Qingshan Xu

Qiancheng Fu, Qingshan Xu, Yew-Soon Ong et al.

Recently, neural implicit surfaces learning by volume rendering has become popular for multi-view reconstruction. However, one key challenge remains: existing approaches lack explicit multi-view geometry constraints, hence usually fail to generate geometry consistent surface reconstruction. To address this challenge, we propose geometry-consistent neural implicit surfaces learning for multi-view reconstruction. We theoretically analyze that there exists a gap between the volume rendering integral and point-based signed distance function (SDF) modeling. To bridge this gap, we directly locate the zero-level set of SDF networks and explicitly perform multi-view geometry optimization by leveraging the sparse geometry from structure from motion (SFM) and photometric consistency in multi-view stereo. This makes our SDF optimization unbiased and allows the multi-view geometry constraints to focus on the true surface optimization. Extensive experiments show that our proposed method achieves high-quality surface reconstruction in both complex thin structures and large smooth regions, thus outperforming the state-of-the-arts by a large margin.

Chunlin Ren, Qingshan Xu, Shikun Zhang et al. · tsinghua

As a fundamental problem in computer vision, multi-view stereo (MVS) aims at recovering the 3D geometry of a target from a set of 2D images. Recent advances in MVS have shown that it is important to perceive non-local structured information for recovering geometry in low-textured areas. In this work, we propose a Hierarchical Prior Mining for Non-local Multi-View Stereo (HPM-MVS). The key characteristics are the following techniques that exploit non-local information to assist MVS: 1) A Non-local Extensible Sampling Pattern (NESP), which is able to adaptively change the size of sampled areas without becoming snared in locally optimal solutions. 2) A new approach to leverage non-local reliable points and construct a planar prior model based on K-Nearest Neighbor (KNN), to obtain potential hypotheses for the regions where prior construction is challenging. 3) A Hierarchical Prior Mining (HPM) framework, which is used to mine extensive non-local prior information at different scales to assist 3D model recovery, this strategy can achieve a considerable balance between the reconstruction of details and low-textured areas. Experimental results on the ETH3D and Tanks \& Temples have verified the superior performance and strong generalization capability of our method. Our code will be released.

Yuan Zhou, Yongzhi Li, Yanqi Dai et al.

Video-driven human reaction generation aims to synthesize 3D human motions that directly react to observed video sequences, which is crucial for building human-like interactive AI systems. However, existing methods often fail to effectively leverage video inputs to steer human reaction synthesis, resulting in reaction motions that are mismatched with the content of video sequences. We reveal that this limitation arises from a severe relational distortion between visual observations and reaction types. In light of this, we propose MuSteerNet, a simple yet effective framework that generates 3D human reactions from videos via observation-reaction mutual steering. Specifically, we first propose a Prototype Feedback Steering mechanism to mitigate relational distortion by refining visual observations with a gated delta-rectification modulator and a relational margin constraint, guided by prototypical vectors learned from human reactions. We then introduce Dual-Coupled Reaction Refinement that fully leverages rectified visual cues to further steer the refinement of generated reaction motions, thereby effectively improving reaction quality and enabling MuSteerNet to achieve competitive performance. Extensive experiments and ablation studies validate the effectiveness of our method. Code coming soon: https://github.com/zhouyuan888888/MuSteerNet.

Chen Zhang, Wanjuan Su, Qingshan Xu et al.

Recently, learning multi-view neural surface reconstruction with the supervision of point clouds or depth maps has been a promising way. However, due to the underutilization of prior information, current methods still struggle with the challenges of limited accuracy and excessive time complexity. In addition, prior data perturbation is also an important but rarely considered issue. To address these challenges, we propose a novel point-guided method named PG-NeuS, which achieves accurate and efficient reconstruction while robustly coping with point noise. Specifically, aleatoric uncertainty of the point cloud is modeled to capture the distribution of noise, leading to noise robustness. Furthermore, a Neural Projection module connecting points and images is proposed to add geometric constraints to implicit surface, achieving precise point guidance. To better compensate for geometric bias between volume rendering and point modeling, high-fidelity points are filtered into a Bias Network to further improve details representation. Benefiting from the effective point guidance, even with a lightweight network, the proposed PG-NeuS achieves fast convergence with an impressive 11x speedup compared to NeuS. Extensive experiments show that our method yields high-quality surfaces with high efficiency, especially for fine-grained details and smooth regions, outperforming the state-of-the-art methods. Moreover, it exhibits strong robustness to noisy data and sparse data.

Xuanyu Yi, Zike Wu, Qingshan Xu et al.

