Advancements in 3D scene reconstruction have transformed 2D images from the real world into 3D models, producing realistic 3D results from hundreds of input photos. Despite great success in dense-view reconstruction scenarios, rendering a detailed scene from sparse views is still an ill-posed optimization problem, often resulting in artifacts and distortions in unseen areas. In this paper, we propose ReconX, a novel 3D scene reconstruction paradigm that reframes the ambiguous reconstruction problem as a temporal generation task. The key insight is to unleash the strong generative prior of large pre-trained video diffusion models for sparse-view reconstruction. Nevertheless, it is challenging to preserve 3D view consistency when directly generating video frames from pre-trained models. To address this issue, given limited input views, the proposed ReconX first constructs a global point cloud and encodes it into a contextual space as the 3D structure condition. Guided by the condition, the video diffusion model then synthesizes video frames that are detail-preserved and exhibit a high degree of 3D consistency, ensuring the coherence of the scene from various perspectives. Finally, we recover the 3D scene from the generated video through a confidence-aware 3D Gaussian Splatting optimization scheme. Extensive experiments on various real-world datasets show the superiority of ReconX over state-of-the-art methods in terms of quality and generalizability.

In this paper, we tackle the open-set temporal action segmentation task, which aims to identify unknown frames while ensuring accurate segmentation of known actions in the temporal domain. Existing open-set methods struggle with identifying unknown frames due to their indistinguishability against ambiguous known frames during action transitions, resulting in significant performance degradation. To address this, we propose the action distribution flow, which models transitions between action sequences to capture the inherent feature discrepancies between unknown and known frames. Specifically, our method first models the distributions of known actions using the training data, and then interpolates these distributions along the optimal transport path for consecutive actions in the testing videos. By evaluating the likelihood of testing frames against the modeled action distribution flow, our approach effectively identifies unknown frames without requiring additional training or prior knowledge of the unknown data. Extensive experiments on open-set versions of the GTEA, 50Salads, and Breakfast datasets demonstrate the superiority of the proposed method across all evaluation metrics.

This paper proposes an adaptive margin contrastive learning method for 3D semantic segmentation on point clouds. Most existing methods use equally penalized objectives, which ignore the per-point ambiguities and less discriminated features stemming from transition regions. However, as highly ambiguous points may be indistinguishable even for humans, their manually annotated labels are less reliable, and hard constraints over these points would lead to sub-optimal models. To address this, we first design AMContrast3D, a method comprising contrastive learning into an ambiguity estimation framework, tailored to adaptive objectives for individual points based on ambiguity levels. As a result, our method promotes model training, which ensures the correctness of low-ambiguity points while allowing mistakes for high-ambiguity points. As ambiguities are formulated based on position discrepancies across labels, optimization during inference is constrained by the assumption that all unlabeled points are uniformly unambiguous, lacking ambiguity awareness. Inspired by the insight of joint training, we further propose AMContrast3D++ integrating with two branches trained in parallel, where a novel ambiguity prediction module concurrently learns point ambiguities from generated embeddings. To this end, we design a masked refinement mechanism that leverages predicted ambiguities to enable the ambiguous embeddings to be more reliable, thereby boosting segmentation performance and enhancing robustness. Experimental results on 3D indoor scene datasets, S3DIS and ScanNet, demonstrate the effectiveness of the proposed method. Code is available athttps://github.com/YangChenApril/AMContrast3D.

Inherently equipped with arbitrary resolution and multi-view consistency, the Neural Radiance Field (NeRF) as an implicit scene representation has drawn extensive attention. While traditional NeRFs excel at novel view synthesis (NVS) under ideal conditions, they overlook the potential of learning consistent geometric representations across varying sight qualities. Current methods mainly focus on optimizing synthesis under clear visibility, which limits their effectiveness in downstream scene understanding tasks where robust geometry comprehension is crucial. In this paper, we propose a NVS pre-training technique named ShadowNeRF which firstly synthesizes degraded views with shadowed regions to challenge the model in inferring complete scene geometries. We then design a self-supervised sight recovery process with a two-stage unshadowing framework, which progressively recovers neighboring areas and reveals geometric properties of invisible regions. This pre-training strategy of degradation synthesis and recovery, when combined with taskspecific fine-tuning, enhances the understanding of underlying scene structure for the model and strengthens its ability to process scenes under varying sight conditions. Through extensive experiments, we demonstrate that our pre-training and finetuning pipeline significantly improves the model performances in semantic segmentation and 3D object detection, as well as the reconstruction quality of complex scenes.

Recovering 3D scenes from sparse views is a challenging task due to its inherent ill-posed problem. Conventional methods have developed specialized solutions (e.g., geometry regularization or feed-forward deterministic model) to mitigate the issue. However, they still suffer from performance degradation by minimal overlap across input views with insufficient visual information. Fortunately, recent video generative models show promise in addressing this challenge as they are capable of generating video clips with plausible 3D structures. Powered by large pretrained video diffusion models, some pioneering research start to explore the potential of video generative prior and create 3D scenes from sparse views. Despite impressive improvements, they are limited by slow inference time and the lack of 3D constraint, leading to inefficiencies and reconstruction artifacts that do not align with real-world geometry structure. In this paper, we propose VideoScene to distill the video diffusion model to generate 3D scenes in one step, aiming to build an efficient and effective tool to bridge the gap from video to 3D. Specifically, we design a 3D-aware leap flow distillation strategy to leap over time-consuming redundant information and train a dynamic denoising policy network to adaptively determine the optimal leap timestep during inference. Extensive experiments demonstrate that our VideoScene achieves faster and superior 3D scene generation results than previous video diffusion models, highlighting its potential as an efficient tool for future video to 3D applications.

4D reconstruction from a single monocular video is an important but challenging task due to its inherent under-constrained nature. While most existing 4D reconstruction methods focus on multi-camera settings, they always suffer from limited multi-view information in monocular videos. Recent studies have attempted to mitigate the ill-posed problem by incorporating data-driven priors as additional supervision. However, they require hours of optimization to align the splatted 2D feature maps of explicit Gaussians with various priors, which limits the range of applications. To address the time-consuming issue, we propose 4D-Fly, an efficient and effective framework for reconstructing the 4D scene from a monocular video (hundreds of frames within 6 minutes), more than 20 × faster and even achieving higher quality than previous optimization methods. Our key insight is to unleash the explicit property of Gaussian primitives and directly apply data priors to them. Specifically, we build a streaming 4D reconstruction paradigm that includes: propagating existing Gaussian to the next timestep with an anchor-based strategy, expanding the 4D scene map with the canonical Gaussian map, and an efficient 4D scene optimization process to further improve visual quality and motion accuracy. Extensive experiments demonstrate the superiority of our 4D-Fly over state-of-the-art methods in terms of speed and quality. Project page: https://diankun-wu.github.io/4D-Fly/.

In this paper, we propose Scene Splatter, a momentum-based paradigm for video diffusion to generate generic scenes from single image. Existing methods, which employ video generation models to synthesize novel views, suffer from limited video length and scene inconsistency, leading to artifacts and distortions during further reconstruction. To address this issue, we construct noisy samples from original features as momentum to enhance video details and maintain scene consistency. However, for latent features with the perception field that spans both known and unknown regions, such latent-level momentum restricts the generative ability of video diffusion in unknown regions. Therefore, we further introduce the aforementioned consistent video as a pixel-level momentum to a directly generated video without momentum for better recovery of unseen regions. Our cascaded momentum enables video diffusion models to generate both high-fidelity and consistent novel views. We further finetune the global Gaussian representations with enhanced frames and render new frames for momentum update in the next step. In this manner, we can iteratively recover a 3D scene, avoiding the limitation of video length. Extensive experiments demonstrate the generalization capability and superior performance of our method in high-fidelity and consistent scene generation.

