Angtian Wang

Yi Zhang, Pengliang Ji, Angtian Wang et al.

Regression-based methods for 3D human pose estimation directly predict the 3D pose parameters from a 2D image using deep networks. While achieving state-of-the-art performance on standard benchmarks, their performance degrades under occlusion. In contrast, optimization-based methods fit a parametric body model to 2D features in an iterative manner. The localized reconstruction loss can potentially make them robust to occlusion, but they suffer from the 2D-3D ambiguity. Motivated by the recent success of generative models in rigid object pose estimation, we propose 3D-aware Neural Body Fitting (3DNBF) - an approximate analysis-by-synthesis approach to 3D human pose estimation with SOTA performance and occlusion robustness. In particular, we propose a generative model of deep features based on a volumetric human representation with Gaussian ellipsoidal kernels emitting 3D pose-dependent feature vectors. The neural features are trained with contrastive learning to become 3D-aware and hence to overcome the 2D-3D ambiguity. Experiments show that 3DNBF outperforms other approaches on both occluded and standard benchmarks. Code is available at https://github.com/edz-o/3DNBF

Angtian Wang, Peng Wang, Jian Sun et al.

The Gaussian reconstruction kernels have been proposed by Westover (1990) and studied by the computer graphics community back in the 90s, which gives an alternative representation of object 3D geometry from meshes and point clouds. On the other hand, current state-of-the-art (SoTA) differentiable renderers, Liu et al. (2019), use rasterization to collect triangles or points on each image pixel and blend them based on the viewing distance. In this paper, we propose VoGE, which utilizes the volumetric Gaussian reconstruction kernels as geometric primitives. The VoGE rendering pipeline uses ray tracing to capture the nearest primitives and blends them as mixtures based on their volume density distributions along the rays. To efficiently render via VoGE, we propose an approximate closeform solution for the volume density aggregation and a coarse-to-fine rendering strategy. Finally, we provide a CUDA implementation of VoGE, which enables real-time level rendering with a competitive rendering speed in comparison to PyTorch3D. Quantitative and qualitative experiment results show VoGE outperforms SoTA counterparts when applied to various vision tasks, e.g., object pose estimation, shape/texture fitting, and occlusion reasoning. The VoGE library and demos are available at: https://github.com/Angtian/VoGE.

Yuanhao Cai, Jiahao Wang, Alan Yuille et al.

X-ray, known for its ability to reveal internal structures of objects, is expected to provide richer information for 3D reconstruction than visible light. Yet, existing neural radiance fields (NeRF) algorithms overlook this important nature of X-ray, leading to their limitations in capturing structural contents of imaged objects. In this paper, we propose a framework, Structure-Aware X-ray Neural Radiodensity Fields (SAX-NeRF), for sparse-view X-ray 3D reconstruction. Firstly, we design a Line Segment-based Transformer (Lineformer) as the backbone of SAX-NeRF. Linefomer captures internal structures of objects in 3D space by modeling the dependencies within each line segment of an X-ray. Secondly, we present a Masked Local-Global (MLG) ray sampling strategy to extract contextual and geometric information in 2D projection. Plus, we collect a larger-scale dataset X3D covering wider X-ray applications. Experiments on X3D show that SAX-NeRF surpasses previous NeRF-based methods by 12.56 and 2.49 dB on novel view synthesis and CT reconstruction. Code, models, and data are released at https://github.com/caiyuanhao1998/SAX-NeRF

Wufei Ma, Angtian Wang, Alan Yuille et al.

We consider the problem of category-level 6D pose estimation from a single RGB image. Our approach represents an object category as a cuboid mesh and learns a generative model of the neural feature activations at each mesh vertex to perform pose estimation through differentiable rendering. A common problem of rendering-based approaches is that they rely on bounding box proposals, which do not convey information about the 3D rotation of the object and are not reliable when objects are partially occluded. Instead, we introduce a coarse-to-fine optimization strategy that utilizes the rendering process to estimate a sparse set of 6D object proposals, which are subsequently refined with gradient-based optimization. The key to enabling the convergence of our approach is a neural feature representation that is trained to be scale- and rotation-invariant using contrastive learning. Our experiments demonstrate an enhanced category-level 6D pose estimation performance compared to prior work, particularly under strong partial occlusion.

Wufei Ma, Qihao Liu, Jiahao Wang et al.