Score distillation sampling~(SDS) has been widely adopted to overcome the absence of unseen views in reconstructing 3D objects from a \textbf{single} image. It leverages pre-trained 2D diffusion models as teacher to guide the reconstruction of student 3D models. Despite their remarkable success, SDS-based methods often encounter geometric artifacts and texture saturation. We find out the crux is the overlooked indiscriminate treatment of diffusion time-steps during optimization: it unreasonably treats the student-teacher knowledge distillation to be equal at all time-steps and thus entangles coarse-grained and fine-grained modeling. Therefore, we propose the Diffusion Time-step Curriculum one-image-to-3D pipeline (DTC123), which involves both the teacher and student models collaborating with the time-step curriculum in a coarse-to-fine manner. Extensive experiments on NeRF4, RealFusion15, GSO and Level50 benchmark demonstrate that DTC123 can produce multi-view consistent, high-quality, and diverse 3D assets. Codes and more generation demos will be released in https://github.com/yxymessi/DTC123.

Zhuang Xiong, Chen Zhang, Qingshan Xu et al.

Despite recent progress in calibration-free monocular SLAM via 3D vision foundation models, scale drift remains severe on long sequences. Motion-agnostic partitioning breaks contextual coherence and causes zero-motion drift, while conventional geometric alignment is computationally expensive. To address these issues, we propose VGGT-Motion, a calibration-free SLAM system for efficient and robust global consistency over kilometer-scale trajectories. Specifically, we first propose a motion-aware submap construction mechanism that uses optical flow to guide adaptive partitioning, prune static redundancy, and encapsulate turns for stable local geometry. We then design an anchor-driven direct Sim(3) registration strategy. By exploiting context-balanced anchors, it achieves search-free, pixel-wise dense alignment and efficient loop closure without costly feature matching. Finally, a lightweight submap-level pose graph optimization enforces global consistency with linear complexity, enabling scalable long-range operation. Experiments show that VGGT-Motion markedly improves trajectory accuracy and efficiency, achieving state-of-the-art performance in zero-shot, long-range calibration-free monocular SLAM.

Jiequan Cui, Beier Zhu, Qingshan Xu et al.

In this paper, we delve deeper into the Kullback-Leibler (KL) Divergence loss and mathematically prove that it is equivalent to the Decoupled Kullback-Leibler (DKL) Divergence loss that consists of (1) a weighted Mean Square Error (wMSE) loss and (2) a Cross-Entropy loss incorporating soft labels. Thanks to the decoupled structure of DKL loss, we have identified two areas for improvement. Firstly, we address the limitation of KL loss in scenarios like knowledge distillation by breaking its asymmetric optimization property along with a smoother weight function. This modification effectively alleviates convergence challenges in optimization, particularly for classes with high predicted scores in soft labels. Secondly, we introduce class-wise global information into KL/DKL to reduce bias arising from individual samples. With these two enhancements, we derive the Generalized Kullback-Leibler (GKL) Divergence loss and evaluate its effectiveness by conducting experiments on CIFAR-10/100, ImageNet, and vision-language datasets, focusing on adversarial training, and knowledge distillation tasks. Specifically, we achieve new state-of-the-art adversarial robustness on the public leaderboard -- RobustBench and competitive knowledge distillation performance across CIFAR/ImageNet models and CLIP models, demonstrating the substantial practical merits. Our code is available at https://github.com/jiequancui/DKL.

Jianyao Xu, Qingshan Xu, Xinyao Liao et al.

Neural surfaces learning has shown impressive performance in multi-view surface reconstruction. However, most existing methods use large multilayer perceptrons (MLPs) to train their models from scratch, resulting in hours of training for a single scene. Recently, how to accelerate the neural surfaces learning has received a lot of attention and remains an open problem. In this work, we propose a prior-based residual learning paradigm for fast multi-view neural surface reconstruction. This paradigm consists of two optimization stages. In the first stage, we propose to leverage generalization models to generate a basis signed distance function (SDF) field. This initial field can be quickly obtained by fusing multiple local SDF fields produced by generalization models. This provides a coarse global geometry prior. Based on this prior, in the second stage, a fast residual learning strategy based on hash-encoding networks is proposed to encode an offset SDF field for the basis SDF field. Moreover, we introduce a prior-guided sampling scheme to help the residual learning stage converge better, and thus recover finer structures. With our designed paradigm, experimental results show that our method only takes about 3 minutes to reconstruct the surface of a single scene, while achieving competitive surface quality. Our code will be released upon publication.

Qingshan Xu, Jiao Liu, Shangshu Yu et al.

In this paper, we aim to create physical digital twins of deformable objects under interaction. Existing methods focus more on the physical learning of current state modeling, but generalize worse to future prediction. This is because existing methods ignore the intrinsic physical properties of deformable objects, resulting in the limited physical learning in the current state modeling. To address this, we present NeuSpring, a neural spring field for the reconstruction and simulation of deformable objects from videos. Built upon spring-mass models for realistic physical simulation, our method consists of two major innovations: 1) a piecewise topology solution that efficiently models multi-region spring connection topologies using zero-order optimization, which considers the material heterogeneity of real-world objects. 2) a neural spring field that represents spring physical properties across different frames using a canonical coordinate-based neural network, which effectively leverages the spatial associativity of springs for physical learning. Experiments on real-world datasets demonstrate that our NeuSping achieves superior reconstruction and simulation performance for current state modeling and future prediction, with Chamfer distance improved by 20% and 25%, respectively.