Autoregressive visual generation has garnered increasing attention due to its scalability and compatibility with other modalities compared with diffusion models. Most existing methods construct visual sequences as spatial patches for autoregressive generation. However, image patches are inherently parallel, contradicting the causal nature of autoregressive modeling. To address this, we propose a Spectral AutoRegressive (SpectralAR) visual generation framework, which realizes causality for visual sequences from the spectral perspective. Specifically, we first transform an image into ordered spectral tokens with Nested Spectral Tokenization, representing lower to higher frequency components. We then perform autoregressive generation in a coarse-to-fine manner with the sequences of spectral tokens. By considering different levels of detail in images, our SpectralAR achieves both sequence causality and token efficiency without bells and whistles. We conduct extensive experiments on ImageNet-1K for image reconstruction and autoregressive generation, and SpectralAR achieves 3.02 gFID with only 64 tokens and 310M parameters. Project page: https://huang-yh.github.io/spectralar/.

Modeling animatable human avatars from videos is a long-standing and challenging problem. While conventional methods require per-instance optimization, recent feed-forward methods have been proposed to generate 3D Gaussians with a learnable network. However, these methods predict Gaussians for each frame independently, without fully capturing the relations of Gaussians from different timestamps. To address this, we propose Human Gaussian Graph to model the connection between predicted Gaussians and human SMPL mesh, so that we can leverage information from all frames to recover an animatable human representation. Specifically, the Human Gaussian Graph contains dual layers where Gaussians are the first layer nodes and mesh vertices serve as the second layer nodes. Based on this structure, we further propose the intra-node operation to aggregate various Gaussians connected to one mesh vertex, and inter-node operation to support message passing among mesh node neighbors. Experimental results on novel view synthesis and novel pose animation demonstrate the efficiency and generalization of our method.

In this paper, we propose a Counterfactually Decoupled Attention Learning (CDAL) method for open-world model attribution. Existing methods rely on handcrafted design of region partitioning or feature space, which could be confounded by the spurious statistical correlations and struggle with novel attacks in open-world scenarios. To address this, CDAL explicitly models the causal relationships between the attentional visual traces and source model attribution, and counterfactually decouples the discriminative model-specific artifacts from confounding source biases for comparison. In this way, the resulting causal effect provides a quantification on the quality of learned attention maps, thus encouraging the network to capture essential generation patterns that generalize to unseen source models by maximizing the effect. Extensive experiments on existing open-world model attribution benchmarks show that with minimal computational overhead, our method consistently improves state-of-the-art models by large margins, particularly for unseen novel attacks. Source code: https://github.com/yzheng97/CDAL.

Skins wrapping around our bodies, leathers covering over the sofa, sheet metal coating the car – it suggests that objects are enclosed by a series of continuous surfaces, which provides us with informative geometry prior for objectness deduction. In this paper, we propose Gaussian-Det which leverages Gaussian Splatting as surface representation for multi-view based 3D object detection. Unlike existing monocular or NeRF-based methods which depict the objects via discrete positional data, Gaussian-Det models the objects in a continuous manner by formulating the input Gaussians as feature descriptors on a mass of partial surfaces. Furthermore, to address the numerous outliers inherently introduced by Gaussian splatting, we accordingly devise a Closure Inferring Module (CIM) for the comprehensive surface-based objectness deduction. CIM firstly estimates the probabilistic feature residuals for partial surfaces given the underdetermined nature of Gaussian Splatting, which are then coalesced into a holistic representation on the overall surface closure of the object proposal. In this way, the surface information Gaussian-Det exploits serves as the prior on the quality and reliability of objectness and the information basis of proposal refinement. Experiments on both synthetic and real-world datasets demonstrate that Gaussian-Det outperforms various existing approaches, in terms of both average precision and recall.

3D Gaussian Splatting (3DGS) has demonstrated impressive performance in 3D scene reconstruction. Beyond novel view synthesis, it shows great potential for multi-view surface reconstruction. Existing methods employ optimization-based reconstruction pipelines that achieve precise and complete surface extractions. However, these approaches typically require dense input views and high time consumption for per-scene optimization. To address these limitations, we propose SurfaceSplat, a feed-forward framework that generates efficient and generalizable pixel-aligned Gaussian surfel representations from sparse-view images. We observe that conventional feed-forward structures struggle to recover accurate geometric attributes of Gaussian surfels because the spatial frequency of pixel-aligned primitives exceeds Nyquist sampling rates. Therefore, we propose a cross-view feature aggregation module based on the Nyquist sampling theorem. Specifically, we first adapt the geometric forms of Gaussian surfels with spatial sampling rate-guided low-pass filters. We then project the filtered surfels across all input views to obtain cross-view feature correlations. By processing these correlations through a specially designed feature fusion network, we can finally regress Gaussian surfels with precise geometry. Extensive experiments on DTU reconstruction benchmarks demonstrate that our model achieves comparable results with state-of-the-art methods, and predict Gaussian surfels within 1 second, offering a 100× speedup without costly per-scene training.

Recent advancements in Multimodal Large Language Models (MLLMs) have significantly enhanced performance on 2D visual tasks. However, improving their spatial intelligence remains a challenge. Existing 3D MLLMs always rely on additional 3D or 2.5D data to incorporate spatial awareness, restricting their utility in scenarios with only 2D inputs, such as images or videos. In this paper, we present Spatial-MLLM, a novel framework for visual-based spatial reasoning from purely 2D observations. Unlike conventional video MLLMs which rely on CLIP-based visual encoders optimized for semantic understanding, our key insight is to unleash the strong structure prior from the feed-forward visual geometry foundation model. Specifically, we propose a dual-encoder architecture: a pretrained 2D visual encoder to extract semantic features, and a 3D spatial encoder—initialized from the backbone of the visual geometry model—to extract 3D structure features. A connector then integrates both features into unified visual tokens for enhanced spatial understanding. Furthermore, we propose a space-aware frame sampling strategy at inference time, which selects the spatially informative frames of a video sequence, ensuring that even under limited token length, the model focuses on frames critical for spatial reasoning. Beyond architecture improvements, we construct a training dataset from multiple sources and train the model on it using supervised fine-tuning and GRPO. Extensive experiments on various real-world datasets demonstrate that Spatial-MLLM achieves state-of-the-art performance in a wide range of visual-based spatial understanding and reasoning tasks.

Timestamp-supervised action segmentation aims to segment and classify actions in untrimmed videos with a random frame annotated per action. Precisely localizing action boundaries from timestamp annotations is crucial for this setting, as it enables generating framewise pseudo-labels and applying the well-explored fully-supervised training. However, prevailing methods struggle with intrinsic uncertainty in boundary localization due to less discriminative features in action-transiting regions. This imprecise boundary estimation significantly reduces the stability and reliability of the generated pseudo-labels in ambiguous action-transiting regions, consequently resulting in performance deterioration of the trained segmentation models. In our paper, we introduce the boundary voting network that mitigates feature ambiguity by hierarchically propagating video-level global prior knowledge into local action-transiting regions. By generating key action representations as votes throughout the video and targeting action-transiting regions, all votes collaboratively contribute to action-transiting feature enhancement and boundary localization refinement. Extensive experiments demonstrate the effectiveness of our method on GTEA, 50Salads, and Breakfast datasets.

In this paper, we propose UniqueSplat, a view-conditioned feed-forward 3D Gaussian Splatting model to reconstruct customized 3D radiance fields for each view query. Existing feed-forward methods such as pixelSplat and MVSplat aim to generate fixed Gaussians across all views of each scene by minimizing the error between rendered views and ground-truth images. However, such fixed Gaussians generally render images from all views and lack the ability to adapt to specific viewpoints, as they do not incorporate target view information when predicting Gaussians. To address this, our UniqueSplat learns the view-conditioned information as a prior and incorporates this knowledge into network parameters, so that Gaussians are dynamically adjusted in accordance with different views. Specifically, we propose a two-branch view-conditioned hyperNetwork to simultaneously learn view-agnostic embeddings and view-specific knowledge, which not only explores the shareable knowledge from various views, but also adapts the model to specific views at test time. Extensive experiments on widely-used datasets including RealEstate10K, ACID and DTU demonstrate the superiority of UniqueSplat over the state-of-the-art methods. Moreover, UniqueSplat encouragingly outperforms existing methods in cross-dataset evaluation, showing its notable generalization ability.