Diffusion models have emerged as a powerful generative method, capable of producing stunning photo-realistic images from natural language descriptions. However, these models lack explicit control over the 3D structure in the generated images. Consequently, this hinders our ability to obtain detailed 3D annotations for the generated images or to craft instances with specific poses and distances. In this paper, we propose 3D Diffusion Style Transfer (3D-DST), which incorporates 3D geometry control into diffusion models. Our method exploits ControlNet, which extends diffusion models by using visual prompts in addition to text prompts. We generate images of the 3D objects taken from 3D shape repositories (e.g., ShapeNet and Objaverse), render them from a variety of poses and viewing directions, compute the edge maps of the rendered images, and use these edge maps as visual prompts to generate realistic images. With explicit 3D geometry control, we can easily change the 3D structures of the objects in the generated images and obtain ground-truth 3D annotations automatically. This allows us to improve a wide range of vision tasks, e.g., classification and 3D pose estimation, in both in-distribution (ID) and out-of-distribution (OOD) settings. We demonstrate the effectiveness of our method through extensive experiments on ImageNet-100/200, ImageNet-R, PASCAL3D+, ObjectNet3D, and OOD-CV. The results show that our method significantly outperforms existing methods, e.g., 3.8 percentage points on ImageNet-100 using DeiT-B.

Chen Wang, Angtian Wang, Junbo Li et al.

Neural Radiance Field (NeRF) has demonstrated excellent quality in novel view synthesis, thanks to its ability to model 3D object geometries in a concise formulation. However, current approaches to NeRF-based models rely on clean images with accurate camera calibration, which can be difficult to obtain in the real world, where data is often subject to corruption and distortion. In this work, we provide the first comprehensive analysis of the robustness of NeRF-based novel view synthesis algorithms in the presence of different types of corruptions. We find that NeRF-based models are significantly degraded in the presence of corruption, and are more sensitive to a different set of corruptions than image recognition models. Furthermore, we analyze the robustness of the feature encoder in generalizable methods, which synthesize images using neural features extracted via convolutional neural networks or transformers, and find that it only contributes marginally to robustness. Finally, we reveal that standard data augmentation techniques, which can significantly improve the robustness of recognition models, do not help the robustness of NeRF-based models. We hope that our findings will attract more researchers to study the robustness of NeRF-based approaches and help to improve their performance in the real world.

Tom Fischer, Yaoyao Liu, Artur Jesslen et al.

Different from human nature, it is still common practice today for vision tasks to train deep learning models only initially and on fixed datasets. A variety of approaches have recently addressed handling continual data streams. However, extending these methods to manage out-of-distribution (OOD) scenarios has not effectively been investigated. On the other hand, it has recently been shown that non-continual neural mesh models exhibit strong performance in generalizing to such OOD scenarios. To leverage this decisive property in a continual learning setting, we propose incremental neural mesh models that can be extended with new meshes over time. In addition, we present a latent space initialization strategy that enables us to allocate feature space for future unseen classes in advance and a positional regularization term that forces the features of the different classes to consistently stay in respective latent space regions. We demonstrate the effectiveness of our method through extensive experiments on the Pascal3D and ObjectNet3D datasets and show that our approach outperforms the baselines for classification by $2-6\%$ in the in-domain and by $6-50\%$ in the OOD setting. Our work also presents the first incremental learning approach for pose estimation. Our code and model can be found at https://github.com/Fischer-Tom/iNeMo.

FSVideo Team, Qingyu Chen, Zhiyuan Fang et al.

We introduce FSVideo, a fast speed transformer-based image-to-video (I2V) diffusion framework. We build our framework on the following key components: 1.) a new video autoencoder with highly-compressed latent space ($64\times64\times4$ spatial-temporal downsampling ratio), achieving competitive reconstruction quality; 2.) a diffusion transformer (DIT) architecture with a new layer memory design to enhance inter-layer information flow and context reuse within DIT, and 3.) a multi-resolution generation strategy via a few-step DIT upsampler to increase video fidelity. Our final model, which contains a 14B DIT base model and a 14B DIT upsampler, achieves competitive performance against other popular open-source models, while being an order of magnitude faster. We discuss our model design as well as training strategies in this report.

Yuanhao Cai, Yixun Liang, Jiahao Wang et al.

X-ray is widely applied for transmission imaging due to its stronger penetration than natural light. When rendering novel view X-ray projections, existing methods mainly based on NeRF suffer from long training time and slow inference speed. In this paper, we propose a 3D Gaussian splatting-based framework, namely X-Gaussian, for X-ray novel view synthesis. Firstly, we redesign a radiative Gaussian point cloud model inspired by the isotropic nature of X-ray imaging. Our model excludes the influence of view direction when learning to predict the radiation intensity of 3D points. Based on this model, we develop a Differentiable Radiative Rasterization (DRR) with CUDA implementation. Secondly, we customize an Angle-pose Cuboid Uniform Initialization (ACUI) strategy that directly uses the parameters of the X-ray scanner to compute the camera information and then uniformly samples point positions within a cuboid enclosing the scanned object. Experiments show that our X-Gaussian outperforms state-of-the-art methods by 6.5 dB while enjoying less than 15% training time and over 73x inference speed. The application on sparse-view CT reconstruction also reveals the practical values of our method. Code is publicly available at https://github.com/caiyuanhao1998/X-Gaussian . A video demo of the training process visualization is at https://www.youtube.com/watch?v=gDVf_Ngeghg .

