Vision-Language Continual Learning (VLCL) has attracted significant research attention for its robust capabilities, and the adoption of Parameter-Efficient Fine-Tuning (PEFT) strategies is enabling these models to achieve competitive performance with substantially reduced resource consumption. However, dominated First-Order (FO) optimization is prone to trap models in suboptimal local minima, especially in limited exploration subspace within PEFT. To overcome this challenge, this paper pioneers a systematic exploration of adopting Zeroth-Order (ZO) optimization for PEFT-based VLCL. We first identify the incompatibility of naive full-ZO adoption in VLCL due to optimization process instability. We then investigate the application of ZO optimization from a modality branch-wise to a fine-grained layer-wise across various training units to identify an optimal strategy. Besides, a key theoretical insight reveals that vision modality exhibit higher variance than language counterparts in VLCL during the ZO optimization process, and we propose a modality-aware stabilized ZO strategy, which adopts gradient sign normalization in ZO and constrains vision modality perturbation to further improve performance. Benefiting from the adoption of ZO optimization, PEFT-based VLCL fulfills better ability to escape local minima during the optimization process, extensive experiments on four benchmarks demonstrate that our method achieves state-of-the-art results.

Kairui Hu, Penghao Wu, Fanyi Pu, Wang Xiao, Xiang Yue, Bo Li, Yuanhan Zhang, Ziwei Liu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Robust deployment of large multimodal models (LMMs) in real-world scenarios requires access to external knowledge sources, given the complexity and dynamic nature of real-world information. Existing approaches such as retrieval-augmented generation (RAG) and prompt engineered search agents rely on rigid pipelines, often leading to inefficient or excessive search behaviors. We present MMSearch-R1, the first end-to-end reinforcement learning framework that enables LMMs to perform on-demand, multi-turn search in real-world Internet environments. Our framework integrates both image and text search tools, allowing the model to reason about when and how to invoke them guided by an outcome-based reward with a search penalty. To support training, We collect a multimodal search VQA dataset through a semi-automated pipeline that covers diverse visual and textual knowledge needs and curate a search-balanced subset with both search-required and search-free samples, which proves essential for shaping efficient and on-demand search behavior. Extensive experiments on knowledge-intensive and info-seeking VQA tasks show that our model not only outperforms RAG-based baselines of the same model size, but also matches the performance of a larger RAG-based model while reducing search calls by over 30%. We further analyze key empirical findings to offer actionable insights for advancing research in multimodal search.

Kai Zou, Ziqi Huang, Yuhao Dong, Shulin Tian, Dian Zheng, Hongbo Liu, Jingwen He, Bin Liu, Yu Qiao, Ziwei Liu. Proceedings of the 64th Annual Meeting of the Association for Computational Linguistics (Volume 1: Long Papers). 2026.

Abstract Immersive applications call for synthesizing spatiotemporal 4D content from casual videos without costly 3D supervision. Existing video‐to‐4D methods typically rely on manually annotated camera poses, which are labor‐intensive and brittle for in‐the‐wild footage. Recent warp‐then‐inpaint approaches mitigate the need for pose labels by warping input frames along a novel camera trajectory and using an inpainting model to fill missing regions, thereby depicting the 4D scene from diverse viewpoints. However, this trajectory‐to‐trajectory formulation often entangles camera motion with scene dynamics and complicates both modeling and inference. We introduce S ee 4D , a pose‐free, trajectory‐to‐camera framework that replaces explicit trajectory prediction with rendering to a bank of fixed virtual cameras, thereby separating camera control from scene modeling. A view‐conditional video inpainting model is trained to learn a robust geometry prior by denoising realistically synthesized warped images and to inpaint occluded or missing regions across virtual viewpoints, eliminating the need for explicit 3D annotations. Building on this inpainting core, we design a spatiotemporal autoregressive inference pipeline that traverses virtual‐camera splines and extends videos with overlapping windows, enabling coherent generation at bounded per‐step complexity. We validate See4D on cross‐view video generation and sparse reconstruction benchmarks. Across quantitative metrics and qualitative assessments, our method achieves superior generalization and improved performance relative to pose‐ or trajectory‐conditioned baselines, advancing practical 4D world modeling from casual videos.

3D content inherently encompasses multi-modal characteristics and can be projected into different modalities (e.g., RGB images, RGBD, and point clouds).Each modality exhibits distinct advantages in 3D asset modeling: RGB images contain vivid 3D textures, whereas point clouds define fine-grained 3D geometries. However, most existing 3D-native generative architectures either operate predominantly within single-modality paradigms-thus overlooking the complementary benefits of multi-modality data-or restrict themselves to 3D structures, thereby limiting the scope of available training datasets. To holistically harness multi-modalities for 3D modeling, we present TriMM, the first feed-forward 3D-native generative model that learns from basic multi-modalities (e.g., RGB, RGBD, and point cloud). Specifically, 1) TriMM first introduces collaborative multi-modal coding, which integrates modality-specific features while preserving their unique representational strengths. 2) Furthermore, auxiliary 2D and 3D supervision are introduced to raise the robustness and performance of multi-modal coding. 3) Based on the embedded multi-modal code, TriMM employs a triplane latent diffusion model to generate 3D assets of superior quality, enhancing both the texture and the geometric detail. Extensive experiments on multiple well-known datasets demonstrate that TriMM, by effectively leveraging multi-modality, achieves competitive performance with models trained on large-scale datasets, despite utilizing a small amount of training data. Furthermore, we conduct additional experiments on recent RGB-D datasets, verifying the feasibility of incorporating other multi-modal datasets into 3D generation.

Video generation has witnessed significant advancements, yet evaluating these models remains a challenge. A comprehensive evaluation benchmark for video generation is indispensable for two reasons: 1) Existing metrics do not fully align with human perceptions; 2) An ideal evaluation system should provide insights to inform future developments of video generation. To this end, we present VBench++, a comprehensive benchmark suite that dissects "video generation quality" into specific, hierarchical, and disentangled dimensions, each with tailored prompts and evaluation methods. VBench++ has several appealing properties: 1) Comprehensive Dimensions: VBench++ comprises 16 dimensions in text-to-video generation (e.g., subject identity inconsistency, motion smoothness, temporal flickering, and spatial relationship, etc). The evaluation metrics with fine-grained levels reveal individual models' strengths and weaknesses. 2) Human Alignment: We also provide a dataset of human preference annotations to validate our benchmarks' alignment with human perception, for each evaluation dimension respectively. 3) Valuable Insights: We look into current models' ability across various evaluation dimensions, and various content types. We also investigate the gaps between video and image generation models. 4) Versatile Benchmarking: VBench++ is designed to evaluate a wide range of video generation tasks, including text-to-video and image-to-video. We introduce a high-quality Image Suite with an adaptive aspect ratio to enable fair evaluations across different image-to-video generation settings. Beyond assessing technical quality, VBench++ evaluates the trustworthiness of video generative models, providing a more holistic view of model performance. 5) Full Open-Sourcing: We fully open-source VBench++, including all prompts, the Image Suite, evaluation methods, generated videos, and human preference annotations.

Recent advancements in vision foundation models (VFMs) have revolutionized visual perception in 2D, yet their potential for 3D scene understanding, particularly in autonomous driving applications, remains underexplored. In this paper, we introduce LargeAD, a versatile and scalable framework designed for large-scale 3D pretraining across diverse real-world driving datasets. Our framework leverages VFMs to extract semantically rich superpixels from 2D images, which are aligned with LiDAR point clouds to generate high-quality contrastive samples. This alignment facilitates cross-modal representation learning, enhancing the semantic consistency between 2D and 3D data. We introduce several key innovations: (i) VFM-driven superpixel generation for detailed semantic representation, (ii) a VFM-assisted contrastive learning strategy to align multimodal features, (iii) superpoint temporal consistency to maintain stable representations across time, and (iv) multi-source data pretraining to generalize across various LiDAR configurations. Our approach achieves substantial gains over state-of-the-art methods in linear probing and fine-tuning for LiDAR-based segmentation and object detection. Extensive experiments on 11 large-scale multi-sensor datasets highlight our superior performance, demonstrating adaptability, efficiency, and robustness in real-world autonomous driving scenarios.

3D scene generation has garnered growing attention in recent years and has made significant progress. Generating 4D cities is more challenging than 3D scenes due to the presence of structurally complex, visually diverse objects like buildings and vehicles, and heightened human sensitivity to distortions in urban environments. To tackle these issues, we propose CityDreamer4D, a compositional generative model specifically tailored for generating unbounded 4D cities. Our main insights are 1) 4D city generation should separate dynamic objects (e.g., vehicles) from static scenes (e.g., buildings and roads), and 2) all objects in the 4D scene should be composed of different types of neural fields for buildings, vehicles, and background stuff. Specifically, we propose Traffic Scenario Generator and Unbounded Layout Generator to produce dynamic traffic scenarios and static city layouts using a highly compact BEV representation. Objects in 4D cities are generated by combining stuff-oriented and instance-oriented neural fields for background stuff, buildings, and vehicles. To suit the distinct characteristics of background stuff and instances, the neural fields employ customized generative hash grids and periodic positional embeddings as scene parameterizations. Furthermore, we offer a comprehensive suite of datasets for city generation, including OSM, GoogleEarth, and CityTopia. The OSM dataset provides a variety of real-world city layouts, while the Google Earth and CityTopia datasets deliver large-scale, high-quality city imagery complete with 3D instance annotations. Leveraging its compositional design, CityDreamer4D supports a range of downstream applications, such as instance editing, city stylization, and urban simulation, while delivering state-of-the-art performance in generating realistic 4D cities.

LiDAR representation learning has emerged as a promising approach to reducing reliance on costly and labor-intensive human annotations. While existing methods primarily focus on spatial alignment between LiDAR and camera sensors, they often overlook the temporal dynamics critical for capturing motion and scene continuity in driving scenarios. To address this limitation, we propose SuperFlow++, a novel framework that integrates spatiotemporal cues in both pretraining and downstream tasks using consecutive LiDAR-camera pairs. SuperFlow++ introduces four key components: (1) a view consistency alignment module to unify semantic information across camera views, (2) a dense-to-sparse consistency regularization mechanism to enhance feature robustness across varying point cloud densities, (3) a flow-based contrastive learning approach that models temporal relationships for improved scene understanding, and (4) a temporal voting strategy that propagates semantic information across LiDAR scans to improve prediction consistency. Extensive evaluations on 11 heterogeneous LiDAR datasets demonstrate that SuperFlow++ outperforms state-of-the-art methods across diverse tasks and driving conditions. Furthermore, by scaling both 2D and 3D backbones during pretraining, we uncover emergent properties that provide deeper insights into developing scalable 3D foundation models. With strong generalizability and computational efficiency, SuperFlow++ establishes a new benchmark for data-efficient LiDAR-based perception in autonomous driving.

Expressive human pose and shape estimation (EHPS) unifies body, hands, and face motion capture with numerous applications. Despite encouraging progress, current state-of-the-art methods focus on training innovative architectural designs on confined datasets. In this work, we investigate the impact of scaling up EHPS towards a family of generalist foundation models. 1) For data scaling, we perform a systematic investigation on 40 EHPS datasets, encompassing a wide range of scenarios that a model trained on any single dataset cannot handle. More importantly, capitalizing on insights obtained from the extensive benchmarking process, we optimize our training scheme and select datasets that lead to a significant leap in EHPS capabilities. Ultimately, we achieve diminishing returns at 10 M training instances from diverse data sources. 2) For model scaling, we take advantage of vision transformers (up to ViT-Huge as the backbone) to study the scaling law of model sizes in EHPS. To exclude the influence of algorithmic design, we base our experiments on two minimalist architectures: SMPLer-X, which consists of an intermediate step for hand and face localization, and SMPLest-X, an even simpler version that reduces the network to its bare essentials and highlights significant advances in the capture of articulated hands. With Big Data and the large model, the foundation models exhibit strong performance across diverse test benchmarks and excellent transferability to even unseen environments. Moreover, our finetuning strategy turns the generalist into specialist models, allowing them to achieve further performance boosts. Notably, our foundation models consistently deliver state-of-the-art results on seven benchmarks such as AGORA, UBody, EgoBody, and our proposed SynHand dataset for comprehensive hand evaluation.

Generating high-quality character animation videos is a fascinating yet challenging task. Existing methods use geometry guidance signals like skeletons, normal maps, or depth maps in a diffusion model to generate character videos from a single reference image. Although these approaches have shown encouraging results, they solely rely on cross attention layers to extract geometry guidance which inevitably leads to temporal inconsistencies and reduced quality. In this paper, we present a novel framework AniFeats to generate high-quality character animation videos. In contrast to existing methods, our key insight is to incorporate explicit features on 3D character meshes during the video generation to achieve significantly improved temporal consistency. Specifically, AniFeats extracts detailed features from the reference image, projects them onto 3D feature meshes based on SMPL-X, and utilizes rendered feature maps from the animated 3D feature meshes as guidance throughout the generation process. This approach directly links local patterns in the input image to those in the output video, effectively strengthening temporal coherence. Extensive experiments demonstrate that AniFeats generates high-quality, temporally consistent character animations with remarkably enhanced realism.