Occupancy prediction reconstructs 3D structures of surrounding environments. It provides detailed information for autonomous driving planning and navigation. However, most existing methods heavily rely on the LiDAR point clouds to generate occupancy ground truth, which is not available in the vision-based system. In this paper, we propose an OccNeRF method for training occupancy networks without 3D ground truth. Different from previous works which consider a bounded scene, we parameterize the reconstructed occupancy fields and reorganize the sampling strategy to align with the cameras' infinite perceptive range. The neural rendering is adopted to convert occupancy fields to multi-camera depth maps, supervised by multi-frame photometric consistency. Moreover, for semantic occupancy prediction, we design several strategies to polish the prompts and filter the outputs of a pretrained open-vocabulary 2D segmentation model. Extensive experiments for both self-supervised depth estimation and 3D occupancy prediction tasks on nuScenes and SemanticKITTI datasets demonstrate the effectiveness of our method. The code is available at https://github.com/LinShan-Bin/OccNeRF.

Shape assembly composes complex shapes geometries by arranging simple part geometries and has wide applications in autonomous robotic assembly and CAD modeling. Existing works focus on geometry reasoning and neglect the actual physical assembly process of matching and fitting joints, which are the contact surfaces connecting different parts. In this paper, we consider contacting joints for the task of multi-part assembly. A successful joint-optimized assembly needs to satisfy the bilateral objectives of shape structure and joint alignment. We propose a hierarchical graph learning approach composed of two levels of graph representation learning. The part graph takes part geometries as input to build the desired shape structure. The joint-level graph uses part joints information and focuses on matching and aligning joints. The two kinds of information are combined to achieve the bilateral objectives. Extensive experiments demonstrate that our method outperforms previous methods, achieving better shape structure and higher joint alignment accuracy.

Recently, 3D content creation from text prompts has demonstrated remarkable progress by utilizing 2D and 3D diffusion models. While 3D diffusion models ensure great multi-view consistency, their ability to generate highquality and diverse 3D assets is hindered by the limited 3D data. In contrast, 2D diffusion models find a distillation approach that achieves excellent generalization and rich details without any 3D data. However, 2D lifting methods suffer from inherent view-agnostic ambiguity thereby leading to serious multi-face Janus issues, where text prompts fail to provide sufficient guidance to learn coherent 3D results. Instead of retraining a costly viewpoint-aware model, we study how to fully exploit easily accessible coarse 3D knowledge to enhance the prompts and guide 2D lifting optimization for refinement. In this paper, we propose Sherpa3D, a new text-to-3D framework that achieves high-fidelity, generalizability, and geometric consistency simultaneously. Specifically, we design a pair of guiding strategies derived from the coarse 3D prior generated by the 3D diffusion model: a structural guidance for geometric fidelity and a semantic guidance for 3D coherence. Employing the two types of guidance, the 2D diffusion model enriches the 3D content with diversified and high - quality results. Extensive experiments show the superiority of our Sherpa3D over the state-of-the-art text-to-3D methods in terms of quality and 3D consistency. Project page: https://liuff19.github.io/Sherpa3D/.

Nowadays, leveraging 2D images and pre-trained mod- els to guide 3D point cloud feature representation has shown a remarkable potential to boost the performance of 3D fundamental models. While some works rely on additional data such as 2D real-world images and their corre- sponding camera poses, recent studies target at using point cloud exclusively by designing 3D-to-2D projection. How- ever, in the indoor scene scenario, existing 3D-to-2D pro- jection strategies suffer from severe occlusions and incoher- ence, which fail to contain sufficient information for fine- grained point cloud segmentation task. In this paper, we ar- gue that the crux of the matter resides in the basic premise of existing projection strategies that the medium is homo- geneous, thereby projection rays propagate along straight lines and behind objects are occluded by front ones. In- spired by the phenomenon of mirage where the occluded objects are exposed by distorted light rays due to heteroge- neous medium refraction rate, we propose MirageRoom by designing parametric mirage projection with heterogeneous medium to obtain series of projected images with various distorted degrees. We further develop a masked reprojection module across 2D and 3D latent space to bridge the gap between pre-trained 2D backbone and 3D point-wise features. Both quantitative and qualitative experimental re- sults on S3DIS and ScanNet V2 demonstrate the effective- ness of our method.11Code will be available here.

In this paper, we propose a new framework for online 3D scene perception. Conventional 3D scene perception methods are offline, i.e., take an already reconstructed 3D scene geometry as input, which is not applicable in robotic applications where the input data is streaming RGB-D videos rather than a complete 3D scene reconstructed from precollected RGB-D videos. To deal with online 3D scene perception tasks where data collection and perception should be performed simultaneously, the model should be able to process 3D scenes frame by frame and make use of the temporal information. To this end, we propose an adapter-based plug-and-play module for the backbone of 3D scene perception model, which constructs memory to cache and aggregate the extracted RGB-D features to empower offline models with temporal learning ability. Specifically, we propose a queued memory mechanism to cache the supporting point cloud and image features. Then we devise aggregation modules which directly perform on the memory and pass temporal information to current frame. We further propose 3D-to-2D adapter to enhance image features with strong global context. Our adapters can be easily inserted into mainstream offline architectures of different tasks and significantly boost their performance on online tasks. Extensive experiments on ScanNet and SceneNN datasets demonstrate our approach achieves leading performance on three 3D scene perception tasks compared with state-of-the-art online methods by simply finetuning existing offline models, without any model and task-specific designs. Project page.

Point clouds captured by different sensors such as RGB-D cameras and LiDAR possess non-negligible domain gaps. Most existing methods design different network architectures and train separately on point clouds from various sensors. Typically, point-based methods achieve outstanding performances on even-distributed dense point clouds from RGB-D cameras, while voxel-based methods are more efficient for large-range sparse LiDAR point clouds. In this paper, we propose geometry-to-voxel auxiliary learning to enable voxel representations to access point-level geometric information, which supports better generalisation of the voxel-based backbone with additional interpretations of multi-sensor point clouds. Specifically, we construct hierarchical geometry pools generated by a voxel-guided dynamic point network, which efficiently provide auxiliary fine-grained geometric information adapted to different stages of voxel features. We conduct experiments on joint multi-sensor datasets to demonstrate the effectiveness of GeoAuxNet. Enjoying elaborate geometric information, our method outperforms other models collectively trained on multi-sensor datasets, and achieve competitive results with the-state-of-art experts on each single dataset.

In this work, we pioneer Semantic Flow, a neural semantic representation of dynamic scenes from monocular videos. In contrast to previous NeRF methods that reconstruct dynamic scenes from the colors and volume densities of individual points, Semantic Flow learns semantics from continuous flows that contain rich 3D motion information. As there is 2D-to-3D ambiguity problem in the viewing direction when extracting 3D flow features from 2D video frames, we consider the volume densities as opacity priors that describe the contributions of flow features to the semantics on the frames. More specifically, we first learn a flow network to predict flows in the dynamic scene, and propose a flow feature aggregation module to extract flow features from video frames. Then, we propose a flow attention module to extract motion information from flow features, which is followed by a semantic network to output semantic logits of flows. We integrate the logits with volume densities in the viewing direction to supervise the flow features with semantic labels on video frames. Experimental results show that our model is able to learn from multiple dynamic scenes and supports a series of new tasks such as instance-level scene editing, semantic completions, dynamic scene tracking and semantic adaption on novel scenes.