Yuxiang Lai, Xiaoxi Chen, Angtian Wang et al.

AI for cancer detection encounters the bottleneck of data scarcity, annotation difficulty, and low prevalence of early tumors. Tumor synthesis seeks to create artificial tumors in medical images, which can greatly diversify the data and annotations for AI training. However, current tumor synthesis approaches are not applicable across different organs due to their need for specific expertise and design. This paper establishes a set of generic rules to simulate tumor development. Each cell (pixel) is initially assigned a state between zero and ten to represent the tumor population, and a tumor can be developed based on three rules to describe the process of growth, invasion, and death. We apply these three generic rules to simulate tumor development--from pixel to cancer--using cellular automata. We then integrate the tumor state into the original computed tomography (CT) images to generate synthetic tumors across different organs. This tumor synthesis approach allows for sampling tumors at multiple stages and analyzing tumor-organ interaction. Clinically, a reader study involving three expert radiologists reveals that the synthetic tumors and their developing trajectories are convincingly realistic. Technically, we analyze and simulate tumor development at various stages using 9,262 raw, unlabeled CT images sourced from 68 hospitals worldwide. The performance in segmenting tumors in the liver, pancreas, and kidneys exceeds prevailing literature benchmarks, underlining the immense potential of tumor synthesis, especially for earlier cancer detection. The code and models are available at https://github.com/MrGiovanni/Pixel2Cancer

Xiaoyan Cong, Haotian Yang, Angtian Wang et al.

Instruction-based video editing aims to modify an input video according to a natural-language instruction while preserving content fidelity and temporal coherence. However, existing diffusion-based approaches are often trained on paired data of simple editing operations, which fundamentally limits their ability to generalize to diverse and complex, real-world instructions. To address this generalization gap, we propose VIVA, a scalable framework for instruction-based video editing that leverages VLM-guided encoding and reward optimization. First, we introduce a VLM-based instructor that encodes the textual instruction, the first frame of the source video, and an optional reference image into visually-grounded instruction representations, providing fine-grained spatial and semantic context for the diffusion transformer backbone. Second, we propose a post-training stage, Edit-GRPO, which adapts Group Relative Policy Optimization to the domain of video editing, directly optimizing the model for instruction-faithful, content-preserving, and aesthetically pleasing edits using relative rewards. Furthermore, we propose a data construction pipeline designed to synthetically generate diverse, high-fidelity paired video-instruction data of basic editing operations. Extensive experiments show that VIVA achieves superior instruction following, generalization, and editing quality over state-of-the-art methods. Website: https://viva-paper.github.io

Fengrui Tian, Yueqi Duan, Angtian Wang et al.

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. Codes are available at https://github.com/tianfr/Semantic-Flow/.

Yuming Gu, Yizhi Wang, Yining Hong et al.

Embodied visual planning aims to enable manipulation tasks by imagining how a scene evolves toward a desired goal and using the imagined trajectories to guide actions. Video diffusion models, through their image-to-video generation capability, provide a promising foundation for such visual imagination. However, existing approaches are largely forward predictive, generating trajectories conditioned on the initial observation without explicit goal modeling, thus often leading to spatial drift and goal misalignment. To address these challenges, we propose Envision, a diffusion-based framework that performs visual planning for embodied agents. By explicitly constraining the generation with a goal image, our method enforces physical plausibility and goal consistency throughout the generated trajectory. Specifically, Envision operates in two stages. First, a Goal Imagery Model identifies task-relevant regions, performs region-aware cross attention between the scene and the instruction, and synthesizes a coherent goal image that captures the desired outcome. Then, an Env-Goal Video Model, built upon a first-and-last-frame-conditioned video diffusion model (FL2V), interpolates between the initial observation and the goal image, producing smooth and physically plausible video trajectories that connect the start and goal states. Experiments on object manipulation and image editing benchmarks demonstrate that Envision achieves superior goal alignment, spatial consistency, and object preservation compared to baselines. The resulting visual plans can directly support downstream robotic planning and control, providing reliable guidance for embodied agents.

Kaiwen Zhang, Liming Jiang, Angtian Wang et al.