Tree of Thoughts (ToT) enhances Large Language Model (LLM) reasoning by structuring problem-solving as a spanning tree. However, recent methods focus on search accuracy while overlooking computational efficiency. The challenges of accelerating the ToT lie in the frequent switching of reasoning focus, and the redundant exploration of suboptimal solutions. To alleviate this dilemma, we propose Dynamic Parallel Tree Search (DPTS), a novel parallelism framework that aims to dynamically optimize the reasoning path in inference. It includes the Parallelism Streamline in the generation phase to build up a flexible and adaptive parallelism with arbitrary paths by cache management and alignment. Meanwhile, the Search and Transition Mechanism filters potential candidates to dynamically maintain the reasoning focus on more possible solutions with less redundancy. Experiments on Qwen-2.5 and Llama-3 on math and code datasets show that DPTS significantly improves efficiency by 2-4\times on average while maintaining or even surpassing existing reasoning algorithms in accuracy, making ToT-based reasoning more scalable and computationally efficient. Codes are released at: https://github.com/yifu-ding/DPTS.

This paper introduces a novel task to evaluate the robust understanding capability of Large Multimodal Models (LMMs), termed Unsolvable Problem Detection (UPD). Multiple-choice question answering (MCQA) is widely used to assess the understanding capability of LMMs, but it does not guarantee that LMMs truly comprehend the answer. UPD assesses the LMM’s ability to withhold answers when encountering unsolvable problems of MCQA, verifying whether the model truly understands the answer. UPD encompasses three problems: Absent Answer Detection (AAD), Incompatible Answer Set Detection (IASD), and Incompatible Visual Question Detection (IVQD), covering unsolvable cases like answer-lacking or incompatible choices and image-question mismatches. For the evaluation, we introduce the MM-UPD Bench, a benchmark for assessing performance across various ability dimensions. Our experiments reveal that even most LMMs, which demonstrate adequate performance on existing benchmarks, struggle significantly with MM-UPD, underscoring a novel aspect of trustworthiness that current benchmarks have overlooked. A detailed analysis shows that LMMs have different bottlenecks and chain-of-thought and self-reflection improved performance for LMMs with the bottleneck in their LLM capability. We hope our insights will enhance the broader understanding and development of more reliable LMMs.

Recent advancements in visual generative models have enabled high-quality image and video generation, opening diverse applications. However, evaluating these models often demands sampling hundreds or thousands of images or videos, making the process computationally expensive, especially for diffusion-based models with inherently slow sampling. Moreover, existing evaluation methods rely on rigid pipelines that overlook specific user needs and provide numerical results without clear explanations. In contrast, humans can quickly form impressions of a model’s capabilities by observing only a few samples. To mimic this, we propose the Evaluation Agent framework, which employs human-like strategies for efficient, dynamic, multi-round evaluations using only a few samples per round, while offering detailed, user-tailored analyses. It offers four key advantages: 1) efficiency, 2) promptable evaluation tailored to diverse user needs, 3) explainability beyond single numerical scores, and 4) scalability across various models and tools. Experiments show that Evaluation Agent reduces evaluation time to 10% of traditional methods while delivering comparable results. The Evaluation Agent framework is fully open-sourced to advance research in visual generative models and their efficient evaluation.

Existing research on avatar creation is typically limited to laboratory datasets, which require high costs against scalability and exhibit insufficient representation of the real world. On the other hand, the web abounds with off-the-shelf real- world human videos, but these videos vary in quality and require accurate annotations for avatar creation. To this end, we propose an automatic annotating pipeline with filtering protocols to curate these humans from the web. Our pipeline surpasses state-of-the-art methods on the EMDB benchmark, and the filtering protocols boost verification metrics on web videos. We then curate WildAvatar, a web-scale in-the-wild human avatar creation dataset extracted from YouTube, with 10,000+ different human subjects and scenes. WildAvatar is at least 10 × richer than previous datasets for 3D human avatar creation and closer to the real world. To explore its potential, we demonstrate the quality and generalizability of avatar creation methods on WildAvatar. We will publicly release our code, data source links and annotations to push forward 3D human avatar creation and other related fields for real-world applications.

LiDAR data pretraining offers a promising approach to leveraging large-scale, readily available datasets for enhanced data utilization. However, existing methods predominantly focus on sparse voxel representation, overlooking the complementary attributes provided by other LiDAR representations. In this work, we propose ${\color{Red}\text{Li}}{\color{Green}\text{MoE}}$, a framework that integrates the Mixture of Experts (MoE) paradigm into LiDAR data representation learning to synergistically combine multiple representations, such as range images, sparse voxels, and raw points. Our approach consists of three stages: i) Image-to-LiDAR Pretraining, which transfers prior knowledge from images to point clouds across different representations; ii) Contrastive Mixture Learning (CML), which uses MoE to adaptively activate relevant attributes from each representation and distills these mixed features into a unified 3D network; iii) Semantic Mixture Supervision (SMS), which combines semantic logits from multiple representations to boost downstream segmentation performance. Extensive experiments across eleven large-scale LiDAR datasets demonstrate our effectiveness and superiority. The code has been made publicly accessible.

The increasing demand for high-quality 3D assets across various industries necessitates efficient and automated 3D content creation. Despite recent advancements in 3D generative models, existing methods still face challenges with optimization speed, geometric fidelity, and the lack of assets for physically based rendering (PBR). In this paper, we introduce 3DTopia-XL, a scalable native 3D generative model designed to overcome these limitations. 3DTopia-XL leverages a novel primitive-based 3D representation, PrimX, which encodes detailed shape, albedo, and material field into a compact tensorial format, facilitating the modeling of high-resolution geometry with PBR assets. On top of the novel representation, we propose a generative framework based on Diffusion Transformer (DiT), which comprises 1) Primitive Patch Compression, 2) and Latent Primitive Diffusion. 3DTopia-XL learns to generate high-quality 3D assets from textual or visual inputs. Extensive qualitative and quantitative evaluations are conducted to demonstrate that 3DTopia-XL significantly outperforms existing methods in generating high-quality 3D assets with fine-grained textures and materials, efficiently bridging the quality gap between generative models and real-world applications.

Texturing is a crucial step in the 3D asset production workflow, which enhances the visual appeal and diversity of 3D assets. Despite recent advancements in Text-to-Texture (T2T) generation, existing methods often yield subpar results, primarily due to local discontinuities, inconsistencies across multiple views, and heavy dependence on UV unwrapping outcomes. To tackle these challenges, we propose a novel generation-refinement 3D texturing framework called MVPaint, which can generate high-resolution, seamless textures while emphasizing multi-view consistency. MVPaint mainly consists of three key modules. 1) Synchronized Multi-view Generation (SMG). Given a 3D mesh model, MVPaint first simultaneously generates multi-view images by employing a SMG model, which leads to coarse texturing results with unpainted parts due to missing observations. 2) Spatial-aware 3D Inpainting (S3I). To ensure complete 3D texturing, we introduce the S3I method, specifically designed to texture previously unobserved areas effectively. 3) UV Refinement (UVR). Furthermore, MVPaint employs a UVR module to improve the texture quality in the UV space, which first performs a UV-space Super-Resolution, followed by a Spatial-aware Seam-Smoothing algorithm for revising spatial texturing discontinuities caused by UV unwrapping. Moreover, we establish two T2T evaluation benchmarks: the Objaverse T2T benchmark and the GSO T2T benchmark, based on selected high-quality 3D meshes from the Objaverse dataset and the entire GSO dataset, respectively. Extensive experimental results demonstrate that MVPaint surpasses existing state-of-the-art methods. Notably, MVPaint could generate high-fidelity textures with minimal Janus issues and highly enhanced cross-view consistency.

Large Language Models (LLMs) demonstrate enhanced capabilities and reliability by reasoning more, evolving from Chain-of-Thought prompting to product-level solutions like OpenAI o1. Despite various efforts to improve LLM reasoning, high-quality long-chain reasoning data and optimized training pipelines still remain inadequately explored in vision-language tasks. In this paper, we present InsightV, an early effort to 1) scalably produce long and robust reasoning data for complex multi-modal tasks, and 2) an effective training pipeline to enhance the reasoning capabilities of multi-modal large language models (MLLMs). Specifically, to create long and structured reasoning data without human labor, we design a two-step pipeline with a progressive strategy to generate sufficiently long and diverse reasoning paths and a multi-granularity assessment method to ensure data quality. We observe that directly supervising MLLMs with such long and complex reasoning data will not yield ideal reasoning ability. To tackle this problem, we design a multi-agent system consisting of a reasoning agent dedicated to performing long-chain reasoning and a summary agent trained to judge and summarize reasoning results. We further incorporate an iterative DPO algorithm to enhance the reasoning agent’s generation stability and quality. Based on the popular LLaVA-NeXT model and our stronger base MLLM, we demonstrate significant performance gains across challenging multi-modal benchmarks requiring visual reasoning. Benefiting from our multi-agent system, Insight-V can also easily maintain or improve performance on perception-focused multi-modal tasks.

Despite the recent progress of audio-driven video generation, existing methods mostly focus on driving facial movements, leading to non-coherent head and body dynamics. Moving forward, it is desirable yet challenging to generate holistic human videos with both accurate lip-sync and delicate co-speech gestures w.r.t. given audio. In this work, we propose AudCast, a generalized audio-driven human video generation framework adopting a cascade Diffusion-Transformers (DiTs) paradigm, which synthesizes holistic human videos based on a reference image and a given audio. 1) Firstly, an audio-conditioned Holistic Human DiT architecture is proposed to directly drive the movements of any human body with vivid gesture dynamics. 2) Then to enhance hand and face details that are well-knownly difficult to handle, a Regional Refinement DiT leverages regional 3D fitting as the bridge to reform the signals, producing the final results. Extensive experiments demonstrate that our framework generates high-fidelity audio-driven holistic human videos with temporal coherence and fine facial and hand details. Resources can be found at https://guanjz20.github.io/projects/AudCast.

Recovering the geometry and materials of objects from a single image is challenging due to its under-constrained nature. In this paper, we present Neural LightRig, a novel framework that boosts intrinsic estimation by leveraging auxiliary multi-lighting conditions from 2D diffusion priors. Specifically, 1) we first leverage illumination priors from large-scale diffusion models to build our multi-light diffusion model on a synthetic relighting dataset with dedicated designs. This diffusion model generates multiple consistent images, each illuminated by point light sources in different directions. 2) By using these varied lighting images to reduce estimation uncertainty, we train a large G-buffer model with a U-Net backbone to accurately predict surface normals and materials. Extensive experiments validate that our approach significantly outperforms state-of-the-art methods, enabling accurate surface normal and PBR material estimation with vivid relighting effects. Code and dataset are available on our project page at https://projects.zxhezexin.com/neural-lightrig.

As wearable devices become more prevalent, understanding the user’s motion is crucial for improving contextual AI systems. We introduce EgoLM, a versatile framework designed for egocentric motion understanding using multi-modal data. EgoLM integrates the rich contextual information from egocentric videos and motion sensors afforded by wearable devices. It also combines dense supervision signals from motion and language, leveraging the vast knowledge encoded in pre-trained large language models (LLMs). EgoLM models the joint distribution of egocentric motions and natural language using LLMs, conditioned on observations from egocentric videos and motion sensors. It unifies a range of motion understanding tasks, including motion narration from video or motion data, as well as motion generation from text or sparse sensor data. Unique to wearable devices, it also enables a novel task to generate text descriptions from sparse sensors. Through extensive experiments, we validate the effectiveness of EgoLM in addressing the challenges of under-constrained egocentric motion learning, and demonstrate its capability as a generalist model through a variety of applications. Project page: https://hongfz16.github.io/projects/EgoLM.

We present Material Anything, a fully-automated, unified diffusion framework designed to generate physically-based materials for 3D objects. Unlike existing methods that rely on complex pipelines or case-specific optimizations, Material Anything offers a robust, end-to-end solution adaptable to objects under diverse lighting conditions. Our approach leverages a pre-trained image diffusion model, enhanced with a triple-head architecture and rendering loss to improve stability and material quality. Additionally, we introduce confidence masks as a dynamic switcher within the diffusion model, enabling it to effectively handle both textured and texture-less objects across varying lighting conditions. By employing a progressive material generation strategy guided by these confidence masks, along with a UV-space material refiner, our method ensures consistent, UV-ready material outputs. Extensive experiments demonstrate our approach outperforms existing methods across a wide range of object categories and lighting conditions.

Human beings are social animals. How to equip 3D autonomous characters with similar social intelligence that can perceive, understand and interact with humans remains an open yet foundamental problem. In this paper, we introduce SOLAMI, the first end-to-end Social vision-Language-Action (VLA) Modeling framework for Immersive interaction with 3D autonomous characters. Specifically, SOLAMI builds 3D autonomous characters from three aspects: 1) Social VLA Architecture: We propose a unified social VLA framework to generate multi-modal response (speech and motion) based on the user’s multimodal input to drive the character for social interaction. 2) Interactive Multimodal Data: We present Syn-MSI, a synthetic multimodal social interaction dataset generated by an automatic pipeline using only existing motion datasets to address the issue of data scarcity. 3) Immersive VR Interface: We develop a VR interface that enables users to immersively interact with these characters driven by various architectures. Extensive quantitative experiments and user study demonstrate that our framework leads to more precise and natural character responses (in both speech and motion) that align with user expectations with lower latency.