In this paper, we propose an adaptive margin contrastive learning method for 3D point cloud semantic segmentation, namely AMContrast3D. Most existing methods use equally penalized objectives, which ignore per-point ambiguities and less discriminated features stemming from transition regions. However, as highly ambiguous points may be indistinguishable even for humans, their manually annotated labels are less reliable, and hard constraints over these points would lead to sub-optimal models. To address this, we design adaptive objectives for individual points based on their ambiguity levels, aiming to ensure the correctness of low-ambiguity points while allowing mistakes for high-ambiguity points. Specifically, we first estimate ambiguities based on position embeddings. Then, we develop a margin generator to shift decision boundaries for contrastive feature embeddings, so margins are narrowed due to increasing ambiguities with even negative margins for extremely high-ambiguity points. Experimental results on large-scale datasets, S3DIS and ScanNet, demonstrate that our method outperforms state-of-the-art methods.

In this paper, we propose ToW3D that enables precise and consistent control over 3D generative adversarial networks (GANs) with the Tug-of-War competition between shape deformation and appearance consistency. Existing point-based GAN editing methods such as DragGAN and GANWarping have yielded impressive performance for 2D image manipulation. However, as 3D generators present weaker generalization ability compared with 2D due to limited training data, they would suffer from drastic changes in global appearance when editing local areas of meshes. To address this, we design a pipeline of “drag locally, shove globally”, which iteratively performs two optimization steps: 1) pull the point towards the target, and 2) push the structure and semantics back to the source. Specifically, we design a structure adaption module based on structure which guarantees the preservation of basic geometric properties, and a semantic preservation module that maintains semantic similarity across different views. Extensive qualitative and quantitative experiments demonstrate superiority of our ToW3D approach over prior methods in terms of appearance consistency and fidelity especially under large deformations.

In this paper, we propose a characteristic-aware adaptive network named AdaForensics for deepfake detection. Most existing methods learn a fixed network to detect deepfakes based on carefully-designed network architectures. However, these methods employ the same deepfake detector for all the images despite of various facial characteristic, which fail to provide customized forgery detection for different individuals. To address this, our AdaForensics simultaneously learns characteristic-agnostic and characteristic-specific embeddings, where the detector dynamically adapts to varying faces with our designed hypernetwork on the fly. More specifically, our AdaForensics not only explores the shareable abstractions from various deepfake images, but also adapts the detector to the given characteristic at test time. To achieve this, we propose a two-branch HyperNetwork to learn an adaptive deepfake detector, which automatically adjusts the parameters based on characteristic of the input. Extensive experiments on widely-used datasets including FaceForensics, Celeb-DF and DFDC demonstrate our AdaForensics outperforms the state-of-the-art works.

In this work, we introduce Unique3D, a novel image-to-3D framework for efficiently generating high-quality 3D meshes from single-view images, featuring state-of-the-art generation fidelity and strong generalizability. Previous methods based on Score Distillation Sampling (SDS) can produce diversified 3D results by distilling 3D knowledge from large 2D diffusion models, but they usually suffer from long per-case optimization time with inconsistent issues. Recent works address the problem and generate better 3D results either by finetuning a multi-view diffusion model or training a fast feed-forward model. However, they still lack intricate textures and complex geometries due to inconsistency and limited generated resolution. To simultaneously achieve high fidelity, consistency, and efficiency in single image-to-3D, we propose a novel framework Unique3D that includes a multi-view diffusion model with a corresponding normal diffusion model to generate multi-view images with their normal maps, a multi-level upscale process to progressively improve the resolution of generated orthographic multi-views, as well as an instant and consistent mesh reconstruction algorithm called ISOMER, which fully integrates the color and geometric priors into mesh results. Extensive experiments demonstrate that our Unique3D significantly outperforms other image-to-3D baselines in terms of geometric and textural details.

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.

In this paper, we propose a dynamic 3D object detector named HyperDet3D, which is adaptively adjusted based on the hyper scene-level knowledge on the fly. Existing methods strive for object-level representations of local elements and their relations without scene-level priors, which suffer from ambiguity between similarly-structured objects only based on the understanding of individual points and object candidates. Instead, we design scene-conditioned hypernetworks to simultaneously learn scene-agnostic embeddings to exploit sharable abstracts from various 3D scenes, and scene-specific knowledge which adapts the 3D detector to the given scene at test time. As a result, the lower-level ambiguity in object representations can be addressed by hierarchical context in scene priors. However, since the upstream hypernetwork in HyperDet3D takes raw scenes as input which contain noises and redundancy, it leads to sub-optimal parameters produced for the 3D detector simply under the constraint of downstream detection losses. Based on the fact that the downstream 3D detection task can be factorized into object-level semantic classification and bounding box regression, we furtherly propose HyperFormer3D by correspondingly designing their scene-level prior tasks in upstream hypernetworks, namely Semantic Occurrence and Objectness Localization. To this end, we design a transformer-based hypernetwork that translates the task-oriented scene priors into parameters of the downstream detector, which refrains from noises and redundancy of the scenes. Extensive experimental results on the ScanNet, SUN RGB-D and MatterPort3D datasets demonstrate the effectiveness of the proposed methods.

In this paper, we investigate the dynamics-aware adversarial attack problem of adaptive neural networks. Most existing adversarial attack algorithms are designed under a basic assumption – the network architecture is fixed throughout the attack process. However, this assumption does not hold for many recently proposed adaptive neural networks, which adaptively deactivate unnecessary execution units based on inputs to improve computational efficiency. It results in a serious issue of lagged gradient, making the learned attack at the current step ineffective due to the architecture change afterward. To address this issue, we propose a Leaded Gradient Method (LGM) and show the significant effects of the lagged gradient. More specifically, we reformulate the gradients to be aware of the potential dynamic changes of network architectures, so that the learned attack better “leads” the next step than the dynamics-unaware methods when network architecture changes dynamically. Extensive experiments on representative types of adaptive neural networks for both 2D images and 3D point clouds show that our LGM achieves impressive adversarial attack performance compared with the dynamic-unaware attack methods. Code is available at https://github.com/antao97/LGM.

Recovering a dense depth map from a pair of indoor RGB and sparse depth images in an unsupervised manner is paramount in applications such as autonomous driving and 3D reconstruction. Most existing methods leverage sparse depth maps to directly estimate the dense depth map with the pixel-wise regression constraints over the input known depth. However, such regression constraints independently compare per-pixel depth values, which ignore the important 3D structures hidden behind depth maps and result in severe structural distortion and poor robustness. In this paper, we propose a Structure-Preserving Encoding (SPE) module by reformulating depth completion as the process of 3D structure generation. The generated structure should recover the complete scene and also consist with the known partial structure, so that the learned depth features from this task are able to encode rich structural information. In addition, SPE hierarchically interpolates and propagates the 3D structures into dense structure-aware positional encodings, which further boosts the information interactions between RGB and depth features via our transformer. Extensive experiments on VOID and NYUv2 demonstrate that SPTR outperforms the state-of-the-art methods by a large margin across various densities of input depths and a strong generalization ability to other datasets.

In this paper, we propose a point-based cross-attention transformer named CrossPoints with parametric Global Porous Sampling (GPS) strategy. The attention module is crucial to capture the correlations between different tokens for transformers. Most existing point-based transformers design multi-scale self-attention operations with down-sampled point clouds by the widely-used Farthest Point Sampling (FPS) strategy. However, FPS only generates sub-clouds with holistic structures, which fails to fully exploit the flexibility of points to generate diversified tokens for the attention module. To address this, we design a cross-attention module with parametric GPS and Complementary GPS (C-GPS) strategies to generate series of diversified tokens through controllable parameters. We show that FPS is a degenerated case of GPS, and the network learns more abundant relational information of the structure and geometry when we perform consecutive cross-attention over the tokens generated by GPS as well as C-GPS sampled points. More specifically, we set evenly-sampled points as queries and design our cross-attention layers with GPS and C-GPS sampled points as keys and values. In order to further improve the diversity of tokens, we design a deformable operation over points to adaptively adjust the points according to the input. Extensive experimental results on both shape classification and indoor scene segmentation tasks indicate promising boosts over the recent point cloud transformers. We also conduct ablation studies to show the effectiveness of our proposed cross-attention module with GPS strategy.