Yuxuan Wang, Xuanyu Yi, Qingshan Xu et al.

Personalizing 3D scenes from a single reference image enables intuitive user-guided editing, which requires achieving both multi-view consistency across perspectives and referential consistency with the input image. However, these goals are particularly challenging due to the viewpoint bias caused by the limited perspective provided in a single image. Lacking the mechanisms to effectively expand reference information beyond the original view, existing methods of image-conditioned 3DGS personalization often suffer from this viewpoint bias and struggle to produce consistent results. Therefore, in this paper, we present Consistent Personalization for 3D Gaussian Splatting (CP-GS), a framework that progressively propagates the single-view reference appearance to novel perspectives. In particular, CP-GS integrates pre-trained image-to-3D generation and iterative LoRA fine-tuning to extract and extend the reference appearance, and finally produces faithful multi-view guidance images and the personalized 3DGS outputs through a view-consistent generation process guided by geometric cues. Extensive experiments on real-world scenes show that our CP-GS effectively mitigates the viewpoint bias, achieving high-quality personalization that significantly outperforms existing methods. The code will be released at https://github.com/Yuxuan-W/CP-GS.

Fangjinhua Wang, Qingshan Xu, Yew-Soon Ong et al.

To reconstruct the 3D geometry from calibrated images, learning-based multi-view stereo (MVS) methods typically perform multi-view depth estimation and then fuse depth maps into a mesh or point cloud. To improve the computational efficiency, many methods initialize a coarse depth map and then gradually refine it in higher resolutions. Recently, diffusion models achieve great success in generation tasks. Starting from a random noise, diffusion models gradually recover the sample with an iterative denoising process. In this paper, we propose a novel MVS framework, which introduces diffusion models in MVS. Specifically, we formulate depth refinement as a conditional diffusion process. Considering the discriminative characteristic of depth estimation, we design a condition encoder to guide the diffusion process. To improve efficiency, we propose a novel diffusion network combining lightweight 2D U-Net and convolutional GRU. Moreover, we propose a novel confidence-based sampling strategy to adaptively sample depth hypotheses based on the confidence estimated by diffusion model. Based on our novel MVS framework, we propose two novel MVS methods, DiffMVS and CasDiffMVS. DiffMVS achieves competitive performance with state-of-the-art efficiency in run-time and GPU memory. CasDiffMVS achieves state-of-the-art performance on DTU, Tanks & Temples and ETH3D. Code is available at: https://github.com/cvg/diffmvs.

Kezheng Xiong, Maoji Zheng, Qingshan Xu et al.

Point cloud registration, a fundamental task in 3D computer vision, has remained largely unexplored in cross-source point clouds and unstructured scenes. The primary challenges arise from noise, outliers, and variations in scale and density. However, neglected geometric natures of point clouds restricts the performance of current methods. In this paper, we propose a novel method termed SPEAL to leverage skeletal representations for effective learning of intrinsic topologies of point clouds, facilitating robust capture of geometric intricacy. Specifically, we design the Skeleton Extraction Module to extract skeleton points and skeletal features in an unsupervised manner, which is inherently robust to noise and density variances. Then, we propose the Skeleton-Aware GeoTransformer to encode high-level skeleton-aware features. It explicitly captures the topological natures and inter-point-cloud skeletal correlations with the noise-robust and density-invariant skeletal representations. Next, we introduce the Correspondence Dual-Sampler to facilitate correspondences by augmenting the correspondence set with skeletal correspondences. Furthermore, we construct a challenging novel large-scale cross-source point cloud dataset named KITTI CrossSource for benchmarking cross-source point cloud registration methods. Extensive quantitative and qualitative experiments are conducted to demonstrate our approach's superiority and robustness on both cross-source and same-source datasets. To the best of our knowledge, our approach is the first to facilitate point cloud registration with skeletal geometric priors.

Hao Fei, Yuan Zhou, Juncheng Li et al.