Visual Auto-Regressive modeling (VAR) has shown promise in bridging the speed and quality gap between autoregressive image models and diffusion models. VAR reformulates autoregressive modeling by decomposing an image into successive resolution scales. During inference, an image is generated by predicting all the tokens in the next (higher-resolution) scale, conditioned on all tokens in all previous (lower-resolution) scales. However, this formulation suffers from reduced image quality due to the parallel generation of all tokens in a resolution scale; has sequence lengths scaling superlinearly in image resolution; and requires retraining to change the sampling schedule.We introduce Hierarchical Masked AutoRegressive modeling (HMAR), a new image generation algorithm that alleviates these issues using next-scale prediction and masked prediction to generate high-quality images with fast sampling. HMAR reformulates next-scale prediction as a Markovian process, wherein the prediction of each resolution scale is conditioned only on tokens in its immediate predecessor instead of the tokens in all predecessor resolutions. When predicting a resolution scale, HMAR uses a controllable multi-step masked generation procedure to generate a subset of the tokens in each step. On ImageNet 256 × 256 and 512 × 512 benchmarks, HMAR models match or outperform parameter-matched VAR, diffusion, and autoregressive baselines. We develop efficient IO-aware block-sparse attention kernels that allow HMAR to achieve faster training and inference times over VAR by over 2.5× and 1.75× respectively, as well as over 3× lower inference memory foot-print. Finally, HMAR yields additional flexibility over VAR; its sampling schedule can be changed without further training, and it can be applied to image editing tasks in a zero-shot manner.

We present Disco4D, a novel Gaussian Splatting framework for 4D human generation and animation from a single image. Different from existing methods, Disco4D distinctively disentangles clothings (with Gaussian models) from the human body (with SMPL-X model), significantly enhancing the generation details and flexibility. Specifically, 1) Disco4D learns to efficiently fit the clothing Gaussians over the SMPL-X Gaussians. 2) Next, Disco4D adopts diffusion models to enhance the 3D generation process, e.g., modeling occluded parts not visible in the input image. 3) Finally, Disco4D learns an identity encoding for each clothing Gaussian to facilitate the separation and extraction of clothing assets. Furthermore, Disco4D naturally supports 4D human animation with vivid dynamics. Extensive experiments demonstrate the superiority of Disco4D on 4D human generation and animation tasks. Our code is available at https://github.com/disco-4d/Disco4D

3D city generation with NeRF-based methods shows promising generation results but is computationally inefficient. Recently 3D Gaussian splatting (3D-GS) has emerged as a highly efficient alternative for object-level 3D generation. However, adapting 3D-GS from finite-scale 3D objects and humans to infinite-scale 3D cities is non-trivial. Unbounded 3D city generation entails significant storage overhead (out-of-memory issues), arising from the need to expand points to billions, often demanding hundreds of Gigabytes of VRAM for a city scene spanning 10km2. In this paper, we propose GaussianCity, a generative Gaussian splatting framework dedicated to efficiently synthesizing unbounded 3D cities with a single feed-forward pass. Our key insights are two-fold: 1) Compact 3D Scene Representation: We introduce BEV-Point as a highly compact intermediate representation, ensuring that the growth in VRAM usage for unbounded scenes remains constant, thus enabling unbounded city generation. 2) Spatial-aware Gaussian Attribute Decoder: We present spatial-aware BEV-Point decoder to produce 3D Gaussian attributes, which leverages Point Serializer to integrate the structural and contextual characteristics of BEV points. Extensive experiments demonstrate that GaussianCity achieves state-of-the-art results in both drone-view and street-view 3D city generation. Notably, compared to CityDreamer, GaussianCity exhibits superior performance with a speedup of 60 times (10.72 FPS v.s. 0.18 FPS).

We introduce EgoLife, a project to develop an egocentric life assistant that accompanies and enhances personal efficiency through AI-powered wearable glasses. To lay the foundation for this assistant, we conducted a comprehensive data collection study where six participants lived together for one week, continuously recording their daily activities—including discussions, shopping, cooking, social-izing, and entertainment—using AI glasses for multimodal person-view video references. This effort resulted in EgoLife Dataset, a comprehensive 300-hour egocentric, terpersonal, multiview, and multimodal daily life with intensive annotation. Leveraging this dataset, we troduce EgoLifeQA, a suite of long-context, life-oriented question-answering tasks designed to provide meaningful sistance in daily life by addressing practical questions as recalling past relevant events, monitoring health and offering personalized recommendations.To address the key technical challenges of 1) developing robust visual-audio models for egocentric data, 2) enabling identity recognition, and 3) facilitating long-context question answering over extensive temporal information, we introduce EgoBulter, an integrated system comprising EgoGPT and EgoRAG. EgoGPT is an omni-modal model trained on egocentric datasets, achieving state-of-the-art performance on egocentric video understanding. EgoRAG is a retrieval-based component that supports answering ultra-long-context questions. Our experimental studies verify their working mechanisms and reveal critical factors and bottlenecks, guiding future improvements. By releasing our datasets, models, and benchmarks, we aim to stimulate further research in egocentric AI assistants.

We present FreeMorph, the first tuning-free method for image morphing that accommodates inputs with different semantics or layouts. Unlike existing methods that rely on finetuning pre-trained diffusion models and are limited by time constraints and semantic/layout discrepancies, FreeMorph delivers high-fidelity image morphing without requiring per-instance training. Despite their efficiency and potential, tuning-free methods face challenges in maintaining high-quality results due to the non-linear nature of the multi-step denoising process and biases inherited from the pre-trained diffusion model. In this paper, we introduce FreeMorph to address these challenges by integrating two key innovations. 1) We first propose a guidance-aware spherical interpolation design that incorporates explicit guidance from the input images by modifying the self-attention modules, thereby addressing identity loss and ensuring directional transitions throughout the generated sequence. 2) We further introduce a step-oriented variation trend that blends self-attention modules derived from each input image to achieve controlled and consistent transitions that respect both inputs. Our extensive evaluations demonstrate that FreeMorph outperforms existing methods, being 10x ~ 50x faster and establishing a new state-of-the-art for image morphing.

We present Free4D, a novel tuning-free framework for 4D scene generation from a single image. Existing methods either focus on object-level generation, making scene-level generation infeasible, or rely on large-scale multi-view video datasets for expensive training, with limited generalization ability due to the scarcity of 4D scene data. In contrast, our key insight is to distill pre-trained foundation models for consistent 4D scene representation, which offers promising advantages such as efficiency and generalizability. 1) To achieve this, we first animate the input image using image-to-video diffusion models followed by 4D geometric structure initialization. 2) To turn this coarse structure into spatial-temporal consistent multiview videos, we design an adaptive guidance mechanism with a point-guided denoising strategy for spatial consistency and a novel latent replacement strategy for temporal coherence. 3) To lift these generated observations into consistent 4D representation, we propose a modulation-based refinement to mitigate inconsistencies while fully leveraging the generated information. The resulting 4D representation enables real-time, controllable rendering, marking a significant advancement in single-image-based 4D scene generation.

We present DPoser-X, a diffusion-based prior model for 3D whole-body human poses. Building a versatile and robust full-body human pose prior remains challenging due to the inherent complexity of articulated human poses and the scarcity of high-quality whole-body pose datasets. To address these limitations, we introduce a Diffusion model as body Pose prior (DPoser) and extend it to DPoser-X for expressive whole-body human pose modeling. Our approach unifies various pose-centric tasks as inverse problems, solving them through variational diffusion sampling. To enhance performance on downstream applications, we introduce a novel truncated timestep scheduling method specifically designed for pose data characteristics. We also propose a masked training mechanism that effectively combines whole-body and part-specific datasets, enabling our model to capture interdependencies between body parts while avoiding overfitting to specific actions. Extensive experiments demonstrate DPoser-X's robustness and versatility across multiple benchmarks for body, hand, face, and full-body pose modeling. Our model consistently outperforms state-of-the-art alternatives, establishing a new benchmark for whole-body human pose prior modeling.

Multimodal Diffusion Transformers (MM-DiTs) have achieved remarkable progress in text-driven visual generation. However, even state-of-the-art MM-DiT models like FLUX struggle with achieving precise alignment between text prompts and generated content. We identify two key issues in the attention mechanism of MM-DiT, namely 1) the suppression of cross-modal attention due to token imbalance between visual and textual modalities and 2) the lack of timestep-aware attention weighting, which hinder the alignment. To address these issues, we propose \textbf{Temperature-Adjusted Cross-modal Attention (TACA)}, a parameter-efficient method that dynamically rebalances multimodal interactions through temperature scaling and timestep-dependent adjustment. When combined with LoRA fine-tuning, TACA significantly enhances text-image alignment on the T2I-CompBench benchmark with minimal computational overhead. We tested TACA on state-of-the-art models like FLUX and SD3.5, demonstrating its ability to improve image-text alignment in terms of object appearance, attribute binding, and spatial relationships. Our findings highlight the importance of balancing cross-modal attention in improving semantic fidelity in text-to-image diffusion models. Our codes are publicly available at \href{https://github.com/Vchitect/TACA}

Visual diffusion models achieve remarkable progress, yet they are typically trained at limited resolutions due to the lack of high-resolution data and constrained computation resources, hampering their ability to generate high-fidelity images or videos at higher resolutions. Recent efforts have explored tuning-free strategies to exhibit the untapped potential higher-resolution visual generation of pre-trained models. However, these methods are still prone to producing low-quality visual content with repetitive patterns. The key obstacle lies in the inevitable increase in high-frequency information when the model generates visual content exceeding its training resolution, leading to undesirable repetitive patterns deriving from the accumulated errors. To tackle this challenge, we propose FreeScale, a tuning-free inference paradigm to enable higher-resolution visual generation via scale fusion. Specifically, FreeScale processes information from different receptive scales and then fuses it by extracting desired frequency components. Extensive experiments validate the superiority of our paradigm in extending the capabilities of higher-resolution visual generation for both image and video models. Notably, compared with previous best-performing methods, FreeScale unlocks the 8k-resolution text-to-image generation for the first time.

3D Gaussian Splatting (3DGS) is a prevailing technique to reconstruct large-scale 3D scenes from multiview images for novel view synthesis, like a room, a block, and even a city. Such large-scale scenes are not static with changes constantly happening in these scenes, like a new building being built or a new decoration being set up. To keep the reconstructed 3D Gaussian fields up-to-date, a naive way is to reconstruct the whole scene after changing, which is extremely costly and inefficient. In this paper, we propose a new method called GauUpdate that allows partially updating an old 3D Gaussian field with new objects from a new 3D Gaussian field. However, simply inserting the new objects leads to inconsistent appearances because the old and new Gaussian fields may have different lighting environments from each other. GauUpdate addresses this problem by applying inverse rendering techniques in the 3DGS to recover both the materials and environmental lights. Based on the materials and lighting, we relight the new objects in the old 3D Gaussian field for consistent global illumination. For an accurate estimation of the materials and lighting, we put additional constraints on the materials and lighting conditions, that these two fields share the same materials but different environment lights, to improve their qualities. We conduct experiments on both synthetic scenes and real-world scenes to evaluate GauUpdate, which demonstrate that GauUpdate achieves realistic object insertion in 3D Gaussian fields with consistent appearances.

In this paper, we propose VistaDream a novel framework to reconstruct a 3D scene from a single-view image. Recent diffusion models enable generating high-quality novel-view images from a single-view input image. Most existing methods only concentrate on building the consistency between the input image and the generated images while losing the consistency between the generated images. VistaDream addresses this problem by a two-stage pipeline. In the first stage, VistaDream begins with building a global coarse 3D scaffold by zooming out a little step with inpainted boundaries and an estimated depth map. Then, on this global scaffold, we use iterative diffusion-based RGB-D inpainting to generate novel-view images to inpaint the holes of the scaffold. In the second stage, we further enhance the consistency between the generated novel-view images by a novel training-free Multiview Consistency Sampling (MCS) that introduces multi-view consistency constraints in the reverse sampling process of diffusion models. Experimental results demonstrate that without training or fine-tuning existing diffusion models, VistaDream achieves consistent and high-quality novel view synthesis using just single-view images and outperforms baseline methods by a large margin. The code, videos, and interactive demos are available at https://vistadream-project-page.github.io/.

Recent advancements in Vision-Language Models (VLMs) have sparked interest in their use for autonomous driving, particularly in generating interpretable driving decisions through natural language. However, the assumption that VLMs inherently provide visually grounded, reliable, and interpretable explanations for driving remains largely unexamined. To address this gap, we introduce DriveBench, a benchmark dataset designed to evaluate VLM reliability across 17 settings (clean, corrupted, and text-only inputs), encompassing 19,200 frames, 20,498 question-answer pairs, three question types, four mainstream driving tasks, and a total of 12 popular VLMs. Our findings reveal that VLMs often generate plausible responses derived from general knowledge or textual cues rather than true visual grounding, especially under degraded or missing visual inputs. This behavior, concealed by dataset imbalances and insufficient evaluation metrics, poses significant risks in safety-critical scenarios like autonomous driving. We further observe that VLMs struggle with multi-modal reasoning and display heightened sensitivity to input corruptions, leading to inconsistencies in performance. To address these challenges, we propose refined evaluation metrics that prioritize robust visual grounding and multi-modal understanding. Additionally, we highlight the potential of leveraging VLMs' awareness of corruptions to enhance their reliability, offering a roadmap for developing more trustworthy and interpretable decision-making systems in real-world autonomous driving contexts. The benchmark toolkit is publicly accessible.