Visual storytelling requires generating multi-shot videos with cinematic quality and long-range consistency. Inspired by human memory, we propose StoryMem, a paradigm that reformulates long-form video storytelling as iterative shot synthesis conditioned on explicit visual memory, transforming pre-trained single-shot video diffusion models into multi-shot storytellers. This is achieved by a novel Memory-to-Video (M2V) design, which maintains a compact and dynamically updated memory bank of keyframes from historical generated shots. The stored memory is then injected into single-shot video diffusion models via latent concatenation and negative RoPE shifts with only LoRA fine-tuning. A semantic keyframe selection strategy, together with aesthetic preference filtering, further ensures informative and stable memory throughout generation. Moreover, the proposed framework naturally accommodates smooth shot transitions and customized story generation applications. To facilitate evaluation, we introduce ST-Bench, a diverse benchmark for multi-shot video storytelling. Extensive experiments demonstrate that StoryMem achieves superior cross-shot consistency over previous methods while preserving high aesthetic quality and prompt adherence, marking a significant step toward coherent minute-long video storytelling.

Yufan Deng, Yuanyang Yin, Xun Guo et al. · bytedance

We tackle the task of any-reference video generation, which aims to synthesize videos conditioned on arbitrary types and combinations of reference subjects, together with textual prompts. This task faces persistent challenges, including identity inconsistency, entanglement among multiple reference subjects, and copy-paste artifacts. To address these issues, we introduce MAGREF, a unified and effective framework for any-reference video generation. Our approach incorporates masked guidance and a subject disentanglement mechanism, enabling flexible synthesis conditioned on diverse reference images and textual prompts. Specifically, masked guidance employs a region-aware masking mechanism combined with pixel-wise channel concatenation to preserve appearance features of multiple subjects along the channel dimension. This design preserves identity consistency and maintains the capabilities of the pre-trained backbone, without requiring any architectural changes. To mitigate subject confusion, we introduce a subject disentanglement mechanism which injects the semantic values of each subject derived from the text condition into its corresponding visual region. Additionally, we establish a four-stage data pipeline to construct diverse training pairs, effectively alleviating copy-paste artifacts. Extensive experiments on a comprehensive benchmark demonstrate that MAGREF consistently outperforms existing state-of-the-art approaches, paving the way for scalable, controllable, and high-fidelity any-reference video synthesis. Code and model can be found at: https://github.com/MAGREF-Video/MAGREF

Angtian Wang, Adam Kortylewski, Alan Yuille

3D pose estimation is a challenging but important task in computer vision. In this work, we show that standard deep learning approaches to 3D pose estimation are not robust when objects are partially occluded or viewed from a previously unseen pose. Inspired by the robustness of generative vision models to partial occlusion, we propose to integrate deep neural networks with 3D generative representations of objects into a unified neural architecture that we term NeMo. In particular, NeMo learns a generative model of neural feature activations at each vertex on a dense 3D mesh. Using differentiable rendering we estimate the 3D object pose by minimizing the reconstruction error between NeMo and the feature representation of the target image. To avoid local optima in the reconstruction loss, we train the feature extractor to maximize the distance between the individual feature representations on the mesh using contrastive learning. Our extensive experiments on PASCAL3D+, occluded-PASCAL3D+ and ObjectNet3D show that NeMo is much more robust to partial occlusion and unseen pose compared to standard deep networks, while retaining competitive performance on regular data. Interestingly, our experiments also show that NeMo performs reasonably well even when the mesh representation only crudely approximates the true object geometry with a cuboid, hence revealing that the detailed 3D geometry is not needed for accurate 3D pose estimation. The code is publicly available at https://github.com/Angtian/NeMo.

Yufan Deng, Xun Guo, Yizhi Wang et al. · bytedance

Video generation has witnessed remarkable progress with the advent of deep generative models, particularly diffusion models. While existing methods excel in generating high-quality videos from text prompts or single images, personalized multi-subject video generation remains a largely unexplored challenge. This task involves synthesizing videos that incorporate multiple distinct subjects, each defined by separate reference images, while ensuring temporal and spatial consistency. Current approaches primarily rely on mapping subject images to keywords in text prompts, which introduces ambiguity and limits their ability to model subject relationships effectively. In this paper, we propose CINEMA, a novel framework for coherent multi-subject video generation by leveraging Multimodal Large Language Model (MLLM). Our approach eliminates the need for explicit correspondences between subject images and text entities, mitigating ambiguity and reducing annotation effort. By leveraging MLLM to interpret subject relationships, our method facilitates scalability, enabling the use of large and diverse datasets for training. Furthermore, our framework can be conditioned on varying numbers of subjects, offering greater flexibility in personalized content creation. Through extensive evaluations, we demonstrate that our approach significantly improves subject consistency, and overall video coherence, paving the way for advanced applications in storytelling, interactive media, and personalized video generation.

