NeurIPS 2024 论文列表

We present a novel bi-directional Transformer architecture (BiXT) which scales linearly with input size in terms of computational cost and memory consumption, but does not suffer the drop in performance or limitation to only one input modality seen with other efficient Transformer-based approaches. BiXT is inspired by the Perceiver architectures but replaces iterative attention with an efficient bi-directional cross-attention module in which input tokens and latent variables attend to each other simultaneously, leveraging a naturally emerging attention-symmetry between the two. This approach unlocks a key bottleneck experienced by Perceiver-like architectures and enables the processing and interpretation of both semantics ('what') and location ('where') to develop alongside each other over multiple layers -- allowing its direct application to dense and instance-based tasks alike. By combining efficiency with the generality and performance of a full Transformer architecture, BiXT can process longer sequences like point clouds, text or images at higher feature resolutions and achieves competitive performance across a range of tasks like point cloud part segmentation, semantic image segmentation, image classification, hierarchical sequence modeling and document retrieval. Our experiments demonstrate that BiXT models outperform larger competitors by leveraging longer sequences more efficiently on vision tasks like classification and segmentation, and perform on par with full Transformer variants on sequence modeling and document retrieval -- but require 28\% fewer FLOPs and are up to $8.4\times$ faster.

The Janus Problem is a common issue in SDS-based text-to-3D methods. Due to view encoding approach and 2D diffusion prior guidance, the 3D representation model tends to learn content with higher certainty from each perspective, leading to view inconsistency. In this work, we first model and analyze the problem, visualizing the specific causes of the Janus Problem, which are associated with discrete view encoding and shared priors in 2D lifting. Based on this, we further propose the LCGen method, which guides text-to-3D to obtain different priors with different certainty from various viewpoints, aiding in view-consistent generation. Experiments have proven that our LCGen method can be directly applied to different SDS-based text-to-3D methods, alleviating the Janus Problem without introducing additional information, increasing excessive training burden, or compromising the generation effect.

Detecting 3D keypoints with semantic consistency is widely used in many scenarios such as pose estimation, shape registration and robotics. Currently, most unsupervised 3D keypoint detection methods focus on the rigid-body objects. However, when faced with deformable objects, the keypoints they identify do not preserve semantic consistency well. In this paper, we introduce an innovative unsupervised keypoint detector Key-Grid for both the rigid-body and deformable objects, which is an autoencoder framework. The encoder predicts keypoints and the decoder utilizes the generated keypoints to reconstruct the objects. Unlike previous work, we leverage the identified keypoint in formation to form a 3D grid feature heatmap called grid heatmap, which is used in the decoder section. Grid heatmap is a novel concept that represents the latent variables for grid points sampled uniformly in the 3D cubic space, where these variables are the shortest distance between the grid points and the “skeleton” connected by keypoint pairs. Meanwhile, we incorporate the information from each layer of the encoder into the decoder section. We conduct an extensive evaluation of Key-Grid on a list of benchmark datasets. Key-Grid achieves the state-of-the-art performance on the semantic consistency and position accuracy of keypoints. Moreover, we demonstrate the robustness of Key-Grid to noise and downsampling. In addition, we achieve SE-(3) invariance of keypoints though generalizing Key-Grid to a SE(3)-invariant backbone.

Although 3D point cloud recognition has achieved substantial progress on standard benchmarks, the typical models are vulnerable to point cloud corruptions, leading to security threats in real-world applications. To improve the corruption robustness, various data augmentation methods have been studied, but they are mainly limited to the spatial domain. As the point cloud has low information density and significant spatial redundancy, it is challenging to analyze the effects of corruptions. In this paper, we focus on the frequency domain to observe the underlying structure of point clouds and their corruptions. Through graph Fourier transform (GFT), we observe a correlation between the corruption robustness of point cloud recognition models and their sensitivity to different frequency bands, which is measured by the GFT spectrum of the model’s Jacobian matrix. To reduce the sensitivity and improve the corruption robustness, we propose Frequency Adversarial Training (FAT) that adopts frequency-domain adversarial examples as data augmentation to train robust point cloud recognition models against corruptions. Theoretically, we provide a guarantee of FAT on its out-of-distribution generalization performance. Empirically, we conduct extensive experiments with various network architectures to validate the effectiveness of FAT, which achieves the new state-of-the-art results.