The Multimodal Large Language Model (MLLM) is currently experiencing rapid growth, driven by the advanced capabilities of LLMs. Unlike earlier specialists, existing MLLMs are evolving towards a Multimodal Generalist paradigm. Initially limited to understanding multiple modalities, these models have advanced to not only comprehend but also generate across modalities. Their capabilities have expanded from coarse-grained to fine-grained multimodal understanding and from supporting limited modalities to arbitrary ones. While many benchmarks exist to assess MLLMs, a critical question arises: Can we simply assume that higher performance across tasks indicates a stronger MLLM capability, bringing us closer to human-level AI? We argue that the answer is not as straightforward as it seems. This project introduces General-Level, an evaluation framework that defines 5-scale levels of MLLM performance and generality, offering a methodology to compare MLLMs and gauge the progress of existing systems towards more robust multimodal generalists and, ultimately, towards AGI. At the core of the framework is the concept of Synergy, which measures whether models maintain consistent capabilities across comprehension and generation, and across multiple modalities. To support this evaluation, we present General-Bench, which encompasses a broader spectrum of skills, modalities, formats, and capabilities, including over 700 tasks and 325,800 instances. The evaluation results that involve over 100 existing state-of-the-art MLLMs uncover the capability rankings of generalists, highlighting the challenges in reaching genuine AI. We expect this project to pave the way for future research on next-generation multimodal foundation models, providing a robust infrastructure to accelerate the realization of AGI. Project page: https://generalist.top/

Qingshan Xu, Xuanyu Yi, Jianyao Xu et al.

Novel view synthesis with sparse inputs poses great challenges to Neural Radiance Field (NeRF). Recent works demonstrate that the frequency regularization of Positional Encoding (PE) can achieve promising results for few-shot NeRF. In this work, we reveal that there exists an inconsistency between the frequency regularization of PE and rendering loss. This prevents few-shot NeRF from synthesizing higher-quality novel views. To mitigate this inconsistency, we propose Adaptive Rendering loss regularization for few-shot NeRF, dubbed AR-NeRF. Specifically, we present a two-phase rendering supervision and an adaptive rendering loss weight learning strategy to align the frequency relationship between PE and 2D-pixel supervision. In this way, AR-NeRF can learn global structures better in the early training phase and adaptively learn local details throughout the training process. Extensive experiments show that our AR-NeRF achieves state-of-the-art performance on different datasets, including object-level and complex scenes.

Yuxuan Wang, Xuanyu Yi, Haohan Weng et al.

Triangle meshes are fundamental to 3D applications, enabling efficient modification and rasterization while maintaining compatibility with standard rendering pipelines. However, current automatic mesh generation methods typically rely on intermediate representations that lack the continuous surface quality inherent to meshes. Converting these representations into meshes produces dense, suboptimal outputs. Although recent autoregressive approaches demonstrate promise in directly modeling mesh vertices and faces, they are constrained by the limitation in face count, scalability, and structural fidelity. To address these challenges, we propose Nautilus, a locality-aware autoencoder for artist-like mesh generation that leverages the local properties of manifold meshes to achieve structural fidelity and efficient representation. Our approach introduces a novel tokenization algorithm that preserves face proximity relationships and compresses sequence length through locally shared vertices and edges, enabling the generation of meshes with an unprecedented scale of up to 5,000 faces. Furthermore, we develop a Dual-stream Point Conditioner that provides multi-scale geometric guidance, ensuring global consistency and local structural fidelity by capturing fine-grained geometric features. Extensive experiments demonstrate that Nautilus significantly outperforms state-of-the-art methods in both fidelity and scalability. The project page is at https://nautilusmeshgen.github.io.

Wanjuan Su, Chen Zhang, Qingshan Xu et al.

Surface reconstruction has traditionally relied on the Multi-View Stereo (MVS)-based pipeline, which often suffers from noisy and incomplete geometry. This is due to that although MVS has been proven to be an effective way to recover the geometry of the scenes, especially for locally detailed areas with rich textures, it struggles to deal with areas with low texture and large variations of illumination where the photometric consistency is unreliable. Recently, Neural Implicit Surface Reconstruction (NISR) combines surface rendering and volume rendering techniques and bypasses the MVS as an intermediate step, which has emerged as a promising alternative to overcome the limitations of traditional pipelines. While NISR has shown impressive results on simple scenes, it remains challenging to recover delicate geometry from uncontrolled real-world scenes which is caused by its underconstrained optimization. To this end, the framework PSDF is proposed which resorts to external geometric priors from a pretrained MVS network and internal geometric priors inherent in the NISR model to facilitate high-quality neural implicit surface learning. Specifically, the visibility-aware feature consistency loss and depth prior-assisted sampling based on external geometric priors are introduced. These proposals provide powerfully geometric consistency constraints and aid in locating surface intersection points, thereby significantly improving the accuracy and delicate reconstruction of NISR. Meanwhile, the internal prior-guided importance rendering is presented to enhance the fidelity of the reconstructed surface mesh by mitigating the biased rendering issue in NISR. Extensive experiments on the Tanks and Temples dataset show that PSDF achieves state-of-the-art performance on complex uncontrolled scenes.

Qingshan Xu, Jiequan Cui, Xuanyu Yi et al.