Angtian Wang, Yuanlu Xu, Nikolaos Sarafianos et al. · meta-ai

Neural reconstruction and rendering strategies have demonstrated state-of-the-art performances due, in part, to their ability to preserve high level shape details. Existing approaches, however, either represent objects as implicit surface functions or neural volumes and still struggle to recover shapes with heterogeneous materials, in particular human skin, hair or clothes. To this aim, we present a new hybrid implicit surface representation to model human shapes. This representation is composed of two surface layers that represent opaque and translucent regions on the clothed human body. We segment different regions automatically using visual cues and learn to reconstruct two signed distance functions (SDFs). We perform surface-based rendering on opaque regions (e.g., body, face, clothes) to preserve high-fidelity surface normals and volume rendering on translucent regions (e.g., hair). Experiments demonstrate that our approach obtains state-of-the-art results on 3D human reconstructions, and also shows competitive performances on other objects.

Angtian Wang, Haibin Huang, Jacob Zhiyuan Fang et al. · bytedance

We propose a unified framework for motion control in video generation that seamlessly integrates camera movement, object-level translation, and fine-grained local motion using trajectory-based inputs. In contrast to prior methods that address these motion types through separate modules or task-specific designs, our approach offers a cohesive solution by projecting user-defined trajectories into the latent space of pre-trained image-to-video generation models via a lightweight motion injector. Users can specify keypoints and their motion paths to control localized deformations, entire object motion, virtual camera dynamics, or combinations of these. The injected trajectory signals guide the generative process to produce temporally consistent and semantically aligned motion sequences. Our framework demonstrates superior performance across multiple video motion control tasks, including stylized motion effects (e.g., motion brushes), dynamic viewpoint changes, and precise local motion manipulation. Experiments show that our method provides significantly better controllability and visual quality compared to prior approaches and commercial solutions, while remaining broadly compatible with various state-of-the-art video generation backbones. Project page: https://anytraj.github.io/.

Zhiheng Liu, Xueqing Deng, Shoufa Chen et al.

Generative video modeling has made significant strides, yet ensuring structural and temporal consistency over long sequences remains a challenge. Current methods predominantly rely on RGB signals, leading to accumulated errors in object structure and motion over extended durations. To address these issues, we introduce WorldWeaver, a robust framework for long video generation that jointly models RGB frames and perceptual conditions within a unified long-horizon modeling scheme. Our training framework offers three key advantages. First, by jointly predicting perceptual conditions and color information from a unified representation, it significantly enhances temporal consistency and motion dynamics. Second, by leveraging depth cues, which we observe to be more resistant to drift than RGB, we construct a memory bank that preserves clearer contextual information, improving quality in long-horizon video generation. Third, we employ segmented noise scheduling for training prediction groups, which further mitigates drift and reduces computational cost. Extensive experiments on both diffusion- and rectified flow-based models demonstrate the effectiveness of WorldWeaver in reducing temporal drift and improving the fidelity of generated videos.

Yifan Yin, Zhengtao Han, Shivam Aarya et al.

Fine-grained robot manipulation, such as lifting and rotating a bottle to display the label on the cap, requires robust reasoning about object parts and their relationships with intended tasks. Despite recent advances in training general-purpose robot manipulation policies guided by language instructions, there is a notable lack of large-scale datasets for fine-grained manipulation tasks with part-level instructions and diverse 3D object instances annotated with part-level labels. In this work, we introduce PartInstruct, the first large-scale benchmark for training and evaluating fine-grained robot manipulation models using part-level instructions. PartInstruct comprises 513 object instances across 14 categories, each annotated with part-level information, and 1302 fine-grained manipulation tasks organized into 16 task classes. Our training set consists of over 10,000 expert demonstrations synthesized in a 3D simulator, where each demonstration is paired with a high-level task instruction, a chain of base part-based skill instructions, and ground-truth 3D information about the object and its parts. Additionally, we designed a comprehensive test suite to evaluate the generalizability of learned policies across new states, objects, and tasks. We evaluated several state-of-the-art robot manipulation approaches, including end-to-end vision-language policy learning and bi-level planning models for robot manipulation on our benchmark. The experimental results reveal that current models struggle to robustly ground part concepts and predict actions in 3D space, and face challenges when manipulating object parts in long-horizon tasks.