In this paper, we propose an effective plug-and-play module called structural relation network (SRN) to model structural dependencies in 3D point clouds for feature representation. Existing network architectures such as PointNet++ and RS-CNN capture local structures individually and ignore the inner interactions between different sub-clouds. Motivated by the fact that structural relation modeling plays critical roles for humans to understand 3D objects, our SRN exploits local information by modeling structural relations in 3D spaces. For a given sub-cloud of point sets, SRN firstly extracts its geometrical and locational relations with the other sub-clouds and maps them into the embedding space, then aggregates both relational features with the other sub-clouds. As the variation of semantics embedded in different sub-clouds is ignored by SRN, we further extend SRN to enable dynamic message passing between different sub-clouds. We propose a graph-based structural relation network (GSRN) where sub-clouds and their pairwise relations are modeled as nodes and edges respectively, so that the node features are updated by the messages along the edges. Since the node features might not be well preserved when acquiring the global representation, we propose a Combined Entropy Readout (CER) function to adaptively aggregate them into the holistic representation, so that GSRN simultaneously models the local-local and local-global region-wise interaction. The proposed SRN and GSRN modules are simple, interpretable, and do not require any additional supervision signals, which can be easily equipped with the existing networks. Experimental results on the benchmark datasets (ScanObjectNN, ModelNet40, ShapeNet Part, S3DIS, ScanNet and SUN-RGBD) indicate promising boosts on the tasks of 3D point cloud classification, segmentation and object detection.

Effectively preserving and encoding structure features from objects in irregular and sparse LiDAR points is a crucial challenge to 3D object detection on the point cloud. Recently, Transformer has demonstrated promising performance on many 2D and even 3D vision tasks. Compared with the fixed and rigid convolution kernels, the self-attention mechanism in Transformer can adaptively exclude the unrelated or noisy points and is thus suitable for preserving the local spatial structure in the irregular LiDAR point cloud. However, Transformer only performs a simple sum on the point features, based on the self-attention mechanism, and all the points share the same transformation for value. A such isotropic operation cannot capture the direction-distance-oriented local structure, which is essential for 3D object detection. In this work, we propose a Structure-Embedding transFormer (SEFormer), which can not only preserve the local structure as a traditional Transformer but also have the ability to encode the local structure. Compared to the self-attention mechanism in traditional Transformer, SEFormer learns different feature transformations for value points based on the relative directions and distances to the query point. Then we propose a SEFormer-based network for high-performance 3D object detection. Extensive experiments show that the proposed architecture can achieve SOTA results on the Waymo Open Dataset, one of the most significant 3D detection benchmarks for autonomous driving. Specifically, SEFormer achieves 79.02% mAP, which is 1.2% higher than existing works. https://github.com/tdzdog/SEFormer.

With the rising industrial attention to 3D virtual mod-eling technology, generating novel 3D content based on specified conditions (e.g. text) has become a hot issue. In this paper, we propose a new generative 3D modeling framework called Diffusion-SDF for the challenging task of text-to-shape synthesis. Previous approaches lack flexibility in both 3D data representation and shape generation, thereby failing to generate highly diversified 3D shapes conforming to the given text descriptions. To address this, we propose a SDF autoencoder together with the voxelized Diffusion model to learn and generate representations for voxelized signed distance fields (SDFs) of 3D shapes. Specifically, we design a novel Uinll-Net architecture that implants a local-focused inner network inside the standard U-Net architecture, which enables better reconstruction of patch-independent SDF representations. We extend our approach to further text-to-shape tasks including text-conditioned shape completion and manipulation. Experimental results show that Diffusion-SDF generates both higher quality and more diversified 3D shapes that conform well to given text descriptions when compared to previous approaches. Code is available at: https://github.com/ttlmh/Diffusion-SDF.

In this paper, we aim to learn a semantic radiance field from multiple scenes that is accurate, efficient and generalizable. While most existing NeRFs target at the tasks of neural scene rendering, image synthesis and multi-view reconstruction, there are a few attempts such as Semantic-NeRF that explore to learn high-level semantic understanding with the NeRF structure. However, Semantic-NeRF simultaneously learns color and semantic label from a single ray with multiple heads, where the single ray fails to provide rich semantic information. As a result, Semantic NeRF relies on positional encoding and needs to train one specific model for each scene. To address this, we propose Semantic Ray (S-Ray) to fully exploit semantic information along the ray direction from its multi-view reprojections. As directly performing dense attention over multi-view reprojected rays would suffer from heavy computational cost, we design a Cross-Reprojection Attention module with consecutive intra-view radial and cross-view sparse attentions, which decomposes contextual information along reprojected rays and cross multiple views and then collects dense connections by stacking the modules. Experiments show that our S-Ray is able to learn from multiple scenes, and it presents strong generalization ability to adapt to unseen scenes. Project page: https://liuff19.github.io/S-Ray/.

In this paper, we target at the problem of learning a generalizable dynamic radiance field from monocular videos. Different from most existing NeRF methods that are based on multiple views, monocular videos only contain one view at each timestamp, thereby suffering from ambiguity along the view direction in estimating point features and scene flows. Previous studies such as DynNeRF disambiguate point features by positional encoding, which is not transferable and severely limits the generalization ability. As a result, these methods have to train one independent model for each scene and suffer from heavy computational costs when applying to increasing monocular videos in real-world applications. To address this, We propose MonoNeRF to simultaneously learn point features and scene flows with point trajectory and feature correspondence constraints across frames. More specifically, we learn an implicit velocity field to estimate point trajectory from temporal features with Neural ODE, which is followed by a flow-based feature aggregation module to obtain spatial features along the point trajectory. We jointly optimize temporal and spatial features in an end-to-end manner. Experiments show that our MonoNeRF is able to learn from multiple scenes and support new applications such as scene editing, unseen frame synthesis, and fast novel scene adaptation. Codes are available at https://github.com/tianfr/MonoNeRF.

In this paper, we propose an HOI-aware adaptive network named AdaAct for weakly-supervised action segmentation. Most existing methods learn a fixed network to predict the action of each frame with the neighboring frames. However, this would result in ambiguity when estimating similar actions, such as pouring juice and pouring coffee. To address this, we aim to exploit temporally global but spatially local human-object interactions (HOI) as video-level prior knowledge for action segmentation. The long-term HOI sequence provides crucial contextual information to distinguish ambiguous actions, where our network dynamically adapts to the given HOI sequence at test time. More specifically, we first design a video HOI encoder that extracts, selects, and integrates the most representative HOI throughout the video. Then, we propose a two-branch HyperNetwork to learn an adaptive temporal encoder, which automatically adjusts the parameters based on the HOI information of various videos on the fly. Extensive experiments on two widely-used datasets including Breakfast and 50Salads demonstrate the effectiveness of our method under different evaluation metrics.

Discovering causal structure from purely observational data (i.e., causal discovery), aiming to identify causal relationships among variables, is a fundamental task in machine learning.The recent invention of differentiable score-based DAG learners is a crucial enabler, which reframes the combinatorial optimization problem into a differentiable optimization with a DAG constraint over directed graph space. Despite their great success, these cutting-edge DAG learners incorporate DAG-ness independent score functions to evaluate the directed graph candidates, lacking in considering graph structure. As a result, measuring the data fitness alone regardless of DAG-ness inevitably leads to discovering suboptimal DAGs and model vulnerabilities.

In this paper, we propose a GraphBit method to learn unsupervised deep binary descriptors for efficient image representation. Conventional binary representation learning methods directly quantize each element according to the threshold without considering the quantization ambiguousness. The elements near the boundary dubbed as "ambiguous bits" fail to collect effective information for reliable binarization and are sensitive to noise that causes reversed bits. We argue that there are implicit inner relationships among bits in binary descriptors called bitwise interaction, where the related bits can provide extra instruction as prior knowledge for ambiguousness reduction. Specifically, we design a deep reinforcement learning model to learn the structure of the graph for bitwise interaction mining, and the uncertainty of binary codes is reduced by maximizing the mutual information with input and related bits. Consequently, the ambiguous bits receive additional instruction from the graph for reliable binarization. Moreover, we further present a differentiable search method (GraphBit+) that mines the bitwise interaction in continuous space, so that the heavy search cost caused by the training difficulties in reinforcement learning is significantly reduced. Since the GraphBit and GraphBit+ methods learn fixed bitwise interaction which is suboptimal for various input, the inaccurate instruction from the fixed bitwise interaction cannot effectively decrease the ambiguousness of binary descriptors. To address this, we further propose the unsupervised binary descriptor learning method via dynamic bitwise interaction mining (D-GraphBit), where a graph convolutional network called GraphMiner reasons the optimal bitwise interaction for each input sample. Extensive experimental results on the CIFAR-10, NUS-WIDE, ImageNet-100, Brown and HPatches datasets demonstrate the efficiency and effectiveness of the proposed GraphBit, GraphBit+ and D-GraphBit.