3D Gaussian Splatting (3DGS) has demonstrated impressive Novel View Synthesis (NVS) results in a real-time rendering manner. During training, it relies heavily on the average magnitude of view-space positional gradients to grow Gaussians to reduce rendering loss. However, this average operation smooths the positional gradients from different viewpoints and rendering errors from different pixels, hindering the growth and optimization of many defective Gaussians. This leads to strong spurious artifacts in some areas. To address this problem, we propose Hard Gaussian Splatting, dubbed HGS, which considers multi-view significant positional gradients and rendering errors to grow hard Gaussians that fill the gaps of classical Gaussian Splatting on 3D scenes, thus achieving superior NVS results. In detail, we present positional gradient driven HGS, which leverages multi-view significant positional gradients to uncover hard Gaussians. Moreover, we propose rendering error guided HGS, which identifies noticeable pixel rendering errors and potentially over-large Gaussians to jointly mine hard Gaussians. By growing and optimizing these hard Gaussians, our method helps to resolve blurring and needle-like artifacts. Experiments on various datasets demonstrate that our method achieves state-of-the-art rendering quality while maintaining real-time efficiency.

Yuan Zhou, Junbao Zhou, Qingshan Xu et al.

Drag-Based Image Editing (DBIE), which allows users to manipulate images by directly dragging objects within them, has recently attracted much attention from the community. However, it faces two key challenges: (\emph{\textcolor{magenta}{i}}) point-based drag is often highly ambiguous and difficult to align with users' intentions; (\emph{\textcolor{magenta}{ii}}) current DBIE methods primarily rely on alternating between motion supervision and point tracking, which is not only cumbersome but also fails to produce high-quality results. These limitations motivate us to explore DBIE from a new perspective -- redefining it as deformation, rotation, and translation of user-specified handle regions. Thereby, by requiring users to explicitly specify both drag areas and types, we can effectively address the ambiguity issue. Furthermore, we propose a simple-yet-effective editing framework, dubbed \textcolor{SkyBlue}{\textbf{DragNeXt}}. It unifies DBIE as a Latent Region Optimization (LRO) problem and solves it through Progressive Backward Self-Intervention (PBSI), simplifying the overall procedure of DBIE while further enhancing quality by fully leveraging region-level structure information and progressive guidance from intermediate drag states. We validate \textcolor{SkyBlue}{\textbf{DragNeXt}} on our NextBench, and extensive experiments demonstrate that our proposed method can significantly outperform existing approaches. Code will be released on github.

Hao Wang, Qingshan Xu, Hongyuan Chen et al.

Recent techniques on implicit geometry representation learning and neural rendering have shown promising results for 3D clothed human reconstruction from sparse video inputs. However, it is still challenging to reconstruct detailed surface geometry and even more difficult to synthesize photorealistic novel views with animated human poses. In this work, we introduce PGAHum, a prior-guided geometry and appearance learning framework for high-fidelity animatable human reconstruction. We thoroughly exploit 3D human priors in three key modules of PGAHum to achieve high-quality geometry reconstruction with intricate details and photorealistic view synthesis on unseen poses. First, a prior-based implicit geometry representation of 3D human, which contains a delta SDF predicted by a tri-plane network and a base SDF derived from the prior SMPL model, is proposed to model the surface details and the body shape in a disentangled manner. Second, we introduce a novel prior-guided sampling strategy that fully leverages the prior information of the human pose and body to sample the query points within or near the body surface. By avoiding unnecessary learning in the empty 3D space, the neural rendering can recover more appearance details. Last, we propose a novel iterative backward deformation strategy to progressively find the correspondence for the query point in observation space. A skinning weights prediction model is learned based on the prior provided by the SMPL model to achieve the iterative backward LBS deformation. Extensive quantitative and qualitative comparisons on various datasets are conducted and the results demonstrate the superiority of our framework. Ablation studies also verify the effectiveness of each scheme for geometry and appearance learning.

Junbao Zhou, Yuan Zhou, Kesen Zhao et al.

Achieving streaming, fine-grained control over the outputs of autoregressive video diffusion models remains challenging, making it difficult to ensure that they consistently align with user expectations. To bridge this gap, we propose \textbf{stReaming drag-oriEnted interactiVe vidEo manipuLation (REVEL)}, a new task that enables users to modify generated videos \emph{anytime} on \emph{anything} via fine-grained, interactive drag. Beyond DragVideo and SG-I2V, REVEL unifies drag-style video manipulation as editing and animating video frames with both supporting user-specified translation, deformation, and rotation effects, making drag operations versatile. In resolving REVEL, we observe: \emph{i}) drag-induced perturbations accumulate in latent space, causing severe latent distribution drift that halts the drag process; \emph{ii}) streaming drag is easily disturbed by context frames, thereby yielding visually unnatural outcomes. We thus propose a training-free approach, \textbf{DragStream}, comprising: \emph{i}) an adaptive distribution self-rectification strategy that leverages neighboring frames' statistics to effectively constrain the drift of latent embeddings; \emph{ii}) a spatial-frequency selective optimization mechanism, allowing the model to fully exploit contextual information while mitigating its interference via selectively propagating visual cues along generation. Our method can be seamlessly integrated into existing autoregressive video diffusion models, and extensive experiments firmly demonstrate the effectiveness of our DragStream.