Recent advances in Large Multimodal Models (LMMs) lead to significant breakthroughs in both academia and industry. One question that arises is how we, as humans, can understand their internal neural representations. This paper takes an initial step towards addressing this question by presenting a versatile framework to identify and interpret the semantics within LMMs. Specifically, 1) we first apply a Sparse Autoencoder(SAE) to disentangle the representations into human understandable features. 2) We then present an automatic interpretation framework to interpreted the open-semantic features learned in SAE by the LMMs themselves. We employ this framework to analyze the LLaVA-NeXT-8B model using the LLaVA-OV-72B model, demonstrating that these features can effectively steer the model's behavior. Our results contribute to a deeper understanding of why LMMs excel in specific tasks, including EQ tests, and illuminate the nature of their mistakes along with potential strategies for their rectification. These findings offer new insights into the internal mechanisms of LMMs and suggest parallels with the cognitive processes of the human brain.

Recent text-to-image diffusion models can produce impressive visuals from textual prompts, but they struggle to reproduce the same subject consistently across multiple generations or contexts. Existing fine-tuning based methods for subject-driven generation face a trade-off between quality and efficiency. Fine-tuning larger models yield higher-quality images but is computationally expensive, while fine-tuning smaller models is more efficient but compromises image quality. To this end, we present Stencil. Stencil resolves this trade-off by leveraging the superior contextual priors of large models and efficient fine-tuning of small models. Stencil uses a small model for fine-tuning while a large pre-trained model provides contextual guidance during inference, injecting rich priors into the generation process with minimal overhead. Stencil excels at generating high-fidelity, novel renditions of the subject in less than a minute, delivering state-of-the-art performance and setting a new benchmark in subject-driven generation. Supplementary materials are available at IEEE SigPort.

Generative 3D modeling has made significant advances recently, but it remains constrained by its inherently ill-posed nature, leading to challenges in quality and controllability. Inspired by the real-world workflow that designers typically refer to existing 3D models when creating new ones, we propose Phidias, a novel generative model that uses diffusion for reference-augmented 3D generation. Given an image, our method leverages a retrieved or user-provided 3D reference model to guide the generation process, thereby enhancing the generation quality, generalization ability, and controllability. Phidias integrates three key components: 1) meta-ControlNet to dynamically modulate the conditioning strength, 2) dynamic reference routing to mitigate misalignment between the input image and 3D reference, and 3) self-reference augmentations to enable self-supervised training with a progressive curriculum. Collectively, these designs result in significant generative improvements over existing methods. Phidias forms a unified framework for 3D generation using text, image, and 3D conditions, offering versatile applications.

Urban scene generation has been developing rapidly recently. However, existing methods primarily focus on generating static and single-frame scenes, overlooking the inherently dynamic nature of real-world driving environments. In this work, we introduce DynamicCity, a novel 4D occupancy generation framework capable of generating large-scale, high-quality dynamic 4D scenes with semantics. DynamicCity mainly consists of two key models. **1)** A VAE model for learning HexPlane as the compact 4D representation. Instead of using naive averaging operations, DynamicCity employs a novel **Projection Module** to effectively compress 4D features into six 2D feature maps for HexPlane construction, which significantly enhances HexPlane fitting quality (up to **12.56** mIoU gain). Furthermore, we utilize an **Expansion & Squeeze Strategy** to reconstruct 3D feature volumes in parallel, which improves both network training efficiency and reconstruction accuracy than naively querying each 3D point (up to **7.05** mIoU gain, **2.06x** training speedup, and **70.84\%** memory reduction). **2)** A DiT-based diffusion model for HexPlane generation. To make HexPlane feasible for DiT generation, a **Padded Rollout Operation** is proposed to reorganize all six feature planes of the HexPlane as a squared 2D feature map. In particular, various conditions could be introduced in the diffusion or sampling process, supporting **versatile 4D generation applications**, such as trajectory- and command-driven generation, inpainting, and layout-conditioned generation. Extensive experiments on the CarlaSC and Waymo datasets demonstrate that DynamicCity significantly outperforms existing state-of-the-art 4D occupancy generation methods across multiple metrics. The code and models have been released to facilitate future research.

Recent advancements in diffusion models have led to significant improvements in the generation and animation of 4D full-body human-object interactions (HOI). Nevertheless, existing methods primarily focus on SMPL-based motion generation, which is limited by the scarcity of realistic large-scale interaction data. This constraint affects their ability to create everyday HOI scenes. This paper addresses this challenge using a zero-shot approach with a pre-trained diffusion model. Despite this potential, achieving our goals is difficult due to the diffusion model's lack of understanding of ''where'' and ''how'' objects interact with the human body. To tackle these issues, we introduce **AvatarGO**, a novel framework designed to generate animatable 4D HOI scenes directly from textual inputs. Specifically, **1)** for the ''where'' challenge, we propose **LLM-guided contact retargeting**, which employs Lang-SAM to identify the contact body part from text prompts, ensuring precise representation of human-object spatial relations. **2)** For the ''how'' challenge, we introduce **correspondence-aware motion optimization** that constructs motion fields for both human and object models using the linear blend skinning function from SMPL-X. Our framework not only generates coherent compositional motions, but also exhibits greater robustness in handling penetration issues. Extensive experiments with existing methods validate AvatarGO's superior generation and animation capabilities on a variety of human-object pairs and diverse poses. As the first attempt to synthesize 4D avatars with object interactions, we hope AvatarGO could open new doors for human-centric 4D content creation.

Visual data comes in various forms, ranging from small icons of just a few pixels to long videos spanning hours. Existing multi-modal LLMs usually standardize these diverse visual inputs to fixed-resolution images or patches for visual encoders and yield similar numbers of tokens for LLMs. This approach is non-optimal for multimodal understanding and inefficient for processing inputs with long and short visual contents. To solve the problem, we propose Oryx, a unified multimodal architecture for the spatial-temporal understanding of images, videos, and multi-view 3D scenes. Oryx offers an on-demand solution to seamlessly and efficiently process visual inputs with arbitrary spatial sizes and temporal lengths through two core innovations: 1) a pre-trained OryxViT model that can encode images at any resolution into LLM-friendly visual representations; 2) a dynamic compressor module that supports 1x to 16x compression on visual tokens by request. These designs enable Oryx to accommodate extremely long visual contexts, such as videos, with lower resolution and high compression while maintaining high recognition precision for tasks like document understanding with native resolution and no compression. Beyond the architectural improvements, enhanced data curation and specialized training on long-context retrieval and spatial-aware data help Oryx achieve strong capabilities in image, video, and 3D multimodal understanding simultaneously.

In this paper, we present \textbf{\textit{FasterCache}}, a novel training-free strategy designed to accelerate the inference of video diffusion models with high-quality generation. By analyzing existing cache-based methods, we observe that \textit{directly reusing adjacent-step features degrades video quality due to the loss of subtle variations}. We further perform a pioneering investigation of the acceleration potential of classifier-free guidance (CFG) and reveal significant redundancy between conditional and unconditional features within the same timestep. Capitalizing on these observations, we introduce FasterCache to substantially accelerate diffusion-based video generation. Our key contributions include a dynamic feature reuse strategy that preserves both feature distinction and temporal continuity, and CFG-Cache which optimizes the reuse of conditional and unconditional outputs to further enhance inference speed without compromising video quality. We empirically evaluate FasterCache on recent video diffusion models. Experimental results show that FasterCache can significantly accelerate video generation (\eg 1.67$\times$ speedup on Vchitect-2.0) while keeping video quality comparable to the baseline, and consistently outperform existing methods in both inference speed and video quality. \textit{Our code will be made public upon publication.}

Large multimodal models excel in multimodal tasks but face significant computational challenges due to excessive visual tokens. Unlike token reduction methods that focus on token-level redundancy, we identify and study the computation-level redundancy on vision tokens to ensure no information loss. Our key insight is that vision tokens from the pretrained vision encoder do not necessarily require all the heavy operations (e.g., self-attention, FFNs) in decoder-only LMMs and could be processed more lightly with proper designs. We designed a series of experiments to discover and progressively squeeze out the vision-related computation redundancy. Based on our findings, we propose ProxyV, a novel approach that utilizes proxy vision tokens to alleviate the computational burden on original vision tokens. ProxyV enhances efficiency without compromising performance and can even yield notable performance gains in scenarios with more moderate efficiency improvements. Furthermore, the flexibility of ProxyV is demonstrated through its combination with token reduction methods to boost efficiency further.

Existing LLM-based motion models fail to fully leverage large models' planning capabilities for motion-related tasks, exhibiting poor generalization, limited text-motion alignment, and an inability to perform multimodal condition joint driven motion generation. We argue that these issues arise from the modality gap and the highly coupled nature of motion tokens. To address this, we proposed the hybrid motion sentence, which is consistant of fine-grained motion decription and atomic body-part motion token that can bridge the gap between motion and text. To obtain a large corpus of hybrid motion sentences, we introduced a novel motion-to-text generation method that combines atomic motion operators with GPT-4o, resulting in 68.2 million fine-grained textual descriptions across diverse modalities. To reconstruct high-quality motion from hybrid sentences and make better motion-text alignment, we introduce Semantic-Aware Decoupled Motion Tokenization. Furthermore, we propose MotionUPG based on LLaMA, leveraging MotionWords dataset for both pretraining and instruction tuning. Our method achieves strong fine-grained text-motion alignment, impressive zero-shot motion generation, and is the first to support multimodal condition joint driven motion generation tasks.

3D modeling is moving from virtual to physical. Existing 3D generation primarily emphasizes geometries and textures while neglecting physical-grounded modeling. Consequently, despite the rapid development of 3D generative models, the synthesized 3D assets often overlook rich and important physical properties, hampering their real-world application in physical domains like simulation and embodied AI. As an initial attempt to address this challenge, we propose \textbf{PhysX}, an end-to-end paradigm for physical-grounded 3D asset generation. \textbf{1)} To bridge the critical gap in physics-annotated 3D datasets, we present \textbf{\ourname}\ - the first physics-grounded 3D dataset systematically annotated across five foundational dimensions: \textbf{\textcolor{color2}{absolute scale}}, \textbf{\textcolor{color3}{material}}, \textbf{\textcolor{color1}{affordance}}, \textbf{\textcolor{color4}{kinematics}}, and \textbf{\textcolor{color5}{function description}}. In particular, we devise a scalable human-in-the-loop annotation pipeline based on vision-language models, which enables efficient creation of physics-first assets from raw 3D assets. \textbf{2)} Furthermore, we propose \textbf{PhysXGen}, a feed-forward framework for physics-grounded 3D asset generation, injecting physical knowledge into the pre-trained 3D structural space. Specifically, PhysXGen employs a dual-branch architecture to explicitly model the latent correlations between 3D structures and physical properties, thereby producing 3D assets with plausible physical predictions while preserving the native geometry quality. Extensive experiments validate the superior performance and promising generalization capability of our framework. All the code, data, and models will be released to facilitate future research in generative physical AI.

Recent advancements have explored text-to-image diffusion models for synthesizing out-of-distribution (OOD) samples, substantially enhancing the performance of OOD detection. However, existing approaches typically rely on perturbing text-conditioned embeddings, resulting in semantic instability and insufficient shift diversity, which limit generalization to realistic OOD. To address these challenges, we propose GOOD, a novel and flexible framework that directly guides diffusion sampling trajectories towards OOD regions using off-the-shelf in-distribution (ID) classifiers. GOOD incorporates dual-level guidance: (1) Image-level guidance based on the gradient of log partition to reduce input likelihood, drives samples toward low-density regions in pixel space. (2) Feature-level guidance, derived from k-NN distance in the classifier’s latent space, promotes sampling in feature-sparse regions. Hence, this dual-guidance design enables more controllable and diverse OOD sample generation. Additionally, we introduce a unified OOD score that adaptively combines image and feature discrepancies, enhancing detection robustness. We perform thorough quantitative and qualitative analyses to evaluate the effectiveness of GOOD, demonstrating that training with samples generated by GOOD can notably enhance OOD detection performance.

Visual grounding in 3D is the key for embodied agents to localize language-referred objects in open-world environments. However, existing benchmarks are limited to indoor focus, single-platform constraints, and small scale. We introduce 3EED, a multi-platform, multi-modal 3D grounding benchmark featuring RGB and LiDAR data from vehicle, drone, and quadruped platforms. We provide over 128,000 objects and 22,000 validated referring expressions across diverse outdoor scenes -- 10x larger than existing datasets. We develop a scalable annotation pipeline combining vision-language model prompting with human verification to ensure high-quality spatial grounding. To support cross-platform learning, we propose platform-aware normalization and cross-modal alignment techniques, and establish benchmark protocols for in-domain and cross-platform evaluations. Our findings reveal significant performance gaps, highlighting the challenges and opportunities of generalizable 3D grounding. The 3EED dataset and benchmark toolkit are released to advance future research in language-driven 3D embodied perception.

Recent advancements in static feed-forward scene reconstruction have demonstrated significant progress in high-quality novel view synthesis. However, these models often struggle with generalizability across diverse environments and fail to effectively handle dynamic content. We present BTimer (short for Bullet Timer), the first motion-aware feed-forward model for real-time reconstruction and novel view synthesis of dynamic scenes. Our approach reconstructs the full scene in a 3D Gaussian Splatting representation at a given target (‘bullet’) timestamp by aggregating information from all the context frames. Such a formulation allows BTimer to gain scalability and generalization by leveraging both static and dynamic scene datasets. Given a casual monocular dynamic video, BTimer reconstructs a bullet-time scene within 150ms while reaching state-of-the-art performance on both static and dynamic scene datasets, even compared with optimization-based approaches.