Novel view acoustic synthesis (NVAS) aims to render binaural audio at any target viewpoint, given a mono audio emitted by a sound source at a 3D scene. Existing methods have proposed NeRF-based implicit models to exploit visual cues as a condition for synthesizing binaural audio. However, in addition to low efficiency originating from heavy NeRF rendering, these methods all have a limited ability of characterizing the entire scene environment such as room geometry, material properties, and the spatial relation between the listener and sound source. To address these issues, we propose a novel Audio-Visual Gaussian Splatting (AV-GS) model. To obtain a material-aware and geometry-aware condition for audio synthesis, we learn an explicit point-based scene representation with audio-guidance parameters on locally initialized Gaussian points, taking into account the space relation from the listener and sound source. To make the visual scene model audio adaptive, we propose a point densification and pruning strategy to optimally distribute the Gaussian points, with the per-point contribution in sound propagation (e.g., more points needed for texture-less wall surfaces as they affect sound path diversion). Extensive experiments validate the superiority of our AV-GS over existing alternatives on the real-world RWAS and simulation-based SoundSpaces datasets. Project page: \url{https://surrey-uplab.github.io/research/avgs/}

We tackle the efficiency problem of learning local feature matching.Recent advancements have given rise to purely CNN-based and transformer-based approaches, each augmented with deep learning techniques. While CNN-based methods often excel in matching speed, transformer-based methods tend to provide more accurate matches. We propose an efficient transformer-based network architecture for local feature matching.This technique is built on constructing multiple homography hypotheses to approximate the continuous correspondence in the real world and uni-directional cross-attention to accelerate the refinement. On the YFCC100M dataset, our matching accuracy is competitive with LoFTR, a state-of-the-art transformer-based architecture, while the inference speed is boosted to 4 times, even outperforming the CNN-based methods.Comprehensive evaluations on other open datasets such as Megadepth, ScanNet, and HPatches demonstrate our method's efficacy, highlighting its potential to significantly enhance a wide array of downstream applications.

The proliferation of 2D foundation models has sparked research into adapting them for open-world 3D instance segmentation. Recent methods introduce a paradigm that leverages superpoints as geometric primitives and incorporates 2D multi-view masks from Segment Anything model (SAM) as merging guidance, achieving outstanding zero-shot instance segmentation results. However, the limited use of 3D priors restricts the segmentation performance. Previous methods calculate the 3D superpoints solely based on estimated normal from spatial coordinates, resulting in under-segmentation for instances with similar geometry. Besides, the heavy reliance on SAM and hand-crafted algorithms in 2D space suffers from over-segmentation due to SAM's inherent part-level segmentation tendency. To address these issues, we propose SA3DIP, a novel method for Segmenting Any 3D Instances via exploiting potential 3D Priors. Specifically, on one hand, we generate complementary 3D primitives based on both geometric and textural priors, which reduces the initial errors that accumulate in subsequent procedures. On the other hand, we introduce supplemental constraints from the 3D space by using a 3D detector to guide a further merging process. Furthermore, we notice a considerable portion of low-quality ground truth annotations in ScanNetV2 benchmark, which affect the fair evaluations. Thus, we present ScanNetV2-INS with complete ground truth labels and supplement additional instances for 3D class-agnostic instance segmentation. Experimental evaluations on various 2D-3D datasets demonstrate the effectiveness and robustness of our approach. Our code and proposed ScanNetV2-INS dataset are available HERE.

Over the past years, we have observed an abundance of approaches for modeling dynamic 3D scenes using Gaussian Splatting (GS). These solutions use GS to represent the scene's structure and the neural network to model dynamics. Such approaches allow fast rendering and extracting each element of such a dynamic scene. However, modifying such objects over time is challenging. SC-GS (Sparse Controlled Gaussian Splatting) enhanced with Deformed Control Points partially solves this issue. However, this approach necessitates selecting elements that need to be kept fixed, as well as centroids that should be adjusted throughout editing. Moreover, this task poses additional difficulties regarding the re-productivity of such editing. To address this, we propose Dynamic Multi-Gaussian Soup (D-MiSo), which allows us to model the mesh-inspired representation of dynamic GS. Additionally, we propose a strategy of linking parameterized Gaussian splats, forming a Triangle Soup with the estimated mesh. Consequently, we can separately construct new trajectories for the 3D objects composing the scene. Thus, we can make the scene's dynamic editable over time or while maintaining partial dynamics.