Jiequan Cui, Beier Zhu, Qingshan Xu et al.

In this paper, we formulate the knowledge distillation (KD) as a conditional generative problem and propose the \textit{Generative Distribution Distillation (GenDD)} framework. A naive \textit{GenDD} baseline encounters two major challenges: the curse of high-dimensional optimization and the lack of semantic supervision from labels. To address these issues, we introduce a \textit{Split Tokenization} strategy, achieving stable and effective unsupervised KD. Additionally, we develop the \textit{Distribution Contraction} technique to integrate label supervision into the reconstruction objective. Our theoretical proof demonstrates that \textit{GenDD} with \textit{Distribution Contraction} serves as a gradient-level surrogate for multi-task learning, realizing efficient supervised training without explicit classification loss on multi-step sampling image representations. To evaluate the effectiveness of our method, we conduct experiments on balanced, imbalanced, and unlabeled data. Experimental results show that \textit{GenDD} performs competitively in the unsupervised setting, significantly surpassing KL baseline by \textbf{16.29\%} on ImageNet validation set. With label supervision, our ResNet-50 achieves \textbf{82.28\%} top-1 accuracy on ImageNet in 600 epochs training, establishing a new state-of-the-art.

Yuan Zhou, Qingshan Xu, Jiequan Cui et al.

Recently, large efforts have been made to design efficient linear-complexity visual Transformers. However, current linear attention models are generally unsuitable to be deployed in resource-constrained mobile devices, due to suffering from either few efficiency gains or significant accuracy drops. In this paper, we propose a new de\textbf{C}oupled du\textbf{A}l-interactive linea\textbf{R} att\textbf{E}ntion (CARE) mechanism, revealing that features' decoupling and interaction can fully unleash the power of linear attention. We first propose an asymmetrical feature decoupling strategy that asymmetrically decouples the learning process for local inductive bias and long-range dependencies, thereby preserving sufficient local and global information while effectively enhancing the efficiency of models. Then, a dynamic memory unit is employed to maintain critical information along the network pipeline. Moreover, we design a dual interaction module to effectively facilitate interaction between local inductive bias and long-range information as well as among features at different layers. By adopting a decoupled learning way and fully exploiting complementarity across features, our method can achieve both high efficiency and accuracy. Extensive experiments on ImageNet-1K, COCO, and ADE20K datasets demonstrate the effectiveness of our approach, e.g., achieving $78.4/82.1\%$ top-1 accuracy on ImagegNet-1K at the cost of only $0.7/1.9$ GMACs. Codes will be released on \href{..}{github}.

Kezheng Xiong, Haoen Xiang, Qingshan Xu et al.

Point cloud registration, a fundamental task in 3D vision, has achieved remarkable success with learning-based methods in outdoor environments. Unsupervised outdoor point cloud registration methods have recently emerged to circumvent the need for costly pose annotations. However, they fail to establish reliable optimization objectives for unsupervised training, either relying on overly strong geometric assumptions, or suffering from poor-quality pseudo-labels due to inadequate integration of low-level geometric and high-level contextual information. We have observed that in the feature space, latent new inlier correspondences tend to cluster around respective positive anchors that summarize features of existing inliers. Motivated by this observation, we propose a novel unsupervised registration method termed INTEGER to incorporate high-level contextual information for reliable pseudo-label mining. Specifically, we propose the Feature-Geometry Coherence Mining module to dynamically adapt the teacher for each mini-batch of data during training and discover reliable pseudo-labels by considering both high-level feature representations and low-level geometric cues. Furthermore, we propose Anchor-Based Contrastive Learning to facilitate contrastive learning with anchors for a robust feature space. Lastly, we introduce a Mixed-Density Student to learn density-invariant features, addressing challenges related to density variation and low overlap in the outdoor scenario. Extensive experiments on KITTI and nuScenes datasets demonstrate that our INTEGER achieves competitive performance in terms of accuracy and generalizability.

Xuanyu Yi, Zike Wu, Qiuhong Shen et al.