Recent Vision-Language Models (VLMs) have shown strong performance in general-purpose visual understanding and reasoning, but their ability to comprehend the visual grammar of movie shots remains underexplored and insufficiently evaluated. To bridge this gap, we present \textbf{ShotBench}, a dedicated benchmark for assessing VLMs’ understanding of cinematic language. ShotBench includes 3,049 still images and 500 video clips drawn from more than 200 films, with each sample annotated by trained annotators or curated from professional cinematography resources, resulting in 3,608 high-quality question-answer pairs. We conduct a comprehensive evaluation of over 20 state-of-the-art VLMs across eight core cinematography dimensions. Our analysis reveals clear limitations in fine-grained perception and cinematic reasoning of current VLMs. To improve VLMs capability in cinematography understanding, we construct a large-scale multimodal dataset, named ShotQA, which contains about 70k Question-Answer pairs derived from movie shots. Besides, we propose ShotVL and train this VLM model with a two-stage training strategy, integrating both supervised fine-tuning and Group Relative Policy Optimization (GRPO). Experimental results demonstrate that our model achieves substantial improvements, surpassing all existing strongest open-source and proprietary models evaluated on ShotBench, establishing a new state-of-the-art performance.

$360^\circ$ videos offer a hyper-immersive experience that allows the viewers to explore a dynamic scene from full 360 degrees. To achieve more accessible and personalized content creation in $360^\circ$ video format, we seek to lift standard perspective videos into $360^\circ$ equirectangular videos. To this end, we introduce **Imagine360**, the first perspective-to-$360^\circ$ video generation framework that creates high-quality $360^\circ$ videos with rich and diverse motion patterns from video anchors. Imagine360 learns fine-grained spherical visual and motion patterns from limited $360^\circ$ video data with several key designs. **1)** Firstly we adopt the dual-branch design, including a perspective and a panorama video denoising branch to provide local and global constraints for $360^\circ$ video generation, with motion module and spatial LoRA layers fine-tuned on $360^\circ$ videos. **2)** Additionally, an antipodal mask is devised to capture long-range motion dependencies, enhancing the reversed camera motion between antipodal pixels across hemispheres. **3)** To handle diverse perspective video inputs, we propose rotation-aware designs that adapt to varying video masking due to changing camera poses across frames. **4)** Lastly, we introduce a new 360 video dataset featuring 10K high-quality, trimmed 360 video clips with structured motion to facilitate training. Extensive experiments show Imagine360 achieves superior graphics quality and motion coherence with our curated dataset among state-of-the-art $360^\circ$ video generation methods. We believe Imagine360 holds promise for advancing personalized, immersive $360^\circ$ video creation.

Given a single in-the-wild human photo, it remains a challenging task to reconstruct a high-fidelity 3D human model. Existing methods face difficulties including a) the varying body proportions captured by in-the-wild human images; b) diverse personal belongings within the shot; and c) ambiguities in human postures and inconsistency in human textures. In addition, the scarcity of high-quality human data intensifies the challenge. To address these problems, we propose a Generalizable image-to-3D huMAN reconstruction framework, dubbed GeneMAN, building upon a comprehensive multi-source collection of high-quality human data, including 3D scans, multi-view videos, single photos, and our generated synthetic human data. GeneMAN encompasses three key modules. 1) Without relying on parametric human models (e.g., SMPL), GeneMAN first trains a human-specific text-to-image diffusion model and a view-conditioned diffusion model, serving as GeneMAN 2D human prior and 3D human prior for reconstruction, respectively. 2) With the help of the pretrained human prior models, the Geometry Initialization-&-Sculpting pipeline is leveraged to recover high-quality 3D human geometry given a single image. 3) To achieve high-fidelity 3D human textures, GeneMAN employs the Multi-Space Texture Refinement pipeline, consecutively refining textures in the latent and the pixel spaces. Extensive experimental results demonstrate that GeneMAN could generate high-quality 3D human models from a single image input, outperforming prior state-of-the-art methods. Notably, GeneMAN could reveal much better generalizability in dealing with in-the-wild images, often yielding high-quality 3D human models in natural poses with common items, regardless of the body proportions in the input images.

Emerging world models autoregressively generate video frames in response to actions, such as camera movements and text prompts, among other control signals. Due to limited temporal context window sizes, these models often struggle to maintain scene consistency during revisits, leading to severe forgetting of previously generated environments. Inspired by the mechanisms of human memory, we introduce a novel framework to enhancing long-term consistency of video world models through a geometry-grounded long-term spatial memory. Our framework includes mechanisms to store and retrieve information from the long-term spatial memory and we curate custom datasets to train and evaluate world models with explicitly stored 3D memory mechanisms. Our evaluations show improved quality, consistency, and context length compared to relevant baselines, paving the way towards long-term consistent world generation.

Despite rapid advances in 3D content generation, quality assessment for the generated 3D assets remains challenging.Existing methods mainly rely on image-based metrics and operate solely at the object level, limiting their ability to capture spatial Despite rapid advances in 3D content generation, quality assessment for the generated 3D assets remains challenging.Existing methods mainly rely on image-based metrics and operate solely at the object level, limiting their ability to capture spatial coherence, material authenticity, and high-fidelity local details.1) To address these challenges, we introduce Hi3DEval, a hierarchical evaluation framework tailored for 3D generative content. It combines both object-level and part-level evaluation, enabling holistic assessments across multiple dimensions as well as fine-grained quality analysis. Additionally, we extend texture evaluation beyond aesthetic appearance by explicitly assessing material realism, focusing on attributes such as albedo, saturation, and metallicness. 2) To support this framework, we construct Hi3DBench, a large-scale dataset comprising diverse 3D assets and high-quality annotations, accompanied by a reliable multi-agent annotation pipeline.We further propose a 3D-aware automated scoring system based on hybrid 3D representations. Specifically, we leverage video-based representations for object-level and material-subject evaluations to enhance modeling of spatio-temporal consistency and employ pretrained 3D features for part-level perception.Extensive experiments demonstrate that our approach outperforms existing image-based metrics in modeling 3D characteristics and achieves superior alignment with human preference, providing a scalable alternative to manual evaluations.

Recent studies have shown that agent-based systems leveraging large language models (LLMs) for key information retrieval and integration have emerged as a promising approach for long video understanding. However, these systems face two major challenges. First, they typically perform modeling and reasoning on individual frames, struggling to capture the temporal context of consecutive frames. Second, to reduce the cost of dense frame-level captioning, they adopt sparse frame sampling, which risks discarding crucial information. To overcome these limitations, we propose VideoLucy, a deep memory backtracking framework for long video understanding. Inspired by the human recollection process from coarse to fine, VideoLucy employs a hierarchical memory structure with progressive granularity. This structure explicitly defines the detail level and temporal scope of memory at different hierarchical depths. Through an agent-based iterative backtracking mechanism, VideoLucy systematically mines video-wide, question-relevant deep memories until sufficient information is gathered to provide a confident answer. This design enables effective temporal understanding of consecutive frames while preserving critical details. In addition, we introduce EgoMem, a new benchmark for long video understanding. EgoMem is designed to comprehensively evaluate a model's ability to understand complex events that unfold over time and capture fine-grained details in extremely long videos. Extensive experiments demonstrate the superiority of VideoLucy. Built on open-source models, VideoLucy significantly outperforms state-of-the-art methods on multiple long video understanding benchmarks, achieving performance even surpassing the latest proprietary models such as GPT-4o. Our code and dataset will be made publicly available.

Effective multi-shot generation demands purposeful, film-like transitions and strict cinematic continuity. Current methods, however, often prioritize basic visual consistency, neglecting crucial editing patterns (e.g., shot/reverse shot, cutaways) that drive narrative flow for compelling storytelling. This yields outputs that may be visually coherent but lack narrative sophistication and true cinematic integrity. To bridge this, we introduce Next Shot Generation (NSG): synthesizing a subsequent, high-quality shot that critically conforms to professional editing patterns while upholding rigorous cinematic continuity. Our framework, Cut2Next, leverages a Diffusion Transformer (DiT). It employs in-context tuning guided by a novel Hierarchical Multi-Prompting strategy. This strategy uses Relational Prompts to define overall context and inter-shot editing styles. Individual Prompts then specify per-shot content and cinematographic attributes. Together, these guide Cut2Next to generate cinematically appropriate next shots. Architectural innovations, Context-Aware Condition Injection (CACI) and Hierarchical Attention Mask (HAM), further integrate these diverse signals without introducing new parameters. We construct RawCuts (large-scale) and CuratedCuts (refined) datasets, both with hierarchical prompts, and introduce CutBench for evaluation. Experiments show Cut2Next excels in visual consistency and text fidelity. Crucially, user studies reveal a strong preference for Cut2Next, particularly for its adherence to intended editing patterns and overall cinematic continuity, validating its ability to generate high-quality, narratively expressive, and cinematically coherent subsequent shots.

Safety-critical 3D scene understanding tasks necessitate not only accurate but also confident predictions from 3D perception models. This study introduces Calib3D, a pioneering effort to benchmark and scrutinize the reliability of 3D scene understanding models from an uncertainty estimation viewpoint. We comprehensively evaluate 28 state of the art models across 10 diverse 3D datasets, uncovering insightful phenomena that cope with both the aleatoric and epistemic uncertainties in 3D scene understanding. We discover that despite achieving impressive levels of accuracy, existing models frequently fail to provide reliable uncertainty estimates - a pitfall that critically undermines their applicability in safety-sensitive contexts. Through extensive analysis of key factors such as network capacity, LiDAR representations, rasterization resolutions, and 3D data augmentation techniques, we correlate these aspects directly with the model calibration efficacy. Furthermore, we introduce DeptS, a novel depth-aware scaling approach aimed at enhancing 3D model calibration. Extensive experiments across a wide range of configurations validate the superiority of our method. We hope this work could serve as a cornerstone for fostering reliable 3D scene understanding. Code and benchmark toolkit are publicly available11https://github.com/ldkong1205/Calib3D.

Generating diverse and high-quality 3D assets automatically poses a fundamental yet challenging task in 3D computer vision. Despite extensive efforts in 3D generation, existing optimization-based approaches struggle to produce large-scale 3D assets efficiently. Meanwhile, feed-forward methods often focus on generating only a single category or a few categories, limiting their generalizability. Therefore, we introduce a diffusion-based feed-forward framework to address these challenges with a single model. To handle the large diversity and complexity in geometry and texture across categories efficiently, we 1) adopt improved triplane to guarantee efficiency; 2) introduce the 3D-aware transformer to aggregate the generalized 3D knowledge with specialized 3D features; and 3) devise the 3D-aware encoder/decoder to enhance the generalized 3D knowledge. Building upon our 3D-aware Diffusion model with TransFormer, DiffTF, we propose a stronger version for 3D generation, i.e., DiffTF++. It boils down to two parts: multi-view reconstruction loss and triplane refinement. Specifically, we utilize multi-view reconstruction loss to fine-tune the diffusion model and triplane decoder, thereby avoiding the negative influence caused by reconstruction errors and improving texture synthesis. By eliminating the mismatch between the two stages, the generative performance is enhanced, especially in texture. Additionally, a 3D-aware refinement process is introduced to filter out artifacts and refine triplanes, resulting in the generation of more intricate and reasonable details. Extensive experiments on ShapeNet and OmniObject3D convincingly demonstrate the effectiveness of our proposed modules and the state-of-the-art 3D object generation performance with large diversity, rich semantics, and high quality.

Efficient data utilization is crucial for advancing 3D scene understanding in autonomous driving, where reliance on heavily human-annotated LiDAR point clouds challenges fully supervised methods. Addressing this, our study extends into semi-supervised learning for LiDAR semantic segmentation, leveraging the intrinsic spatial priors of driving scenes and multi-sensor complements to augment the efficacy of unlabeled datasets. We introduce LaserMix++, an evolved framework that integrates laser beam manipulations from disparate LiDAR scans and incorporates LiDAR-camera correspondences to further assist data-efficient learning. Our framework is tailored to enhance 3D scene consistency regularization by incorporating multi-modality, including 1) multi-modal LaserMix operation for fine-grained cross-sensor interactions; 2) camera-to-LiDAR feature distillation that enhances LiDAR feature learning; and 3) language-driven knowledge guidance generating auxiliary supervisions using open-vocabulary models. The versatility of LaserMix++ enables applications across LiDAR representations, establishing it as a universally applicable solution. Our framework is rigorously validated through theoretical analysis and extensive experiments on popular driving perception datasets. Results demonstrate that LaserMix++ markedly outperforms fully supervised alternatives, achieving comparable accuracy with five times fewer annotations and significantly improving the supervised-only baselines. This substantial advancement underscores the potential of semi-supervised approaches in reducing the reliance on extensive labeled data in LiDAR-based 3D scene understanding systems.

Advancements in large pre-trained generative models have expanded their potential as effective data generators in visual recognition. This work delves into the impact of generative images, primarily comparing paradigms that harness external data (i.e. generative vs. retrieval vs. original). Our key contributions are: 1) GenBench Construction: We devise GenBench, a broad benchmark comprising 22 datasets with 2548 categories, to appraise generative data across various visual recognition tasks. 2) CLER Score: To address the insufficient correlation of existing metrics (e.g., FID, CLIP score) with downstream recognition performance, we propose CLER, a training-free metric indicating generative data's efficiency for recognition tasks prior to training. 3) New Baselines: Comparisons of generative data with retrieved data from the same external pool help to elucidate the unique traits of generative data. 4) External Knowledge Injection: By fine-tuning special token embeddings for each category via Textual Inversion, performance improves across 17 datasets, except when dealing with low-resolution reference images. Our exhaustive benchmark and analysis spotlight generative data's promise in visual recognition, while identifying key challenges for future investigation.

