reconstruction of fractured bones, bones with implants, robustness to population shift, and error in estimating clinical parameters. Our open-source platform provides reference implementations of 8 models (many of whose implementations were not publicly available), APIs to easily collect and preprocess 6 public datasets, and the implementation of automatic clinical parameter and landmark extraction methods. We present an extensive evaluation of 8 2D-3D models on equal footing using 6 public datasets comprising images for four different anatomies. Our results show that attention-based methods that capture global spatial relationships tend to perform better across all anatomies and datasets; performance on clinically relevant subgroups may be overestimated without disaggregated reporting; ribs are substantially more difficult to reconstruct compared to femur, hip and spine; and the dice score improvement does not always bring corresponding improvement in the automatic estimation of clinically relevant parameters.

3D scene segmentation based on neural implicit representation has emerged recently with the advantage of training only on 2D supervision. However, existing approaches still requires expensive per-scene optimization that prohibits generalization to novel scenes during inference. To circumvent this problem, we introduce a \textit{generalizable} 3D segmentation framework based on implicit representation. Specifically, our framework takes in multi-view image features and semantic maps as the inputs instead of only spatial information to avoid overfitting to scene-specific geometric and semantic information. We propose a novel soft voting mechanism to aggregate the 2D semantic information from different views for each 3D point. In addition to the image features, view difference information is also encoded in our framework to predict the voting scores. Intuitively, this allows the semantic information from nearby views to contribute more compared to distant ones. Furthermore, a visibility module is also designed to detect and filter out detrimental information from occluded views. Due to the generalizability of our proposed method, we can synthesize semantic maps or conduct 3D semantic segmentation for novel scenes with solely 2D semantic supervision. Experimental results show that our approach achieves comparable performance with scene-specific approaches. More importantly, our approach can even outperform existing strong supervision-based approaches with only 2D annotations.

While Vision Transformers (ViTs) have undoubtedly made impressive strides in computer vision (CV), their intricate network structures necessitate substantial computation and memory resources. A decision-making process for CV tasks typically entails performing computations with low latency, which is a tricky problem for ViT models. Model quantization is a widely-used technique to optimize the hardware efficiency of deep neural networks. Full quantization under Sub-8-bit precision, in particular, is a promising solution to reduce inference latency significantly. Unfortunately, current commodity hardware, such as CPUs and GPUs, still struggles to efficiently execute these sub-8-bit quantized networks, as their SIMD instructions only support a granularity of 8 bits or wider. Also, there is a scarcity of literature that presents a full quantization paradigm for ViTs. In this paper, we propose an activation-aware fully sub-8-bit quantization-aware training (QAT) framework called PackQViT for efficient yet accurate ViT acceleration on mobile devices to facilitate real-time AI-powered decision-making. Specifically, in revisiting data activation within the ViT dataflow, two characteristics are relevant to quantization strategy and precision: the long-tailed distribution and systematic channel-wise outliers. In response, we employ either log2 quantization or clipping to address the long-tailed distribution and incorporate outlier-aware training for residual link quantization to regulate the various channel-wise outliers more consistently. Notably, due to the systematic fixed pattern, outlier-aware training approach can predict the channel indices and regularized scales of outliers in advance, thus avoiding the runtime data-adaptive selection during inference. Furthermore, we employ Int-$2^{n}$-Softmax, Int-LayerNorm, and Integer GELU to enable integer-only computation flow. Finally, we develop a SIMD-based 4-bit packed multiplier to achieve end-to-end ViT acceleration on mobile phones. Compared to prior studies on ViT quantization using 8-bit precision, PackQViT surpasses other works by an improved accuracy ranging from 0.4\% to 17.9\% for various widely used ViTs on ImageNet dataset; under 4-bit precision, PackQViT demonstrates 0.4%$\sim$2.8% higher accuracy. Compared to the baseline multiplier, our implementations on the Realme GT Android smartphone with Snapdragon 870 SoC CPU achieve 2.6x$\sim$3.7x speedup under 8-bit scenario and 3.8x$\sim$5.9x speedup under 4-bit which ensures practical real-time performance.

We present AircraftVerse, a publicly available aerial vehicle design dataset. Aircraft design encompasses different physics domains and, hence, multiple modalities of representation. The evaluation of these designs requires the use of scientific analytical and simulation models ranging from computer-aided design tools for structural and manufacturing analysis, computational fluid dynamics tools for drag and lift computation, battery models for energy estimation, and simulation models for flight control and dynamics. AircraftVerse contains $27{,}714$ diverse air vehicle designs - the largest corpus of designs with this level of complexity. Each design comprises the following artifacts: a symbolic design tree describing topology, propulsion subsystem, battery subsystem, and other design details; a STandard for the Exchange of Product (STEP) model data; a 3D CAD design using a stereolithography (STL) file format; a 3D point cloud for the shape of the design; and evaluation results from high fidelity state-of-the-art physics models that characterize performance metrics such as maximum flight distance and hover-time. We also present baseline surrogate models that use different modalities of design representation to predict design performance metrics, which we provide as part of our dataset release. Finally, we discuss the potential impact of this dataset on the use of learning in aircraft design, and more generally, in the emerging field of deep learning for scientific design. AircraftVerse is accompanied by a datasheet as suggested in the recent literature, and it is released under Creative Commons Attribution-ShareAlike (CC BY-SA) license. The dataset with baseline models are hosted at http://doi.org/10.5281/zenodo.6525446, code at https://github.com/SRI-CSL/AircraftVerse, and the dataset description at https://uavdesignverse.onrender.com/.