Recent 3D large reconstruction models (LRMs) can generate high-quality 3D content in sub-seconds by integrating multi-view diffusion models with scalable multi-view reconstructors. Current works further leverage 3D Gaussian Splatting as 3D representation for improved visual quality and rendering efficiency. However, we observe that existing Gaussian reconstruction models often suffer from multi-view inconsistency and blurred textures. We attribute this to the compromise of multi-view information propagation in favor of adopting powerful yet computationally intensive architectures (e.g., Transformers). To address this issue, we introduce MVGamba, a general and lightweight Gaussian reconstruction model featuring a multi-view Gaussian reconstructor based on the RNN-like State Space Model (SSM). Our Gaussian reconstructor propagates causal context containing multi-view information for cross-view self-refinement while generating a long sequence of Gaussians for fine-detail modeling with linear complexity. With off-the-shelf multi-view diffusion models integrated, MVGamba unifies 3D generation tasks from a single image, sparse images, or text prompts. Extensive experiments demonstrate that MVGamba outperforms state-of-the-art baselines in all 3D content generation scenarios with approximately only $0.1\times$ of the model size.

Salvatore Esposito, Qingshan Xu, Kacper Kania et al.

We introduce a new generative approach for synthesizing 3D geometry and images from single-view collections. Most existing approaches predict volumetric density to render multi-view consistent images. By employing volumetric rendering using neural radiance fields, they inherit a key limitation: the generated geometry is noisy and unconstrained, limiting the quality and utility of the output meshes. To address this issue, we propose GeoGen, a new SDF-based 3D generative model trained in an end-to-end manner. Initially, we reinterpret the volumetric density as a Signed Distance Function (SDF). This allows us to introduce useful priors to generate valid meshes. However, those priors prevent the generative model from learning details, limiting the applicability of the method to real-world scenarios. To alleviate that problem, we make the transformation learnable and constrain the rendered depth map to be consistent with the zero-level set of the SDF. Through the lens of adversarial training, we encourage the network to produce higher fidelity details on the output meshes. For evaluation, we introduce a synthetic dataset of human avatars captured from 360-degree camera angles, to overcome the challenges presented by real-world datasets, which often lack 3D consistency and do not cover all camera angles. Our experiments on multiple datasets show that GeoGen produces visually and quantitatively better geometry than the previous generative models based on neural radiance fields.

Qingshan Xu, Jiao Liu, Melvin Wong et al.

Text-to-3D generation has shown great promise in generating novel 3D content based on given text prompts. However, existing generative methods mostly focus on geometric or visual plausibility while ignoring precise physics perception for the generated 3D shapes. This greatly hinders the practicality of generated 3D shapes in real-world applications. In this work, we propose Phy3DGen, a precise-physics-driven text-to-3D generation method. By analyzing the solid mechanics of generated 3D shapes, we reveal that the 3D shapes generated by existing text-to-3D generation methods are impractical for real-world applications as the generated 3D shapes do not conform to the laws of physics. To this end, we leverage 3D diffusion models to provide 3D shape priors and design a data-driven differentiable physics layer to optimize 3D shape priors with solid mechanics. This allows us to optimize geometry efficiently and learn precise physics information about 3D shapes at the same time. Experimental results demonstrate that our method can consider both geometric plausibility and precise physics perception, further bridging 3D virtual modeling and precise physical worlds.

Renyi Mao, Qingshan Xu, Peng Zheng et al.

Coordinate-based neural implicit representation or implicit fields have been widely studied for 3D geometry representation or novel view synthesis. Recently, a series of efforts have been devoted to accelerating the speed and improving the quality of the coordinate-based implicit field learning. Instead of learning heavy MLPs to predict the neural implicit values for the query coordinates, neural voxels or grids combined with shallow MLPs have been proposed to achieve high-quality implicit field learning with reduced optimization time. On the other hand, lightweight field representations such as linear grid have been proposed to further improve the learning speed. In this paper, we aim for both fast and high-quality implicit field learning, and propose TaylorGrid, a novel implicit field representation which can be efficiently computed via direct Taylor expansion optimization on 2D or 3D grids. As a general representation, TaylorGrid can be adapted to different implicit fields learning tasks such as SDF learning or NeRF. From extensive quantitative and qualitative comparisons, TaylorGrid achieves a balance between the linear grid and neural voxels, showing its superiority in fast and high-quality implicit field learning.

Qingshan Xu, Martin R. Oswald, Wenbing Tao et al.

In deep multi-view stereo networks, cost regularization is crucial to achieve accurate depth estimation. Since 3D cost volume filtering is usually memory-consuming, recurrent 2D cost map regularization has recently become popular and has shown great potential in reconstructing 3D models of different scales. However, existing recurrent methods only model the local dependencies in the depth domain, which greatly limits the capability of capturing the global scene context along the depth dimension. To tackle this limitation, we propose a novel non-local recurrent regularization network for multi-view stereo, named NR2-Net. Specifically, we design a depth attention module to capture non-local depth interactions within a sliding depth block. Then, the global scene context between different blocks is modeled in a gated recurrent manner. This way, the long-range dependencies along the depth dimension are captured to facilitate the cost regularization. Moreover, we design a dynamic depth map fusion strategy to improve the algorithm robustness. Our method achieves state-of-the-art reconstruction results on both DTU and Tanks and Temples datasets.