Guofeng Zhang, Angtian Wang, Jacob Zhiyuan Fang et al. · bytedance

Text-to-video generation has advanced rapidly in visual fidelity, whereas standard methods still have limited ability to control the subject composition of generated scenes. Prior work shows that adding localized text control signals, such as bounding boxes or segmentation masks, can help. However, these methods struggle in complex scenarios and degrade in multi-object settings, offering limited precision and lacking a clear correspondence between individual trajectories and visual entities as the number of controllable objects increases. We introduce Text-Grounded Trajectories (TGT), a framework that conditions video generation on trajectories paired with localized text descriptions. We propose Location-Aware Cross-Attention (LACA) to integrate these signals and adopt a dual-CFG scheme to separately modulate local and global text guidance. In addition, we develop a data processing pipeline that produces trajectories with localized descriptions of tracked entities, and we annotate two million high quality video clips to train TGT. Together, these components enable TGT to use point trajectories as intuitive motion handles, pairing each trajectory with text to control both appearance and motion. Extensive experiments show that TGT achieves higher visual quality, more accurate text alignment, and improved motion controllability compared with prior approaches. Website: https://textgroundedtraj.github.io.

Xingrui Wang, Wufei Ma, Angtian Wang et al.

For vision-language models (VLMs), understanding the dynamic properties of objects and their interactions in 3D scenes from videos is crucial for effective reasoning about high-level temporal and action semantics. Although humans are adept at understanding these properties by constructing 3D and temporal (4D) representations of the world, current video understanding models struggle to extract these dynamic semantics, arguably because these models use cross-frame reasoning without underlying knowledge of the 3D/4D scenes. In this work, we introduce DynSuperCLEVR, the first video question answering dataset that focuses on language understanding of the dynamic properties of 3D objects. We concentrate on three physical concepts -- velocity, acceleration, and collisions within 4D scenes. We further generate three types of questions, including factual queries, future predictions, and counterfactual reasoning that involve different aspects of reasoning about these 4D dynamic properties. To further demonstrate the importance of explicit scene representations in answering these 4D dynamics questions, we propose NS-4DPhysics, a Neural-Symbolic VideoQA model integrating Physics prior for 4D dynamic properties with explicit scene representation of videos. Instead of answering the questions directly from the video text input, our method first estimates the 4D world states with a 3D generative model powered by physical priors, and then uses neural symbolic reasoning to answer the questions based on the 4D world states. Our evaluation on all three types of questions in DynSuperCLEVR shows that previous video question answering models and large multimodal models struggle with questions about 4D dynamics, while our NS-4DPhysics significantly outperforms previous state-of-the-art models. Our code and data are released in https://xingruiwang.github.io/projects/DynSuperCLEVR/.

Fengrui Tian, Yaoyao Liu, Adam Kortylewski et al.

3D object pose estimation is a challenging task. Previous works always require thousands of object images with annotated poses for learning the 3D pose correspondence, which is laborious and time-consuming for labeling. In this paper, we propose to learn a category-level 3D object pose estimator without pose annotations. Instead of using manually annotated images, we leverage diffusion models (e.g., Zero-1-to-3) to generate a set of images under controlled pose differences and propose to learn our object pose estimator with those images. Directly using the original diffusion model leads to images with noisy poses and artifacts. To tackle this issue, firstly, we exploit an image encoder, which is learned from a specially designed contrastive pose learning, to filter the unreasonable details and extract image feature maps. Additionally, we propose a novel learning strategy that allows the model to learn object poses from those generated image sets without knowing the alignment of their canonical poses. Experimental results show that our method has the capability of category-level object pose estimation from a single shot setting (as pose definition), while significantly outperforming other state-of-the-art methods on the few-shot category-level object pose estimation benchmarks.

Angtian Wang, Wufei Ma, Alan Yuille et al.

Human vision demonstrates higher robustness than current AI algorithms under out-of-distribution scenarios. It has been conjectured such robustness benefits from performing analysis-by-synthesis. Our paper formulates triple vision tasks in a consistent manner using approximate analysis-by-synthesis by render-and-compare algorithms on neural features. In this work, we introduce Neural Textured Deformable Meshes, which involve the object model with deformable geometry that allows optimization on both camera parameters and object geometries. The deformable mesh is parameterized as a neural field, and covered by whole-surface neural texture maps, which are trained to have spatial discriminability. During inference, we extract the feature map of the test image and subsequently optimize the 3D pose and shape parameters of our model using differentiable rendering to best reconstruct the target feature map. We show that our analysis-by-synthesis is much more robust than conventional neural networks when evaluated on real-world images and even in challenging out-of-distribution scenarios, such as occlusion and domain shift. Our algorithms are competitive with standard algorithms when tested on conventional performance measures.

Jiahao Yang, Wufei Ma, Angtian Wang et al.