In this paper, we propose a Mask-Robust Inpainting Network (MRIN) approach to recover the masked areas of an image. Most existing methods learn a single model for image inpainting, under a basic assumption that all masks are from the same type. However, we discover that the masks are usually complex and exhibit various shapes and sizes at different locations of an image, where a single model cannot fully capture the large domain gap across different masks. To address this, we learn to decompose a complex mask area into several basic types and recover the damaged image in a patch-wise manner with a type-specific generator. More specifically, our MRIN consists of a mask-robust agent and an adaptive patch generative network. The mask-robust agent contains a mask selector and a patch locator, which generates mask attention maps to select a patch at each step. Based on the predicted mask attention maps, the adaptive patch generative network inpaints the selected patch with the generators bank, so that it sequentially inpaints each patch with different patch generators according to its mask type. Extensive experiments demonstrate that our approach outperforms most state-of-the-art approaches on the Place2, CelebA, and Paris Street View datasets.

Action recognition models have shown a promising capability to classify human actions in short video clips. In a real scenario, multiple correlated human actions commonly occur in particular orders, forming semantically meaningful human activities. Conventional action recognition approaches focus on analyzing single actions. However, they fail to fully reason about the contextual relations between adjacent actions, which provide potential temporal logic for understanding long videos. In this paper, we propose a prompt-based framework, Bridge-Prompt (Br-Prompt), to model the semantics across adjacent actions, so that it simultaneously exploits both out-of-context and contextual information from a series of ordinal actions in instructional videos. More specifically, we reformulate the individual action labels as integrated text prompts for super-vision, which bridge the gap between individual action semantics. The generated text prompts are paired with corresponding video clips, and together co-train the text encoder and the video encoder via a contrastive approach. The learned vision encoder has a stronger capability for ordinal-action-related downstream tasks, e.g. action segmentation and human activity recognition. We evaluate the performances of our approach on several video datasets: Georgia Tech Egocentric Activities (GTEA), 50Salads, and the Breakfast dataset. Br-Prompt achieves state-of-the-art on multiple benchmarks. Code is available at: https://github.com/ttlmh/Bridge-Prompt.

It is difficult to construct a data collection including all possible combinations of human actions and interacting objects due to the combinatorial nature of human-object interactions (HOI). In this work, we aim to develop a transferable HOI detector for unseen interactions. Existing HOI detectors often treat interactions as discrete labels and learn a classifier according to a predetermined category space. This is inherently inapt for detecting unseen interactions which are out of the predefined categories. Conversely, we treat independent HOI labels as the natural language supervision of interactions and embed them into a joint visual-and-text space to capture their correlations. More specifically, we propose a new HOI visual encoder to detect the interacting humans and objects, and map them to a joint feature space to perform interaction recognition. Our visual encoder is instantiated as a Vision Transformer with new learnable HOI tokens and a sequence parser to generate unique HOI predictions. It distills and leverages the transferable knowledge from the pretrained CLIP model to perform the zero-shot interaction detection. Experiments on two datasets, SWIG-HOI and HICO-DET, validate that our proposed method can achieve a notable mAP improvement on detecting both seen and unseen HOIs. Our code is available at https://github.com/scwangdyd/promting_hoi.

A bathtub in a library, a sink in an office, a bed in a laundry room - the counter-intuition suggests that scene provides important prior knowledge for 3D object detection, which instructs to eliminate the ambiguous detection of similar objects. In this paper, we propose HyperDet3D to explore scene-conditioned prior knowledge for 3D object detection. Existing methods strive for better representation of local elements and their relations without scene-conditioned knowledge, which may cause ambiguity merely based on the understanding of individual points and object candidates. Instead, HyperDet3D simultaneously learns scene-agnostic embeddings and scene-specific knowledge through scene-conditioned hypernetworks. More specifically, our HyperDet3D not only explores the sharable abstracts from various 3D scenes, but also adapts the detector to the given scene at test time. We propose a discriminative Multi-head Scene-specific Attention (MSA) module to dynamically control the layer parameters of the detector conditioned on the fusion of scene-conditioned knowledge. Our HyperDet3D achieves state-of-the-art results on the 3D object detection benchmark of the ScanNet and SUN RGB-D datasets. Moreover, through cross-dataset evaluation, we show the acquired scene-conditioned prior knowledge still takes effect when facing 3D scenes with domain gap.

In this paper, we propose an uncertainty-aware representation Learning (UARL) method for action segmentation. Most existing action segmentation methods exploit continuity information of the action period to predict frame-level labels, which ignores the temporal ambiguity of the transition region between two actions. Moreover, similar periods of different actions, e.g., the beginning of some actions, will confuse the network if they are annotated with different labels, which causes spatial ambiguity. To address this, we design the UARL to exploit the transitional expression between two action periods by uncertainty learning. Specially, we model every frame of actions with an active distribution that represents the probabilities of different actions, which captures the uncertainty of the action and exploits the tendency during the action. We evaluate our method on three popular action prediction datasets: Breakfast, Georgia Tech Egocentric Activities (GTEA), and 50Salads. The experimental results demonstrate that our method achieves the performance with state-of-the-art.

Most existing point cloud instance and semantic segmentation methods rely heavily on strong supervision signals, which require point-level labels for every point in the scene. However, such strong supervision suffers from large annotation costs, arousing the need to study efficient annotating. In this paper, we discover that the locations of instances matter for both instance and semantic 3D scene segmentation. By fully taking advantage of locations, we design a weakly-supervised point cloud segmentation method that only requires clicking on one point per instance to indicate its location for annotation. With over-segmentation for pre-processing, we extend these location annotations into segments as seg-level labels. We further design a segment grouping network (SegGroup) to generate point-level pseudo labels under seg-level labels by hierarchically grouping the unlabeled segments into the relevant nearby labeled segments, so that existing point-level supervised segmentation models can directly consume these pseudo labels for training. Experimental results show that our seg-level supervised method (SegGroup) achieves comparable results with the fully annotated point-level supervised methods. Moreover, it outperforms the recent weakly-supervised methods given a fixed annotation budget. Code is available at https://github.com/antao97/SegGroup.

Invariance and equivariance to the rotation group have been widely discussed in the 3D deep learning community for pointclouds. Yet most proposed methods either use complex mathematical tools that may limit their accessibility, or are tied to specific input data types and network architectures. In this paper, we introduce a general framework built on top of what we call Vector Neuron representations for creating SO (3) -equivariant neural networks for pointcloud processing. Extending neurons from 1D scalars to 3D vectors, our vector neurons enable a simple mapping of SO (3) actions to latent spaces thereby providing a framework for building equivariance in common neural operations – including linear layers, non-linearities, pooling, and normalizations. Due to their simplicity, vector neurons are versatile and, as we demonstrate, can be incorporated into diverse network architecture backbones, allowing them to process geometry inputs in arbitrary poses. Despite its simplicity, our method performs comparably well in accuracy and generalization with other more complex and specialized state-of-the-art methods on classification and segmentation tasks. We also show for the first time a rotation equivariant reconstruction network. Source code is available at https://github.com/FlyingGiraffe/vnn.