Recent advances in Large Multimodal Models (LMMs) have unveiled great potential as visual assistants. However, most existing works focus on responding to individual instructions or using previous dialogues for contextual understanding. There is little discussion on employing both images and text as in-context examples to enhance the instruction following capability. To bridge this gap, we introduce the Otter model to leverage both textual and visual in-context examples for instruction tuning. Specifically, Otter builds upon Flamingo with Perceiver architecture, and has been instruction tuned for general purpose multi-modal assistant. Otter seamlessly processes multi-modal inputs, supporting modalities including text, multiple images, and dynamic video content. To support the training of Otter, we present the MIMIC-IT (MultI-Modal In-Context Instruction Tuning) dataset, which encompasses over 3 million multi-modal instruction-response pairs, including approximately 2.2 million unique instructions across a broad spectrum of images and videos. MIMIC-IT has been carefully curated to feature a diverse array of in-context examples for each entry. Comprehensive evaluations suggest that instruction tuning with these in-context examples substantially enhances model convergence and generalization capabilities. Notably, the extensive scenario coverage provided by the MIMIC-IT dataset empowers the Otter model to excel in tasks involving complex video and multi-image understanding.

Recent advancements in bird's eye view (BEV) representations have shown remarkable promise for in-vehicle 3D perception. However, while these methods have achieved impressive results on standard benchmarks, their robustness in varied conditions remains insufficiently assessed. In this study, we present RoboBEV, an extensive benchmark suite designed to evaluate the resilience of BEV algorithms. This suite incorporates a diverse set of camera corruption types, each examined over three severity levels. Our benchmarks also consider the impact of complete sensor failures that occur when using multi-modal models. Through RoboBEV, we assess 33 state-of-the-art BEV-based perception models spanning tasks like detection, map segmentation, depth estimation, and occupancy prediction. Our analyses reveal a noticeable correlation between the model's performance on in-distribution datasets and its resilience to out-of-distribution challenges. Our experimental results also underline the efficacy of strategies like pre-training and depth-free BEV transformations in enhancing robustness against out-of-distribution data. Furthermore, we observe that leveraging extensive temporal information significantly improves the model's robustness. Based on our observations, we design an effective robustness enhancement strategy based on the CLIP model. The insights from this study pave the way for the development of future BEV models that seamlessly combine accuracy with real-world robustness.

The size of vision models has grown exponentially over the last few years, especially after the emergence of Vision Transformer. This has motivated the development of parameter-efficient tuning methods, such as learning adapter layers or visual prompt tokens, which allow a tiny portion of model parameters to be trained whereas the vast majority obtained from pre-training are frozen. However, designing a proper tuning method is non-trivial: one might need to try out a lengthy list of design choices, not to mention that each downstream dataset often requires custom designs. In this paper, we view the existing parameter-efficient tuning methods as “prompt modules” and propose Neural prOmpt seArcH (NOAH), a novel approach that learns, for large vision models, the optimal design of prompt modules through a neural architecture search algorithm, specifically for each downstream dataset. By conducting extensive experiments on over 20 vision datasets, we demonstrate that NOAH (i) is superior to individual prompt modules, (ii) has good few-shot learning ability, and (iii) is domain-generalizable.

Class-Incremental Learning (CIL) or continual learning is a desired capability in the real world, which requires a learning system to adapt to new tasks without forgetting former ones. While traditional CIL methods focus on visual information to grasp core features, recent advances in Vision-Language Models (VLM) have shown promising capabilities in learning generalizable representations with the aid of textual information. However, when continually trained with new classes, VLMs often suffer from catastrophic forgetting of former knowledge. Applying VLMs to CIL poses two major challenges: 1) how to adapt the model without forgetting and 2) how to make full use of the multi-modal information. To this end, we propose PROjectiOn Fusion (Proof) that enables VLMs to learn without forgetting. To handle the first challenge, we propose training task-specific projections based on the frozen image/text encoders. When facing new tasks, new projections are expanded, and former projections are fixed, alleviating the forgetting of old concepts. For the second challenge, we propose the fusion module to better utilize the cross-modality information. By jointly adjusting visual and textual features, the model can capture better task-specific semantic information that facilitates recognition. Extensive experiments on nine benchmark datasets with various continual learning scenarios and various VLMs validate that Proof achieves state-of-the-art performance.

As a main approach towards touch-free human-computer interaction,hand gesture recognition(HGR) has long been a research focus for both academia and industry. Meanwhile,visible light communication(VLC) has become increasingly popular with VLC-ready commercial products (e.g., Philips lamps) available on the market. These facts provoke us to ask: can we leverage a VLC-ready lamp to realizeintegrated sensing and communication(ISAC) by conducting both HGR and VLC simultaneously? To this end, we propose ReflexGest as our answer to this question. ReflexGest is implemented upon a table lamp for the sake of practicality; this VLC-ready lamp is equipped with a ring-shaped light-emitting diode (LED) array and a photodiode (PD, for light intensity sensing) originally aiming for up/down-link VLCs. Demanding hand gestures to be performed between the lamp and a table surface, ReflexGest exploits the variation of the reflection and their unique correlation with the corresponding hand gestures to achieve HGR. In particular, ReflexGest first handles the limited sensing ability of the PD by enhancing the LED lamp and thus diversifying the light emission patterns. Moreover, ReflexGest combats the reflection interference from varying table surfaces via an adversarial learning technique to distill only the features relevant to hand gestures. Our extensive evaluations demonstrate that ReflexGest is able to deliver accurate HGR under realistic VLC traffic.

Text-driven video generation witnesses rapid progress. However, merely using text prompts is not enough to depict the desired subject appearance that accurately aligns with users' intents, especially for customized content creation. In this paper, we study the task of video generation with image prompts, which provide more accurate and direct content control beyond the text prompts. Specifically, we propose a feed-forward framework VideoBooth, with two dedicated designs: 1) We propose to embed image prompts in a coarse-to-fine manner. Coarse visual embeddings from image encoder provide high-level encodings of image prompts, while fine visual embeddings from the proposed attention injection module provide multi-scale and detailed encoding of image prompts. These two complementary embeddings can faithfully capture the desired appearance. 2) In the attention injection module at fine level, multi-scale image prompts are fed into different cross-frame attention layers as additional keys and values. This extra spatial in-formation refines the details in the first frame and then it is propagated to the remaining frames, which maintains temporal consistency. Extensive experiments demonstrate that Video Booth achieves state-of-the-art performance in gener-ating customized high-quality videos with subjects specified in image prompts. Notably, VideoBooth is a generalizable framework where a single model works for a wide range of image prompts with only feed-forward passes.

Existing photorealistic relightable hand models require extensive identity-specific observations in different views, poses, and illuminations, and face challenges in generalizing to natural illuminations and novel identities. To bridge this gap, we present URHand, the first universal relightable hand model that generalizes across viewpoints, poses, illuminations, and identities. Our model allows few-shot personalization using images captured with a mobile phone, and is ready to be photorealistically rendered under novel illuminations. To simplify the personalization process while retaining photorealism, we build a powerful universal relightable prior based on neural relighting from multi-view images of hands captured in a light stage with hundreds of identities. The key challenge is scaling the cross-identity training while maintaining personalized fidelity and sharp details without compromising generalization under natural illuminations. To this end, we propose a spatially varying linear lighting model as the neural renderer that takes physics-inspired shading as input feature. By removing non-linear activations and bias, our specifically designed lighting model explicitly keeps the linearity of light transport. This enables single-stage training from light-stage data while generalizing to real-time rendering under arbitrary continuous illuminations across diverse identities. In addition, we introduce the joint learning of a physically based model and our neural relighting model, which further improves fidelity and generalization. Extensive experiments show that our approach achieves superior performance over existing methods in terms of both quality and generalizability. We also demonstrate quick personalization of URHand from a short phone scan of an unseen identity.

In this work, we present Digital Life Project, a framework utilizing language as the universal medium to build autonomous 3D characters, who are capable of engaging in social interactions and expressing with articulated body motions, thereby simulating life in a digital environment. Our framework comprises two primary components: 1) SocioMind: a meticulously crafted digital brain that models personalities with systematic few-shot exemplars, incorporates a reflection process based on psychology principles, and emulates autonomy by initiating dialogue topics; 2) MoMat-MoGen: a text-driven motion synthesis paradigm for controlling the character's digital body. It integrates motion matching, a proven industry technique to ensure motion quality, with cutting-edge advancements in motion generation for diversity. Extensive experiments demonstrate that each module achieves state-of-the-art performance in its respective domain. Collectively, they enable virtual characters to initiate and sustain dialogues autonomously, while evolving their socio-psychological states. Concurrently, these characters can perform contextually relevant bodily movements. Additionally, an extension of DLP enables a virtual character to recognize and appropriately respond to human players' actions.

Dynamic human rendering from video sequences has achieved remarkable progress by formulating the rendering as a mapping from static poses to human images. However, existing methods focus on the human appearance reconstruction of every single frame while the temporal motion relations are not fully explored. In this paper, we propose a new 4D motion modeling paradigm, SurMo, that jointly models the temporal dynamics and human appearances in a unified framework with three key designs: 1) Surface-based motion encoding that models 4D human motions with an efficient compact surface-based triplane. It encodes both spatial and temporal motion relations on the dense surface manifold of a statistical body template, which inherits body topology priors for generalizable novel view synthesis with sparse training observations. 2) Physical motion decoding that is designed to encourage physical motion learning by decoding the motion triplane features at timestep t to pre-dict both spatial derivatives and temporal derivatives at the next timestep t + 1 in the training stage. 3) 4D appearance decoding that renders the motion trip lanes into images by an efficient volumetric surface-conditioned renderer that fo-cuses on the rendering of body surfaces with motion learning conditioning. Extensive experiments validate the state-of-the-art performance of our new paradigm and illus-trate the expressiveness of surface-based motion triplanes for rendering high-fidelity view-consistent humans with fast motions and even motion-dependent shadows. Our project page is at: https://taohuumd.github.io/projects/SurMo.

We present, GauHuman, a 3D human model with Gaussian Splatting for both fast training minutes) and real-time rendering (up to 189 FPS), compared with existing NeRF-based implicit representation modelling frameworks demanding hours of training and seconds of rendering per frame. Specifically, GauHuman encodes Gaussian Splatting in the canonical space and transforms 3D Gaussians from canonical space to posed space with linear blend skinning (LBS), in which effective pose and LBS refinement modules are designed to learn fine details of 3D humans under negligible computational cost. Moreover, to enable fast optimization of GauHuman, we initialize and prune 3D Gaussians with 3D human prior, while splitting/cloning via KL divergence guidance, along with a novel merge operation for further speeding up. Extensive experiments on ZJU_Mocap and MonoCap datasets demonstrate that GauHuman achieves state-of-the-art performance quantitatively and qualitatively with fast training and real-time rendering speed. Notably, without sacrificing rendering quality, GauHuman can fast model the 3D human performer with 3D Gaussians. Our code is available at https://github.com/skhu101/GauHuman.

Video generation has witnessed significant advance-ments, yet evaluating these models remains a challenge. A comprehensive evaluation benchmark for video generation is indispensable for two reasons: 1) Existing metrics do not fully align with human perceptions; 2) An ideal eval-uation system should provide insights to inform future de-velopments of video generation. To this end, we present VBench, a comprehensive benchmark suite that dissects “video generation quality” into specific, hierarchical, and disentangled dimensions, each with tailored prompts and evaluation methods. VBench has three appealing proper-ties: 1) Comprehensive Dimensions: VBench comprises 16 dimensions in video generation (e.g., subject identity in-consistency, motion smoothness, temporal flickering, and spatial relationship, etc.). The evaluation metrics with fine-grained levels reveal individual models' strengths and weaknesses. 2) Human Alignment: We also provide a dataset of human preference annotations to validate our benchmarks' alignment with human perception, for each evaluation dimension respectively. 3) Valuable Insights: We look into current models' ability across various evaluation dimensions, and various content types. We also investi-gate the gaps between video and image generation models. We will open-source VBench, including all prompts, evaluation methods, generated videos, and human preference an-notations, and also include more video generation models in VBench to drive forward the field of video generation.

A unified and versatile LiDAR segmentation model with strong robustness and generalizability is desirable for safe autonomous driving perception. This work presents M3Net, a one-of-a-kind framework for fulfilling multitask, multi-dataset, multimodality LiDAR segmentation in a universal manner using just a single set of parameters. To better exploit data volume and diversity, we first combine large-scale driving datasets acquired by different types of sensors from diverse scenes and then conduct alignments in three spaces, namely data, feature, and label spaces, during the training. As a result, M3Net is capable of taming heterogeneous data for training state-of-the-art LiDAR segmentation models. Extensive experiments on twelve LiDAR segmentation datasets verify our effectiveness. Notably, using a shared set of parameters, M3Net achieves 75.1%,83.1%, and 72.4% mIoU scores, respectively, on the official benchmarks of SemanticKITTI, nuScenes, and Waymo Open.