3D Gaussian Splatting (3DGS) has already become the emerging research focus in the fields of 3D scene reconstruction and novel view synthesis. Given that training a 3DGS requires a significant amount of time and computational cost, it is crucial to protect the copyright, integrity, and privacy of such 3D assets. Steganography, as a crucial technique for encrypted transmission and copyright protection, has been extensively studied. However, it still lacks profound exploration targeted at 3DGS. Unlike its predecessor NeRF, 3DGS possesses two distinct features: 1) explicit 3D representation; and 2) real-time rendering speeds. These characteristics result in the 3DGS point cloud files being public and transparent, with each Gaussian point having a clear physical significance. Therefore, ensuring the security and fidelity of the original 3D scene while embedding information into the 3DGS point cloud files is an extremely challenging task. To solve the above-mentioned issue, we first propose a steganography framework for 3DGS, dubbed GS-Hider, which can embed 3D scenes and images into original GS point clouds in an invisible manner and accurately extract the hidden messages. Specifically, we design a coupled secured feature attribute to replace the original 3DGS's spherical harmonics coefficients and then use a scene decoder and a message decoder to disentangle the original RGB scene and the hidden message. Extensive experiments demonstrated that the proposed GS-Hider can effectively conceal multimodal messages without compromising rendering quality and possesses exceptional security, robustness, capacity, and flexibility. Our project is available at: https://xuanyuzhang21.github.io/project/gshider.

This paper introduces OpenGaussian, a method based on 3D Gaussian Splatting (3DGS) that possesses the capability for 3D point-level open vocabulary understanding. Our primary motivation stems from observing that existing 3DGS-based open vocabulary methods mainly focus on 2D pixel-level parsing. These methods struggle with 3D point-level tasks due to weak feature expressiveness and inaccurate 2D-3D feature associations. To ensure robust feature presentation and 3D point-level understanding, we first employ SAM masks without cross-frame associations to train instance features with 3D consistency. These features exhibit both intra-object consistency and inter-object distinction. Then, we propose a two-stage codebook to discretize these features from coarse to fine levels. At the coarse level, we consider the positional information of 3D points to achieve location-based clustering, which is then refined at the fine level. Finally, we introduce an instance-level 3D-2D feature association method that links 3D points to 2D masks, which are further associated with 2D CLIP features. Extensive experiments, including open vocabulary-based 3D object selection, 3D point cloud understanding, click-based 3D object selection, and ablation studies, demonstrate the effectiveness of our proposed method. The source code is available at our project page https://3d-aigc.github.io/OpenGaussian.

3D reconstruction and relighting of objects made from scattering materials present a significant challenge due to the complex light transport beneath the surface. 3D Gaussian Splatting introduced high-quality novel view synthesis at real-time speeds. While 3D Gaussians efficiently approximate an object's surface, they fail to capture the volumetric properties of subsurface scattering. We propose a framework for optimizing an object's shape together with the radiance transfer field given multi-view OLAT (one light at a time) data. Our method decomposes the scene into an explicit surface represented as 3D Gaussians, with a spatially varying BRDF, and an implicit volumetric representation of the scattering component. A learned incident light field accounts for shadowing. We optimize all parameters jointly via ray-traced differentiable rendering. Our approach enables material editing, relighting, and novel view synthesis at interactive rates. We show successful application on synthetic data and contribute a newly acquired multi-view multi-light dataset of objects in a light-stage setup. Compared to previous work we achieve comparable or better results at a fraction of optimization and rendering time while enabling detailed control over material attributes.