Qingshan Xu, Wenbing Tao

Recently, learning-based multi-view stereo methods have achieved promising results. However, they all overlook the visibility difference among different views, which leads to an indiscriminate multi-view similarity definition and greatly limits their performance on datasets with strong viewpoint variations. In this paper, a Pixelwise Visibility-aware multi-view Stereo Network (PVSNet) is proposed for robust dense 3D reconstruction. We present a pixelwise visibility network to learn the visibility information for different neighboring images before computing the multi-view similarity, and then construct an adaptive weighted cost volume with the visibility information. Moreover, we present an anti-noise training strategy that introduces disturbing views during model training to make the pixelwise visibility network more distinguishable to unrelated views, which is different with the existing learning methods that only use two best neighboring views for training. To the best of our knowledge, PVSNet is the first deep learning framework that is able to capture the visibility information of different neighboring views. In this way, our method can be generalized well to different types of datasets, especially the ETH3D high-res benchmark with strong viewpoint variations. Extensive experiments show that PVSNet achieves the state-of-the-art performance on different datasets.

Qingshan Xu, Wenbing Tao

Deep learning has shown to be effective for depth inference in multi-view stereo (MVS). However, the scalability and accuracy still remain an open problem in this domain. This can be attributed to the memory-consuming cost volume representation and inappropriate depth inference. Inspired by the group-wise correlation in stereo matching, we propose an average group-wise correlation similarity measure to construct a lightweight cost volume. This can not only reduce the memory consumption but also reduce the computational burden in the cost volume filtering. Based on our effective cost volume representation, we propose a cascade 3D U-Net module to regularize the cost volume to further boost the performance. Unlike the previous methods that treat multi-view depth inference as a depth regression problem or an inverse depth classification problem, we recast multi-view depth inference as an inverse depth regression task. This allows our network to achieve sub-pixel estimation and be applicable to large-scale scenes. Through extensive experiments on DTU dataset and Tanks and Temples dataset, we show that our proposed network with Correlation cost volume and Inverse DEpth Regression (CIDER), achieves state-of-the-art results, demonstrating its superior performance on scalability and accuracy.

Qingshan Xu, Wenbing Tao

The completeness of 3D models is still a challenging problem in multi-view stereo (MVS) due to the unreliable photometric consistency in low-textured areas. Since low-textured areas usually exhibit strong planarity, planar models are advantageous to the depth estimation of low-textured areas. On the other hand, PatchMatch multi-view stereo is very efficient for its sampling and propagation scheme. By taking advantage of planar models and PatchMatch multi-view stereo, we propose a planar prior assisted PatchMatch multi-view stereo framework in this paper. In detail, we utilize a probabilistic graphical model to embed planar models into PatchMatch multi-view stereo and contribute a novel multi-view aggregated matching cost. This novel cost takes both photometric consistency and planar compatibility into consideration, making it suited for the depth estimation of both non-planar and planar regions. Experimental results demonstrate that our method can efficiently recover the depth information of extremely low-textured areas, thus obtaining high complete 3D models and achieving state-of-the-art performance.

Qingshan Xu, Wenbing Tao

In this paper, we propose an efficient multi-scale geometric consistency guided multi-view stereo method for accurate and complete depth map estimation. We first present our basic multi-view stereo method with Adaptive Checkerboard sampling and Multi-Hypothesis joint view selection (ACMH). It leverages structured region information to sample better candidate hypotheses for propagation and infer the aggregation view subset at each pixel. For the depth estimation of low-textured areas, we further propose to combine ACMH with multi-scale geometric consistency guidance (ACMM) to obtain the reliable depth estimates for low-textured areas at coarser scales and guarantee that they can be propagated to finer scales. To correct the erroneous estimates propagated from the coarser scales, we present a novel detail restorer. Experiments on extensive datasets show our method achieves state-of-the-art performance, recovering the depth estimation not only in low-textured areas but also in details.

Qingshan Xu, Wenbing Tao

In computer vision domain, how to fast and accurately perform multiview stereo (MVS) is still a challenging problem. In this paper we present a fast yet accurate method for 3D dense reconstruction, called AMHMVS, built on the PatchMatch based stereo algorithm. Different from the regular symmetric propagation scheme, our approach adopts an asymmetric checkerboard propagation strategy, which can adaptively make effective hypotheses expand further according to the confidence of current neighbor hypotheses. In order to aggregate visual information from multiple images better, we propose the multi-hypothesis joint view selection for each pixel, which leverages a cost matrix based on the multiple propagated hypotheses to robustly infer an appropriate aggregation subset parallel. Combined with the above two steps, our approach not only has the capacity of massively parallel computation, but also obtains high accuracy and completeness. Experiments on extensive datasets show that our method achieves more accurate and robust results, and runs faster than the competing methods.