Realistic 3D human generationfrom text prompts is a de-sirable yet challenging task. Existing methods optimize 3D representations like mesh or neural fields via score distil-lation sampling (SDS), which suffers from inadequate fine details or excessive training time. In this paper, we pro-pose an efficient yet effective framework, HumanGaussian, that generates high-quality 3D humans with fine-grained geometry and realistic appearance. Our key insight is that 3D Gaussian Splatting is an efficient renderer with peri-odic Gaussian shrinkage or growing, where such adaptive density control can be naturally guided by intrinsic human structures. Specifically, 1) we first propose a Structure-Aware SDS that simultaneously optimizes human appear-ance and geometry. The multi-modal score function from both RGB and depth space is leveraged to distill the Gaus-sian densification and pruning process. 2) Moreover, we devise an Annealed Negative Prompt Guidance by decom-posing SDS into a noisier generative score and a cleaner classifier score, which well addresses the over-saturation issue. The floating artifacts are further eliminated based on Gaussian size in a prune-only phase to enhance generation smoothness. Extensive experiments demonstrate the supe-rior efficiency and competitive quality of our framework, rendering vivid 3D humans under diverse scenarios.

Diffusion models generate images with an unprece-dented level of quality, but how can we freely rearrange image layouts? Recent works generate controllable scenes via learning spatially disentangled latent codes, but these methods do not apply to diffusion models due to their fixed forward process. In this work, we propose SceneDiffusion to optimize a layered scene representation during the diffusion sampling process. Our key insight is that spatial disentanglement can be obtained by jointly denoising scene rende rings at different spatial layouts. Our generated scenes support a wide range of spatial editing operations, including moving, resizing, cloning, and layer-wise appearance editing operations, including object restyling and replacing. Moreover, a scene can be generated conditioned on a ref-erence image, thus enabling object moving for in-the- wild images. Notably, this approach is training-free, compatible with general text-to-image diffusion models, and responsive in less than a second.

In this paper, we uncover the untapped potential of dif-fusion U-Net, which serves as a “free lunch” that substan-tially improves the generation quality on the fly. We initially investigate the key contributions of the U-Net architecture to the denoising process and identify that its main backbone primarily contributes to denoising, whereas its skip connections mainly introduce high-frequency features into the de-coder module, causing the potential neglect of crucial functions intrinsic to the backbone network. Capitalizing on this discovery, we propose a simple yet effective method, termed “FreeU”, which enhances generation quality without additional training or finetuning. Our key insight is to strategi-cally re-weight the contributions sourced from the U-Net's skip connections and backbone feature maps, to leverage the strengths of both components of the U-Net architec-ture. Promising results on image and video generation tasks demonstrate that our FreeU can be readily integrated to ex-isting diffusion models, e.g., Stable Diffusion, DreamBooth and ControlNet, to improve the generation quality with only a few lines of code. All you need is to adjust two scaling factors during inference.

Expressive human pose and shape estimation (a.k.a. 3D whole-body mesh recovery) involves the human body, hand, and expression estimation. Most existing methods have tack-led this task in a two-stage manner, first detecting the human body part with an off-the-shelf detection model and then in-ferring the different human body parts individually. Despite the impressive results achieved, these methods suffer from 1) loss of valuable contextual information via cropping, 2) introducing distractions, and 3) lacking inter-association among different persons and body parts, inevitably causing performance degradation, especially for crowded scenes. To address these issues, we introduce a novel ali-in-one-stage framework, AiOS, for multiple expressive human pose and shape recovery without an additional human detection step. Specifically, our method is built upon DETR, which treats multi-person whole-body mesh recovery task as a progressive set prediction problem with various sequential detection. We devise the decoder tokens and extend them to our task. Specifically, we first employ a human token to probe a hu-man location in the image and encode global features for each instance, which provides a coarse location for the later transformer block. Then, we introduce a joint-related token to probe the human joint in the image and encoder a fine-grained local feature, which collaborates with the global feature to regress the whole-body mesh. This straightfor-ward but effective model outperforms previous state-of-the-art methods by a 9% reduction in NMVE on AGORA, a 30% reduction in PVE on EHF, a 10% reduction in PVE on ARCTIC, and a 3% reduction in PVE on EgoBody.

The ability to learn from context with novel concepts, and deliver appropriate responses are essential in human conversations. Despite current Multimodal Large Language Models (MLLMs) and Large Language Models (LLMs) being trained on mega-scale datasets, recognizing unseen images or understanding novel concepts in a training-free manner remains a challenge. In-Context Learning (ICL) explores training-free few-shot learning, where models are encouraged to “learn to learn” from limited tasks and generalize to unseen tasks. In this work, we propose link-context learning (LCL), which emphasizes “reasoning from cause and effect” to augment the learning capabilities of MLLMs. LCL goes beyond traditional ICL by explicitly strengthening the causal relationship between the support set and the query set. By providing demonstrations with causal links, LCL guides the model to discern not only the analogy but also the underlying causal associations between data points, which empowers MLLMs to recognize unseen images and understand novel concepts more effectively. To facilitate the evaluation of this novel approach, we introduce the ISEKAI dataset, comprising exclusively of unseen generated image-label pairs designed for link-context learning. Extensive experiments show that our LCL-MLLM exhibits strong link-context learning capabilities to novel concepts over vanilla MLLMs. Code, demo, and dataset have been released.

While super-resolution (SR) methods based on diffusion models exhibit promising results, their practical application is hindered by the substantial number of required inference steps. Recent methods utilize the degraded images in the initial state, thereby shortening the Markov chain. Nevertheless, these solutions either rely on a precise formulation of the degradation process or still necessitate a relatively lengthy generation path (e.g., 15 iterations). To enhance inference speed, we propose a simple yet effective method for achieving single-step SR generation, named SinSR. Specifically, we first derive a deterministic sampling process from the most recent state-of-the-art (SOTA) method for accelerating diffusion-based SR. This allows the mapping between the input random noise and the generated high-resolution image to be obtained in a reduced and acceptable number of inference steps during training. We show that this deterministic mapping can be distilled into a student model that performs SR within only one inference step. Additionally, we propose a novel consistency-preserving loss to simultaneously leverage the ground-truth image during the distillation process, ensuring that the performance of the student model is not solely bound by the feature manifold of the teacher model, resulting in further performance improvement. Extensive experiments conducted on synthetic and real-world datasets demonstrate that the proposed method can achieve comparable or even superior performance compared to both previous SOTA methods and the teacher model, in just one sampling step, resulting in a remarkable up to × 10 speedup for inference. Our code will be released at https://github.com/wyf0912/SinSR/.

We introduce a new task - language-driven video inpainting, which uses natural language instructions to guide the inpainting process. This approach overcomes the limitations of traditional video inpainting methods that depend on manually labeled binary masks, a process often tedious and labor-intensive. We present the Remove Objects from Videos by Instructions (ROVI) dataset, containing 5,650 videos and 9,091 inpainting results, to support training and evaluation for this task. We also propose a novel diffusion-based language-driven video inpainting framework, the first end-to-end baseline for this task, integrating Multimodal Large Language Models to understand and execute complex language-based inpaintingrequests effectively. Our comprehensive results showcase the dataset's versatility and the model's effectiveness in various language-instructed inpainting scenarios. We have made datasets, code, and models publicly available at https://github.com/jianzongwu/Language-Driven-Video-Inpainting.

Despite recent advances in text-to-3D generative methods, there is a notable absence of reliable evaluation metrics. Existing metrics usually focus on a single criterion each, such as how well the asset aligned with the input text. These metrics lack the flexibility to generalize to different evaluation criteria and might not align well with human preferences. Conducting user preference studies is an alternative that offers both adaptability and human-aligned results. User studies, however, can be very ex-pensive to scale. This paper presents an automatic, ver-satile, and human-aligned evaluation metric for text-to-3D generative models. To this end, we first develop a prompt generator using GPT-4V to generate evaluating prompts, which serve as input to compare text-to-3D models. We further design a method instructing GPT-4V to compare two 3D assets according to user-defined crite-ria. Finally, we use these pairwise comparison results to assign these models Elo ratings. Experimental results suggest our metric strongly aligns with human preference across different evaluation criteria. Our code is available at https://github.com/3DTopia/GPTEval3D.

3D city generation is a desirable yet challenging task, since humans are more sensitive to structural distortions in urban environments. Additionally, generating 3D cities is more complex than 3D natural scenes since buildings, as objects of the same class, exhibit a wider range of appear-ances compared to the relatively consistent appearance of objects like trees in natural scenes. To address these challenges, we propose CityDreamer, a compositional generative model designed specifically for unbounded 3D cities. Our key insight is that 3D city generation should be a com-position of different types of neural fields: 1) various building instances, and 2) background stuff, such as roads and green lands. Specifically, we adopt the bird's eye view scene representation and employ a volumetric render for both instance-oriented and stuff-oriented neural fields. The generative hash grid and periodic positional embedding are tailored as scene parameterization to suit the distinct characteristics of building instances and background stuff. Furthermore, we contribute a suite of CityGen Datasets, including OSM and GoogleEarth, which comprises a vast amount of real-world city imagery to enhance the realism of the generated 3D cities both in their layouts and appear-ances. CityDreamer achieves state-of-the-art performance not only in generating realistic 3D cities but also in local-ized editing within the generated cities.

The remarkable efficacy of text-to-image diffusion models has motivated extensive exploration of their potential application in video domains. Zero-shot methods seek to extend image diffusion models to videos without necessitating model training. Recent methods mainly focus on incorporating inter-frame correspondence into attention mechanisms. However, the soft constraint imposed on determining where to attend to valid features can sometimes be insufficient, resulting in temporal inconsistency. In this paper, we introduce FRESCO, intra-frame correspondence alongside inter-frame correspondence to establish a more robust spatial-temporal constraint. This enhancement ensures a more consistent transformation of semantically similar content across frames. Beyond mere attention guidance, our approach involves an explicit update of features to achieve high spatial-temporal consistency with the input video, significantly improving the visual coherence of the resulting translated videos. Extensive experiments demonstrate the effectiveness of our proposed framework in producing high-quality, coherent videos, marking a notable improvement over existing zero-shot methods.

Diffusion models have emerged as the de facto paradigm for video generation. However, their reliance on web-scale data of varied quality often yields results that are visually unappealing and misaligned with the textual prompts. To tackle this problem, we propose InstructVideo to instruct text-to-video diffusion models with human feedback by reward fine-tuning. InstructVideo has two key ingredients: 1) To ameliorate the cost of reward fine-tuning induced by generating through the full DDIM sampling chain, we recast reward fine-tuning as editing. By leveraging the diffusion process to corrupt a sampled video, InstructVideo requires only partial inference of the DDIM sampling chain, reducing fine-tuning cost while improving fine-tuning efficiency. 2) To mitigate the absence of a dedicated video reward model for human preferences, we repurpose established image reward models, e.g., HPSv2. To this end, we propose Segmental Video Reward, a mechanism to provide reward signals based on segmental sparse sampling, and Temporally Attenuated Reward, a method that mitigates temporal modeling degradation during fine-tuning. Extensive experiments, both qualitative and quantitative, validate the practicality and efficacy of using image reward models in InstructVideo, significantly enhancing the visual quality of generated videos without compromising generalization capabilities. Code and models can be accessed through our project page https://instructvideo.github.io/.

In this work, we present Vlogger, a generic AI systemfor generating a minute-level video blog (i.e., vlog) of user de-scriptions. Different from short videos with a few seconds, vlog often contains a complex storyline with diversified scenes, which is challenging for most existing video generation approaches. To break through this bottleneck, our Vlogger smartly leverages Large Language Model (LLM) as Director and decomposes a long video generation task of vlog into four key stages, where we invoke various foundation models to play the critical roles of vlog profession-als, including (1) Script, (2) Actor, (3) ShowMaker, and (4) Voicer. With such a design of mimicking human beings, our Vlogger can generate vlogs through explainable cooperation of top-down planning and bottom-up shooting. More-over, we introduce a novel video diffusion model, Show-Maker, which serves as a videographer in our Vlogger for generating the video snippet of each shooting scene. By incorporating Script and Actor attentively as textual and visual prompts, it can effectively enhance spatial-temporal coherence in the snippet. Besides, we design a concise mixed training paradigm for ShowMaker, boosting its ca-pacity for both T2V generation and prediction. Finally, the extensive experiments show that our method achieves state-of-the-art performance on zero-shot T2V generation and prediction tasks. More importantly, Vlogger can generate over 5-minute vlogs from open-world descriptions, without loss of video coherence on script and actor.