3D occupancy-based perception pipeline has significantly advanced autonomous driving by capturing detailed scene descriptions and demonstrating strong generalizability across various object categories and shapes. Current methods predominantly rely on LiDAR or camera inputs for 3D occupancy prediction. These methods are susceptible to adverse weather conditions, limiting the all-weather deployment of self-driving cars. To improve perception robustness, we leverage the recent advances in automotive radars and introduce a novel approach that utilizes 4D imaging radar sensors for 3D occupancy prediction. Our method, RadarOcc, circumvents the limitations of sparse radar point clouds by directly processing the 4D radar tensor, thus preserving essential scene details. RadarOcc innovatively addresses the challenges associated with the voluminous and noisy 4D radar data by employing Doppler bins descriptors, sidelobe-aware spatial sparsification, and range-wise self-attention mechanisms. To minimize the interpolation errors associated with direct coordinate transformations, we also devise a spherical-based feature encoding followed by spherical-to-Cartesian feature aggregation. We benchmark various baseline methods based on distinct modalities on the public K-Radar dataset. The results demonstrate RadarOcc's state-of-the-art performance in radar-based 3D occupancy prediction and promising results even when compared with LiDAR- or camera-based methods. Additionally, we present qualitative evidence of the superior performance of 4D radar in adverse weather conditions and explore the impact of key pipeline components through ablation studies.

Learning-based simulators show great potential for simulating particle dynamics when 3D groundtruth is available, but per-particle correspondences are not always accessible. The development of neural rendering presents a new solution to this field to learn 3D dynamics from 2D images by inverse rendering. However, existing approaches still suffer from ill-posed natures resulting from the 2D to 3D uncertainty, for example, specific 2D images can correspond with various 3D particle distributions. To mitigate such uncertainty, we consider a conventional, mechanically interpretable framework as the physical priors and extend it to a learning-based version. In brief, we incorporate the learnable graph kernels into the classic Discrete Element Analysis (DEA) framework to implement a novel mechanics-informed network architecture. In this case, the graph networks are only used for approximating some specific mechanical operators in the DEA framework rather than the whole dynamics mapping. By integrating the strong physics priors, our methods can effectively learn the dynamics of various materials from the partial 2D observations in a unified manner. Experiments show that our approach outperforms other learned simulators by a large margin in this context and is robust to different renderers, fewer training samples, and fewer camera views.

3D Gaussian Splatting (3DGS) has demonstrated impressive novel view synthesis performance. While conventional methods require per-scene optimization, more recently several feed-forward methods have been proposed to generate pixel-aligned Gaussian representations with a learnable network, which are generalizable to different scenes. However, these methods simply combine pixel-aligned Gaussians from multiple views as scene representations, thereby leading to artifacts and extra memory cost without fully capturing the relations of Gaussians from different images. In this paper, we propose Gaussian Graph Network (GGN) to generate efficient and generalizable Gaussian representations. Specifically, we construct Gaussian Graphs to model the relations of Gaussian groups from different views. To support message passing at Gaussian level, we reformulate the basic graph operations over Gaussian representations, enabling each Gaussian to benefit from its connected Gaussian groups with Gaussian feature fusion. Furthermore, we design a Gaussian pooling layer to aggregate various Gaussian groups for efficient representations. We conduct experiments on the large-scale RealEstate10K and ACID datasets to demonstrate the efficiency and generalization of our method. Compared to the state-of-the-art methods, our model uses fewer Gaussians and achieves better image quality with higher rendering speed.

Cryo-EM is an increasingly popular method for determining the atomic resolution 3D structure of macromolecular complexes (eg, proteins) from noisy 2D images captured by an electron microscope. The computational task is to reconstruct the 3D density of the particle, along with 3D pose of the particle in each 2D image, for which the posterior pose distribution is highly multi-modal. Recent developments in cryo-EM have focused on deep learning for which amortized inference has been used to predict pose. Here, we address key problems with this approach, and propose a new semi-amortized method, cryoSPIN, in which reconstruction begins with amortized inference and then switches to a form of auto-decoding to refine poses locally using stochastic gradient descent. Through evaluation on synthetic datasets, we demonstrate that cryoSPIN is able to handle multi-modal pose distributions during the amortized inference stage, while the later, more flexible stage of direct pose optimization yields faster and more accurate convergence of poses compared to baselines. On experimental data, we show that cryoSPIN outperforms the state-of-the-art cryoAI in speed and reconstruction quality.