In this work, we tackle the problem of category-level online pose tracking of objects from point cloud sequences. For the first time, we propose a unified framework that can handle 9DoF pose tracking for novel rigid object instances as well as per-part pose tracking for articulated objects from known categories. Here the 9DoF pose, comprising 6D pose and 3D size, is equivalent to a 3D amodal bounding box representation with free 6D pose. Given the depth point cloud at the current frame and the estimated pose from the last frame, our novel end-to-end pipeline learns to accurately update the pose. Our pipeline is composed of three modules: 1) a pose canonicalization module that normalizes the pose of the input depth point cloud; 2) RotationNet, a module that directly regresses small interframe delta rotations; and 3) CoordinateNet, a module that predicts the normalized coordinates and segmentation, enabling analytical computation of the 3D size and translation. Leveraging the small pose regime in the pose-canonicalized point clouds, our method integrates the best of both worlds by combining dense coordinate prediction and direct rotation regression, thus yielding an end-to-end differentiable pipeline optimized for 9DoF pose accuracy (without using non-differentiable RANSAC). Our extensive experiments demonstrate that our method achieves new state-of-the-art performance on category-level rigid object pose (NOCSREAL275 [29]) and articulated object pose benchmarks (SAPIEN [34], BMVC [18]) at the fastest FPS ∼ 12.

Learning an effective distance measurement between sample pairs plays an important role in visual analysis, where the training procedure largely relies on hard negative samples. However, hard negative samples usually account for the tiny minority in the training set, which may fail to fully describe the data distribution close to the decision boundary. In this paper, we present a deep adversarial metric learning (DAML) framework to generate synthetic hard negatives from the original negative samples, which is widely applicable to existing supervised deep metric learning algorithms. Different from existing sampling strategies which simply ignore numerous easy negatives, our DAML aim to exploit them by generating synthetic hard negatives adversarial to the learned metric as complements. We simultaneously train the feature embedding and hard negative generator in an adversarial manner, so that adequate and targeted synthetic hard negatives are created to learn more precise distance metrics. As a single transformation may not be powerful enough to describe the global input space under the attack of the hard negative generator, we further propose a deep adversarial multi-metric learning (DAMML) method by learning multiple local transformations for more complete description. We simultaneously exploit the collaborative and competitive relationships among multiple metrics, where the metrics display unity against the generator for effective distance measurement as well as compete for more training data through a metric discriminator to avoid overlapping. Extensive experimental results on five benchmark datasets show that our DAML and DAMML effectively boost the performance of existing deep metric learning approaches through adversarial learning.

Deep embedding learning aims to learn a distance metric for effective similarity measurement, which has achieved promising performance in various tasks. As the vast majority of training samples produce gradients with magnitudes close to zero, hard example mining is usually employed to improve the effectiveness and efficiency of the training procedure. However, most existing sampling methods are designed by hand, which ignores the dependence between examples and suffer from exhaustive searching. In this paper, we propose a deep embedding with discriminative sampling policy (DE-DSP) learning framework by simultaneously training two models: a deep sampler network that learns effective sampling strategies, and a feature embedding that maps samples to the feature space. Rather than exhaustively calculating the hardness of all the examples for mining through forward-propagation, the deep sampler network exploits the strong prior of relations among samples to learn discriminative sampling policy in an more efficient manner. Experimental results demonstrate faster convergence and stronger discriminative power of our DE-DSP framework under different embedding objectives.

In this paper, we propose a new supervision objective named uniform loss to learn deep equidistributed representations for face recognition. Most existing methods aim to learn discriminative face features, encouraging large inter-class distances and small intra-class variations. However, they ignore the distribution of faces in the holistic feature space, which may lead to severe locality and unbalance. With the prior that faces lie on a hypersphere manifold, we impose an equidistributed constraint by uniformly spreading the class centers on the manifold, so that the minimum distance between class centers can be maximized through complete exploitation of the feature space. To this end, we consider the class centers as like charges on the surface of hypersphere with inter-class repulsion, and minimize the total electric potential energy as the uniform loss. Extensive experimental results on the MegaFace Challenge I, IARPA Janus Benchmark A (IJB-A), Youtube Faces (YTF) and Labeled Faces in the Wild (LFW) datasets show the effectiveness of the proposed uniform loss.

The symmetry for the corners of a box, the continuity for the surfaces of a monitor, the linkage between the torso and other body parts --- it suggests that 3D objects may have common and underlying inner relations between local structures, and it is a fundamental ability for intelligent species to reason for them. In this paper, we propose an effective plug-and-play module called the structural relation network (SRN) to reason about the structural dependencies of local regions in 3D point clouds. Existing network architectures on point sets such as PointNet++ capture local structures individually, without considering their inner interactions. Instead, our SRN simultaneously exploits local information by modeling their geometrical and locational relations, which play critical roles for our humans to understand 3D objects. The proposed SRN module is simple, interpretable, and does not require any additional supervision signals, which can be easily equipped with the existing networks. Experimental results on benchmark datasets indicate promising boosts on the tasks of 3D point cloud classification and segmentation by capturing structural relations with the SRN module.

In this paper, we propose an unsupervised feature learning method called deep binary descriptor with multi-quantization (DBD-MQ) for visual analysis. Existing learning-based binary descriptors such as compact binary face descriptor (CBFD) and DeepBit utilize the rigid sign function for binarization despite of data distributions, which usually suffer from severe quantization loss. In order to address the limitation, we propose a deep multi-quantization network to learn a data-dependent binarization in an unsupervised manner. More specifically, we design a K-Autoencoders (KAEs) network to jointly learn the parameters of feature extractor and the binarization functions under a deep learning framework, so that discriminative binary descriptors can be obtained with a fine-grained multi-quantization. As DBD-MQ simply allocates the same number of quantizers to each real-valued feature dimension ignoring the elementwise diversity of informativeness, we further propose a deep competitive binary descriptor with multi-quantization (DCBD-MQ) method to learn optimal allocation of bits with the fixed binary length in a competitive manner, where informative dimensions gain more bits for complete representation. Moreover, we present a similarity-aware binary encoding strategy based on the earth mover's distance of Autoencoders, so that elements that are quantized into similar Autoencoders will have smaller Hamming distances. Extensive experimental results on six widely-used datasets show that our DBD-MQ and DCBD-MQ outperform most state-of-the-art unsupervised binary descriptors.

In this paper, we propose a GraphBit method to learn deep binary descriptors in a directed acyclic graph unsupervisedly, representing bitwise interactions as edges between the nodes of bits. Conventional binary representation learning methods enforce each element to be binarized into zero or one. However, there are elements lying in the boundary which suffer from doubtful binarization as "ambiguous bits". Ambiguous bits fail to collect effective information for confident binarization, which are unreliable and sensitive to noise. We argue that there are implicit inner relationships between bits in binary descriptors, where the related bits can provide extra instruction as prior knowledge for ambiguity elimination. Specifically, we design a deep reinforcement learning model to learn the structure of the graph for bitwise interaction mining, reducing the uncertainty of binary codes by maximizing the mutual information with inputs and related bits, so that the ambiguous bits receive additional instruction from the graph for confident binarization. Due to the reliability of the proposed binary codes with bitwise interaction, we obtain an average improvement of 9.64%, 8.84% and 3.22% on the CIFAR-10, Brown and HPatches datasets respectively compared with the state-of-the-art unsupervised binary descriptors.

Learning an effective distance metric between image pairs plays an important role in visual analysis, where the training procedure largely relies on hard negative samples. However, hard negatives in the training set usually account for the tiny minority, which may fail to fully describe the distribution of negative samples close to the margin. In this paper, we propose a deep adversarial metric learning (DAML) framework to generate synthetic hard negatives from the observed negative samples, which is widely applicable to supervised deep metric learning methods. Different from existing metric learning approaches which simply ignore numerous easy negatives, the proposed DAML exploits them to generate potential hard negatives adversarial to the learned metric as complements. We simultaneously train the hard negative generator and feature embedding in an adversarial manner, so that more precise distance metrics can be learned with adequate and targeted synthetic hard negatives. Extensive experimental results on three benchmark datasets including CUB-200-2011, Cars196 and Stanford Online Products show that DAML effectively boosts the performance of existing deep metric learning approaches through adversarial learning.