Obtaining accurate 3D object poses is vital for numerous computer vision applications, such as 3D reconstruction and scene understanding. However, annotating real-world objects is time-consuming and challenging. While synthetically generated training data is a viable alternative, the domain shift between real and synthetic data is a significant challenge. In this work, we aim to narrow the performance gap between models trained on synthetic data and few real images and fully supervised models trained on large-scale data. We achieve this by approaching the problem from two perspectives: 1) We introduce SyntheticP3D, a new synthetic dataset for object pose estimation generated from CAD models and enhanced with a novel algorithm. 2) We propose a novel approach (CC3D) for training neural mesh models that perform pose estimation via inverse rendering. In particular, we exploit the spatial relationships between features on the mesh surface and a contrastive learning scheme to guide the domain adaptation process. Combined, these two approaches enable our models to perform competitively with state-of-the-art models using only 10% of the respective real training images, while outperforming the SOTA model by 10.4% with a threshold of pi/18 using only 50% of the real training data. Our trained model further demonstrates robust generalization to out-of-distribution scenarios despite being trained with minimal real data.

Artur Jesslen, Guofeng Zhang, Angtian Wang et al.

Discriminative models for object classification typically learn image-based representations that do not capture the compositional and 3D nature of objects. In this work, we show that explicitly integrating 3D compositional object representations into deep networks for image classification leads to a largely enhanced generalization in out-of-distribution scenarios. In particular, we introduce a novel architecture, referred to as NOVUM, that consists of a feature extractor and a neural object volume for every target object class. Each neural object volume is a composition of 3D Gaussians that emit feature vectors. This compositional object representation allows for a highly robust and fast estimation of the object class by independently matching the features of the 3D Gaussians of each category to features extracted from an input image. Additionally, the object pose can be estimated via inverse rendering of the corresponding neural object volume. To enable the classification of objects, the neural features at each 3D Gaussian are trained discriminatively to be distinct from (i) the features of 3D Gaussians in other categories, (ii) features of other 3D Gaussians of the same object, and (iii) the background features. Our experiments show that NOVUM offers intriguing advantages over standard architectures due to the 3D compositional structure of the object representation, namely: (1) An exceptional robustness across a spectrum of real-world and synthetic out-of-distribution shifts and (2) an enhanced human interpretability compared to standard models, all while maintaining real-time inference and a competitive accuracy on in-distribution data.

Bingchen Zhao, Shaozuo Yu, Wufei Ma et al.

Enhancing the robustness of vision algorithms in real-world scenarios is challenging. One reason is that existing robustness benchmarks are limited, as they either rely on synthetic data or ignore the effects of individual nuisance factors. We introduce OOD-CV, a benchmark dataset that includes out-of-distribution examples of 10 object categories in terms of pose, shape, texture, context and the weather conditions, and enables benchmarking models for image classification, object detection, and 3D pose estimation. In addition to this novel dataset, we contribute extensive experiments using popular baseline methods, which reveal that: 1. Some nuisance factors have a much stronger negative effect on the performance compared to others, also depending on the vision task. 2. Current approaches to enhance robustness have only marginal effects, and can even reduce robustness. 3. We do not observe significant differences between convolutional and transformer architectures. We believe our dataset provides a rich testbed to study robustness and will help push forward research in this area.

Angtian Wang, Shenxiao Mei, Alan Yuille et al.

We study the problem of learning to estimate the 3D object pose from a few labelled examples and a collection of unlabelled data. Our main contribution is a learning framework, neural view synthesis and matching, that can transfer the 3D pose annotation from the labelled to unlabelled images reliably, despite unseen 3D views and nuisance variations such as the object shape, texture, illumination or scene context. In our approach, objects are represented as 3D cuboid meshes composed of feature vectors at each mesh vertex. The model is initialized from a few labelled images and is subsequently used to synthesize feature representations of unseen 3D views. The synthesized views are matched with the feature representations of unlabelled images to generate pseudo-labels of the 3D pose. The pseudo-labelled data is, in turn, used to train the feature extractor such that the features at each mesh vertex are more invariant across varying 3D views of the object. Our model is trained in an EM-type manner alternating between increasing the 3D pose invariance of the feature extractor and annotating unlabelled data through neural view synthesis and matching. We demonstrate the effectiveness of the proposed semi-supervised learning framework for 3D pose estimation on the PASCAL3D+ and KITTI datasets. We find that our approach outperforms all baselines by a wide margin, particularly in an extreme few-shot setting where only 7 annotated images are given. Remarkably, we observe that our model also achieves an exceptional robustness in out-of-distribution scenarios that involve partial occlusion.

Yutong Bai, Angtian Wang, Adam Kortylewski et al.