A proper scene representation is central to the pursuit of spatial intelligence where agents can robustly reconstruct and efficiently understand 3D scenes. A scene representation is either metric, such as landmark maps in 3D reconstruction, 3D bounding boxes in object detection, or voxel grids in occupancy prediction, or topological, such as pose graphs with loop closures in SLAM or visibility graphs in SfM. In this work, we propose to build Multiview Scene Graphs (MSG) from unposed images, representing a scene topologically with interconnected place and object nodes. The task of building MSG is challenging for existing representation learning methods since it needs to jointly address both visual place recognition, object detection, and object association from images with limited fields of view and potentially large viewpoint changes. To evaluate any method tackling this task, we developed an MSG dataset and annotation based on a public 3D dataset. We also propose an evaluation metric based on the intersection-over-union score of MSG edges. Moreover, we develop a novel baseline method built on mainstream pretrained vision models, combining visual place recognition and object association into one Transformer decoder architecture. Experiments demonstrate that our method has superior performance compared to existing relevant baselines.

Novel view synthesis of dynamic scenes is becoming important in various applications, including augmented and virtual reality. We propose a novel 4D Gaussian Splatting (4DGS) algorithm for dynamic scenes from casually recorded monocular videos. To overcome the overfitting problem of existing work for these real-world videos, we introduce an uncertainty-aware regularization that identifies uncertain regions with few observations and selectively imposes additional priors based on diffusion models and depth smoothness on such regions. This approach improves both the performance of novel view synthesis and the quality of training image reconstruction. We also identify the initialization problem of 4DGS in fast-moving dynamic regions, where the Structure from Motion (SfM) algorithm fails to provide reliable 3D landmarks. To initialize Gaussian primitives in such regions, we present a dynamic region densification method using the estimated depth maps and scene flow. Our experiments show that the proposed method improves the performance of 4DGS reconstruction from a video captured by a handheld monocular camera and also exhibits promising results in few-shot static scene reconstruction.

Multi-instance point cloud registration aims to estimate the pose of all instances of a model point cloud in the whole scene. Existing methods all adopt the strategy of first obtaining the global correspondence and then clustering to obtain the pose of each instance. However, due to the cluttered and occluded objects in the scene, it is difficult to obtain an accurate correspondence between the model point cloud and all instances in the scene. To this end, we propose a simple yet powerful 3D focusing-and-matching network for multi-instance point cloud registration by learning the multiple pair-wise point cloud registration. Specifically, we first present a 3D multi-object focusing module to locate the center of each object and generate object proposals. By using self-attention and cross-attention to associate the model point cloud with structurally similar objects, we can locate potential matching instances by regressing object centers. Then, we propose a 3D dual-masking instance matching module to estimate the pose between the model point cloud and each object proposal. It performs instance mask and overlap mask masks to accurately predict the pair-wise correspondence. Extensive experiments on two public benchmarks, Scan2CAD and ROBI, show that our method achieves a new state-of-the-art performance on the multi-instance point cloud registration task.

Our objective in this paper is to probe large vision models to determine to what extent they ‘understand’ different physical properties of the 3D scene depicted in an image. To this end, we make the following contributions: (i) We introduce a general and lightweight protocol to evaluate whether features of an off-the-shelf large vision model encode a number of physical ‘properties’ of the 3D scene, by training discriminative classifiers on the features for these properties. The probes are applied on datasets of real images with annotations for the property. (ii) We apply this protocol to properties covering scene geometry, scene material, support relations, lighting, and view-dependent measures, and large vision models including CLIP, DINOv1, DINOv2, VQGAN, Stable Diffusion. (iii) We find that features from Stable Diffusion and DINOv2 are good for discriminative learning of a number of properties, including scene geometry, support relations, shadows and depth, but less performant for occlusion and material, while outperforming DINOv1, CLIP and VQGAN for all properties. (iv) It is observed that different time steps of Stable Diffusion features, as well as different transformer layers of DINO/CLIP/VQGAN, are good at different properties, unlocking potential applications of 3D physical understanding.