In this paper, we propose a context-aware local binary feature learning (CA-LBFL) method for face recognition. Unlike existing learning-based local face descriptors such as discriminant face descriptor (DFD) and compact binary face descriptor (CBFD) which learn each feature code individually, our CA-LBFL exploits the contextual information of adjacent bits by constraining the number of shifts from different binary bits, so that more robust information can be exploited for face representation. Given a face image, we first extract pixel difference vectors (PDV) in local patches, and learn a discriminative mapping in an unsupervised manner to project each pixel difference vector into a context-aware binary vector. Then, we perform clustering on the learned binary codes to construct a codebook, and extract a histogram feature for each face image with the learned codebook as the final representation. In order to exploit local information from different scales, we propose a context-aware local binary multi-scale feature learning (CA-LBMFL) method to jointly learn multiple projection matrices for face representation. To make the proposed methods applicable for heterogeneous face recognition, we present a coupled CA-LBFL (C-CA-LBFL) method and a coupled CA-LBMFL (C-CA-LBMFL) method to reduce the modality gap of corresponding heterogeneous faces in the feature level, respectively. Extensive experimental results on four widely used face datasets clearly show that our methods outperform most state-of-the-art face descriptors.

Metric learning has been widely used in many visual analysis applications, which learns new distance metrics to measure the similarities of samples effectively. Conventional metric learning methods learn a single linear Mahalanobis metric, yet such linear projections are not powerful enough to capture the nonlinear relationships. Recently, deep metric learning approaches, such as discriminative deep metric learning and deep transfer metric learning, have been introduced to fully exploit the nonlinearity of samples by learning hierarchical nonlinear transformations. However, these methods only learn holistic metrics over the input space and are limited for the heterogeneous data sets, where data varies locally. In this paper, we propose a deep localized metric learning approach for visual recognition by learning multiple fine-grained deep localized metrics. We first learn K local subspaces and one holistic subspace with the K-auto-encoders-based clustering. Then, given an input pair, we compute its localized distance on each learned subspace and obtain the final distance representation. Finally, we train the entire neural networks to ensure the distances of positive pairs smaller than negative pairs by a large margin. Experimental results on three visual recognition applications, including face recognition, person re-identification, and scene recognition, show that our DLML outperforms most existing metric learning approaches.

In this paper, we propose a topology preserving graph matching (TPGM) method for partial face recognition. Most existing face recognition methods extract features from holistic facial images. However, faces in real-world unconstrained environments may be occluded by objects or other faces, which cannot provide the whole face images for description. Keypoint-based partial face recognition methods such as multi-keypoint descriptor with Gabor ternary pattern and robust point set matching match the local keypoints for partial face recognition. However, they simply measure the nodewise similarity without higher order geometric graph information, which are susceptible to noises. To address this, our TPGM method estimates a non-rigid transformation encoding the second-order geometric structure of the graph, so that more accurate and robust correspondence can be computed with the topological information. In order to exploit higher order topological information, we propose a topology preserving structural matching method to construct a higher order structure for each face and estimate the transformation. Experimental results on four widely used face data sets demonstrate that our method outperforms most existing state-of-the-art face recognition methods.

In this paper, we propose an unsupervised feature learning method called deep binary descriptor with multi-quantization (DBD-MQ) for visual matching. Existing learning-based binary descriptors such as compact binary face descriptor (CBFD) and DeepBit utilize the rigid sign function for binarization despite of data distributions, thereby suffering from severe quantization loss. In order to address the limitation, our DBD-MQ considers the binarization as a multi-quantization task. Specifically, we apply a K-AutoEncoders (KAEs) network to jointly learn the parameters and the binarization functions under a deep learning framework, so that discriminative binary descriptors can be obtained with a fine-grained multi-quantization. Extensive experimental results on different visual analysis including patch retrieval, image matching and image retrieval show that our DBD-MQ outperforms most existing binary feature descriptors.

In this paper, we propose a topology preserving graph matching (TPGM) method for partial face recognition. Most existing face recognition methods extract features from holistic face images, yet faces in real-world unconstrained environments are usually occluded by objects or other faces, which cannot provide the whole face images for recognition. Latest keypoint-based partial face recognition methods only match on the detected keypoints to remove the occluded regions. However, they simply measure the node-wise similarity without higher order geometrical graph information, thereby depending heavily on descriptors which are susceptible to noises. To address this, our TPGM method estimates a non-rigid transformation encoding the second order geometric structure of the graph, so that more accurate and robust correspondence can be computed with the topological information. Experimental results on three widely used face datasets show that the proposed TPGM outperforms most existing state-of-the-art partial face recognition methods.

In this paper, we propose a rotation-invariant local binary descriptor (RI-LBD) learning method for visual recognition. Compared with hand-crafted local binary descriptors, such as local binary pattern and its variants, which require strong prior knowledge, local binary feature learning methods are more efficient and data-adaptive. Unlike existing learning-based local binary descriptors, such as compact binary face descriptor and simultaneous local binary feature learning and encoding, which are susceptible to rotations, our RI-LBD first categorizes each local patch into a rotational binary pattern (RBP), and then jointly learns the orientation for each pattern and the projection matrix to obtain RI-LBDs. As all the rotation variants of a patch belong to the same RBP, they are rotated into the same orientation and projected into the same binary descriptor. Then, we construct a codebook by a clustering method on the learned binary codes, and obtain a histogram feature for each image as the final representation. In order to exploit higher order statistical information, we extend our RI-LBD to the triple rotation-invariant co-occurrence local binary descriptor (TRICo-LBD) learning method, which learns a triple co-occurrence binary code for each local patch. Extensive experimental results on four different visual recognition tasks, including image patch matching, texture classification, face recognition, and scene classification, show that our RI-LBD and TRICo-LBD outperform most existing local descriptors.

Automatic building change detection at different periods is very important for city monitoring, disaster assessment, map updating, etc. Some existing data sources could be used in this task such as 3D geometry model (e.g. Digital Surface Model, Geographic Information System) and radiometric images from satellites or special aircrafts. However, it is too expensive for timely change detection by using these above methods. With the rapid development of UAV technique, capturing the city building images with high resolution camera at a low altitude becomes cheaper and cheaper. Using these easily acquired aerial images, we proposed a novel change detection framework based on RGB-D map generated by 3D reconstruction, which can overcome the large illumination changes. Firstly, an image-based 3D reconstruction is applied to retrieve two point clouds and related camera poses from two aerial image sets captured at different periods. Then an RGB-D map could be generated from each 3D model, followed by a 2D-3D registration procedure to align the two reconstructed 3D point clouds together. At last, a difference depth map could be generated and from which we can use random forest classification and component connectivity analysis techniques to segment the changed building areas out. Experimental results have illustrated the effectiveness and applicability of the proposed framework.

Feature based image matching is essential for many computer vision applications. Recently, progressive methods which iteratively enrich the candidate matches and reject the wrong ones have attracted a lot of attentions due to its high precision/recall and efficiency. Its quality of enrichment and rejection relies heavily on the accuracy of the estimated local affine transformation and the capability of the geometric constraint constructed between features. In this paper, we propose a novel progressive feature matching algorithms based on triplet graph, which will produce a more general local affine transformation estimation method, and results in a powerful affine invariant constraint and efficient MRF optimization in rejecting mismatches. Experimental results on several challenging datasets have illustrated our method can achieve much higher precision/recall than existing methods.

Conventional single image based localization methods usually fail to localize a querying image when there exist large variations between the querying image and the pre-built scene. To address this, we propose an image-set querying based localization approach. When the localization by a single image fails to work, the system will ask the user to capture more auxiliary images. First, a local 3D model is established for the querying image set. Then, the pose of the querying image set is estimated by solving a nonlinear optimization problem, which aims to match the local 3D model against the pre-built scene. Experiments have shown the effectiveness and feasibility of the proposed approach.