In this paper, we introduce a contrastive learning framework for keypoint detection (CoKe). Keypoint detection differs from other visual tasks where contrastive learning has been applied because the input is a set of images in which multiple keypoints are annotated. This requires the contrastive learning to be extended such that the keypoints are represented and detected independently, which enables the contrastive loss to make the keypoint features different from each other and from the background. Our approach has two benefits: It enables us to exploit contrastive learning for keypoint detection, and by detecting each keypoint independently the detection becomes more robust to occlusion compared to holistic methods, such as stacked hourglass networks, which attempt to detect all keypoints jointly. Our CoKe framework introduces several technical innovations. In particular, we introduce: (i) A clutter bank to represent non-keypoint features; (ii) a keypoint bank that stores prototypical representations of keypoints to approximate the contrastive loss between keypoints; and (iii) a cumulative moving average update to learn the keypoint prototypes while training the feature extractor. Our experiments on a range of diverse datasets (PASCAL3D+, MPII, ObjectNet3D) show that our approach works as well, or better than, alternative methods for keypoint detection, even for human keypoints, for which the literature is vast. Moreover, we observe that CoKe is exceptionally robust to partial occlusion and previously unseen object poses.

Adam Kortylewski, Qing Liu, Angtian Wang et al.

Computer vision systems in real-world applications need to be robust to partial occlusion while also being explainable. In this work, we show that black-box deep convolutional neural networks (DCNNs) have only limited robustness to partial occlusion. We overcome these limitations by unifying DCNNs with part-based models into Compositional Convolutional Neural Networks (CompositionalNets) - an interpretable deep architecture with innate robustness to partial occlusion. Specifically, we propose to replace the fully connected classification head of DCNNs with a differentiable compositional model that can be trained end-to-end. The structure of the compositional model enables CompositionalNets to decompose images into objects and context, as well as to further decompose object representations in terms of individual parts and the objects' pose. The generative nature of our compositional model enables it to localize occluders and to recognize objects based on their non-occluded parts. We conduct extensive experiments in terms of image classification and object detection on images of artificially occluded objects from the PASCAL3D+ and ImageNet dataset, and real images of partially occluded vehicles from the MS-COCO dataset. Our experiments show that CompositionalNets made from several popular DCNN backbones (VGG-16, ResNet50, ResNext) improve by a large margin over their non-compositional counterparts at classifying and detecting partially occluded objects. Furthermore, they can localize occluders accurately despite being trained with class-level supervision only. Finally, we demonstrate that CompositionalNets provide human interpretable predictions as their individual components can be understood as detecting parts and estimating an objects' viewpoint.

Angtian Wang, Yihong Sun, Adam Kortylewski et al.

Detecting partially occluded objects is a difficult task. Our experimental results show that deep learning approaches, such as Faster R-CNN, are not robust at object detection under occlusion. Compositional convolutional neural networks (CompositionalNets) have been shown to be robust at classifying occluded objects by explicitly representing the object as a composition of parts. In this work, we propose to overcome two limitations of CompositionalNets which will enable them to detect partially occluded objects: 1) CompositionalNets, as well as other DCNN architectures, do not explicitly separate the representation of the context from the object itself. Under strong object occlusion, the influence of the context is amplified which can have severe negative effects for detection at test time. In order to overcome this, we propose to segment the context during training via bounding box annotations. We then use the segmentation to learn a context-aware CompositionalNet that disentangles the representation of the context and the object. 2) We extend the part-based voting scheme in CompositionalNets to vote for the corners of the object's bounding box, which enables the model to reliably estimate bounding boxes for partially occluded objects. Our extensive experiments show that our proposed model can detect objects robustly, increasing the detection performance of strongly occluded vehicles from PASCAL3D+ and MS-COCO by 41% and 35% respectively in absolute performance relative to Faster R-CNN.

Yuyin Zhou, Yingwei Li, Zhishuai Zhang et al.

Pancreatic ductal adenocarcinoma (PDAC) is one of the most lethal cancers with an overall five-year survival rate of 8%. Due to subtle texture changes of PDAC, pancreatic dual-phase imaging is recommended for better diagnosis of pancreatic disease. In this study, we aim at enhancing PDAC automatic segmentation by integrating multi-phase information (i.e., arterial phase and venous phase). To this end, we present Hyper-Pairing Network (HPN), a 3D fully convolution neural network which effectively integrates information from different phases. The proposed approach consists of a dual path network where the two parallel streams are interconnected with hyper-connections for intensive information exchange. Additionally, a pairing loss is added to encourage the commonality between high-level feature representations of different phases. Compared to prior arts which use single phase data, HPN reports a significant improvement up to 7.73% (from 56.21% to 63.94%) in terms of DSC.