Creating diverse and high-quality 3D assets with an automatic generative model is highly desirable. Despite extensive efforts on 3D generation, most existing works focus on the generation of a single category or a few categories. In this paper, we introduce a diffusion-based feed-forward framework for synthesizing massive categories of real-world 3D objects \textit{with a single generative model}. Notably, there are three major challenges for this large-vocabulary 3D generation: \textbf{a}) the need for expressive yet efficient 3D representation; \textbf{b}) large diversity in geometry and texture across categories; \textbf{c}) complexity in the appearances of real-world objects. To this end, we propose a novel triplane-based 3D-aware \textbf{Diff}usion model with \textbf{T}rans\textbf{F}ormer, \textbf{DiffTF}, for handling challenges via three aspects. \textbf{1}) Considering efficiency and robustness, we adopt a revised triplane representation and improve the fitting speed and accuracy. \textbf{2}) To handle the drastic variations in geometry and texture, we regard the features of all 3D objects as a combination of generalized 3D knowledge and specialized 3D features. To extract generalized 3D knowledge from diverse categories, we propose a novel 3D-aware transformer with shared cross-plane attention. It learns the cross-plane relations across different planes and aggregates the generalized 3D knowledge with specialized 3D features. \textbf{3}) In addition, we devise the 3D-aware encoder/decoder to enhance the generalized 3D knowledge in the encoded triplanes for handling categories with complex appearances. Extensive experiments on ShapeNet and OmniObject3D (over 200 diverse real-world categories) convincingly demonstrate that a single DiffTF model achieves state-of-the-art large-vocabulary 3D object generation performance with large diversity, rich semantics, and high quality. More results are available at https://difftf.github.io/

Recently video generation has achieved substantial progress with realistic results. Nevertheless, existing AI-generated videos are usually very short clips ("shot-level'') depicting a single scene. To deliver a coherent long video ("story-level''), it is desirable to have creative transition and prediction effects across different clips. This paper presents a short-to-long video diffusion model, SEINE, that focuses on generative transition and prediction. The goal is to generate high-quality long videos with smooth and creative transitions between scenes and varying lengths of shot-level videos. Specifically, we propose a random-mask video diffusion model to automatically generate transitions based on textual descriptions. By providing the images of different scenes as inputs, combined with text-based control, our model generates transition videos that ensure coherence and visual quality. Furthermore, the model can be readily extended to various tasks such as image-to-video animation and autoregressive video prediction. To conduct a comprehensive evaluation of this new generative task, we propose three assessing criteria for smooth and creative transition: temporal consistency, semantic similarity, and video-text semantic alignment. Extensive experiments validate the effectiveness of our approach over existing methods for generative transition and prediction, enabling the creation of story-level long videos.

Despite significant advances in large-scale text-to-image models, achieving hyper-realistic human image generation remains a desirable yet unsolved task. Existing models like Stable Diffusion and DALL·E 2 tend to generate human images with incoherent parts or unnatural poses. To tackle these challenges, our key insight is that human image is inherently structural over multiple granularities, from the coarse-level body skeleton to fine-grained spatial geometry. Therefore, capturing such correlations between the explicit appearance and latent structure in one model is essential to generate coherent and natural human images. To this end, we propose a unified framework, HyperHuman, that generates in-the-wild human images of high realism and diverse layouts. Specifically, 1) we first build a large-scale human-centric dataset, named HumanVerse, which consists of 340M images with comprehensive annotations like human pose, depth, and surface normal. 2) Next, we propose a Latent Structural Diffusion Model that simultaneously denoises the depth and surface normal along with the synthesized RGB image. Our model enforces the joint learning of image appearance, spatial relationship, and geometry in a unified network, where each branch in the model complements to each other with both structural awareness and textural richness. 3) Finally, to further boost the visual quality, we propose a Structure-Guided Refiner to compose the predicted conditions for more detailed generation of higher resolution. Extensive experiments demonstrate that our framework yields the state-of-the-art performance, generating hyper-realistic human images under diverse scenarios.

With the availability of large-scale video datasets and the advances of diffusion models, text-driven video generation has achieved substantial progress. However, existing video generation models are typically trained on a limited number of frames, resulting in the inability to generate high-fidelity long videos during inference. Furthermore, these models only support single-text conditions, whereas real-life scenarios often require multi-text conditions as the video content changes over time. To tackle these challenges, this study explores the potential of extending the text-driven capability to generate longer videos conditioned on multiple texts. 1) We first analyze the impact of initial noise in video diffusion models. Then building upon the observation of noise, we propose FreeNoise, a tuning-free and time-efficient paradigm to enhance the generative capabilities of pretrained video diffusion models while preserving content consistency. Specifically, instead of initializing noises for all frames, we reschedule a sequence of noises for long-range correlation and perform temporal attention over them by window-based fusion. 2) Additionally, we design a novel motion injection method to support the generation of videos conditioned on multiple text prompts. Extensive experiments validate the superiority of our paradigm in extending the generative capabilities of video diffusion models. It is noteworthy that compared with the previous best-performing method which brought about 255% extra time cost, our method incurs only negligible time cost of approximately 17%. Generated video samples are available at our website: http://haonanqiu.com/projects/FreeNoise.html.

We introduce a novel task within the field of human motion generation, termed dance accompaniment, which necessitates the generation of responsive movements from a dance partner, the "follower", synchronized with the lead dancer’s movements and the underlying musical rhythm. Unlike existing solo or group dance generation tasks, a duet dance scenario entails a heightened degree of interaction between the two participants, requiring delicate coordination in both pose and position. To support this task, we first build a large-scale and diverse duet interactive dance dataset, DD100, by recording about 117 minutes of professional dancers’ performances. To address the challenges inherent in this task, we propose a GPT based model, Duolando, which autoregressively predicts the subsequent tokenized motion conditioned on the coordinated information of the music, the leader’s and the follower’s movements. To further enhance the GPT’s capabilities of generating stable results on unseen conditions (music and leader motions), we devise an off-policy reinforcement learning strategy that allows the model to explore viable trajectories from out-of-distribution samplings, guided by human-defined rewards. Based on the collected dataset and proposed method, we establish a benchmark with several carefully designed metrics.

Recent advances in 3D content creation mostly leverage optimization-based 3D generation via score distillation sampling (SDS). Though promising results have been exhibited, these methods often suffer from slow per-sample optimization, limiting their practical usage. In this paper, we propose DreamGaussian, a novel 3D content generation framework that achieves both efficiency and quality simultaneously. Our key insight is to design a generative 3D Gaussian Splatting model with companioned mesh extraction and texture refinement in UV space. In contrast to the occupancy pruning used in Neural Radiance Fields, we demonstrate that the progressive densification of 3D Gaussians converges significantly faster for 3D generative tasks. To further enhance the texture quality and facilitate downstream applications, we introduce an efficient algorithm to convert 3D Gaussians into textured meshes and apply a fine-tuning stage to refine the details. Extensive experiments demonstrate the superior efficiency and competitive generation quality of our proposed approach. Notably, DreamGaussian produces high-quality textured meshes in just 2 minutes from a single-view image, achieving approximately 10 times acceleration compared to existing methods.

This paper introduces InternVid, a large-scale video-centric multimodal dataset that enables learning powerful and transferable video-text representations for multimodal understanding and generation. InternVid contains over 7 million videos lasting nearly 760K hours, yielding 234M video clips accompanied by detailed descriptions of total 4.1B words. Our core contribution is to develop a scalable approach to autonomously build a high-quality video-text dataset with large language models (LLM), thereby showcasing its efficacy in learning video-language representation at scale. Specifically, we utilize a multi-scale approach to generate video-related descriptions. Furthermore, we introduce ViCLIP, a video-text representation learning model based on ViT-L. Learned on InternVid via contrastive learning, this model demonstrates leading zero-shot action recognition and competitive video retrieval performance. Beyond basic video understanding tasks like recognition and retrieval, our dataset and model have broad applications. They are particularly beneficial for generating interleaved video-text data for learning a video-centric dialogue system, advancing video-to-text and text-to-video generation research. These proposed resources provide a tool for researchers and practitioners interested in multimodal video understanding and generation.

3D facial animation has attracted considerable attention due to its extensive applications in the multimedia field. Audio-driven 3D facial animation has been widely explored with promising results. However, multi-modal 3D facial animation, especially text-guided 3D facial animation is rarely explored due to the lack of multi-modal 3D facial animation dataset. To fill this gap, we first construct a large-scale multi-modal 3D facial animation dataset, MMHead, which consists of 49 hours of 3D facial motion sequences, speech audios, and rich hierarchical text annotations. Each text annotation contains abstract action and emotion descriptions, fine-grained facial and head movements (i.e., expression and head pose) descriptions, and three possible scenarios that may cause such emotion. Concretely, we integrate five public 2D portrait video datasets, and propose an automatic pipeline to 1) reconstruct 3D facial motion sequences from monocular videos; and 2) obtain hierarchical text annotations with the help of AU detection and ChatGPT. Based on the MMHead dataset, we establish benchmarks for two new tasks: text-induced 3D talking head animation and text-to-3D facial motion generation. Moreover, a simple but efficient VQ-VAE-based method named MM2Face is proposed to unify the multi-modal information and generate diverse and plausible 3D facial motions, which achieves competitive results on both benchmarks. Extensive experiments and comprehensive analysis demonstrate the significant potential of our dataset and benchmarks in promoting the development of multi-modal 3D facial animation. The dataset will be released at: https://wsj-sjtu.github.io/MMHead/.

Physically realistic materials are pivotal in augmenting the realism of 3D assets across various applications and lighting conditions. However, existing 3D assets and generative models often lack authentic material properties. Manual assignment of materials using graphic software is a tedious and time-consuming task. In this paper, we exploit advancements in Multimodal Large Language Models (MLLMs), particularly GPT-4V, to present a novel approach, Make-it-Real: 1) We demonstrate that GPT-4V can effectively recognize and describe materials, allowing the construction of a detailed material library. 2) Utilizing a combination of visual cues and hierarchical text prompts, GPT-4V precisely identifies and aligns materials with the corresponding components of 3D objects. 3) The correctly matched materials are then meticulously applied as reference for the new SVBRDF material generation according to the original albedo map, significantly enhancing their visual authenticity. Make-it-Real offers a streamlined integration into the 3D content creation workflow, showcasing its utility as an essential tool for developers of 3D assets.

Conditional diffusion models can create unseen images in various settings, aiding image interpolation. Interpolation in latent spaces is well-studied, but interpolation with specific conditions like text or image is less understood. Common approaches interpolate linearly in the conditioning space but tend to result in inconsistent images with poor fidelity. This work introduces a novel training-free technique named \textbf{Attention Interpolation via Diffusion (AID)}. AID has two key contributions: \textbf{1)} a fused inner/outer interpolated attention layer to boost image consistency and fidelity; and \textbf{2)} selection of interpolation coefficients via a beta distribution to increase smoothness. Additionally, we present an AID variant called \textbf{Prompt-guided Attention Interpolation via Diffusion (PAID)}, which \textbf{3)} treats interpolation as a condition-dependent generative process. Experiments demonstrate that our method achieves greater consistency, smoothness, and efficiency in condition-based interpolation, aligning closely with human preferences. Furthermore, PAID offers substantial benefits for compositional generation, controlled image editing, image morphing and image-controlled generation, all while remaining training-free.

We present L4GM, the first 4D Large Reconstruction Model that produces animated objects from a single-view video input -- in a single feed-forward pass that takes only a second. Key to our success is a novel dataset of multiview videos containing curated, rendered animated objects from Objaverse. This dataset depicts 44K diverse objects with 110K animations rendered in 48 viewpoints, resulting in 12M videos with a total of 300M frames. We keep our L4GM simple for scalability and build directly on top of LGM, a pretrained 3D Large Reconstruction Model that outputs 3D Gaussian ellipsoids from multiview image input. L4GM outputs a per-frame 3D Gaussian splat representation from video frames sampled at a low fps and then upsamples the representation to a higher fps to achieve temporal smoothness. We add temporal self-attention layers to the base LGM to help it learn consistency across time, and utilize a per-timestep multiview rendering loss to train the model. The representation is upsampled to a higher framerate by training an interpolation model which produces intermediate 3D Gaussian representations. We showcase that L4GM that is only trained on synthetic data generalizes well on in-the-wild videos, producing high quality animated 3D assets.

Recent advances in text-to-image generation have enabled the creation of high-quality images with diverse applications. However, accurately describing desired visual attributes can be challenging, especially for non-experts in art and photography. An intuitive solution involves adopting favorable attributes from source images. Current methods attempt to distill identity and style from source images. However, "style" is a broad concept that includes texture, color, and artistic elements, but does not cover other important attributes like lighting and dynamics. Additionally, a simplified "style" adaptation prevents combining multiple attributes from different sources into one generated image. In this work, we formulate a more effective approach to decompose the aesthetics of a picture into specific visual attributes, letting users apply characteristics like lighting, texture, and dynamics from different images. To achieve this goal, we constructed the first fine-grained visual attributes dataset (FiVA) to the best of our knowledge. This FiVA dataset features a well-organized taxonomy for visual attributes and includes 1 M high-quality generated images with visual attribute annotations. Leveraging this dataset, we propose a fine-grained visual attributes adaptation framework (FiVA-Adapter) , which decouples and adapts visual attributes from one or more source images into a generated one. This approach enhances user-friendly customization, allowing users to selectively apply desired attributes to create images that meet their unique preferences and specific content requirements.

Diffusion models gain increasing popularity for their generative capabilities.Recently, there have been surging needs to generate customized images by inverting diffusion models from exemplar images, and existing inversion methods mainly focus on capturing object appearances (i.e., the "look").However, how to invert object relations, another important pillar in the visual world, remains unexplored.In this work, we propose the Relation Inversion task, which aims to learn a specific relation (represented as "relation prompt") from exemplar images.Specifically, we learn a relation prompt with a frozen pre-trained text-to-image diffusion model.The learned relation prompt can then be applied to generate relation-specific images with new objects, backgrounds, and styles.To tackle the Relation Inversion task, we propose the ReVersion Framework.Specifically, we propose a novel "relation-steering contrastive learning" scheme to steer the relation prompt towards relation-dense regions, and disentangle it away from object appearances.We further devise "relation-focal importance sampling" to emphasize high-level interactions over low-level appearances (e.g., texture, color).To comprehensively evaluate this new task, we contribute the ReVersion Benchmark, which provides various exemplar images with diverse

Image- and video-based 3D human recovery (i.e., pose and shape estimation) have achieved substantial progress. However, due to the prohibitive cost of motion capture, existing datasets are often limited in scale and diversity. In this work, we obtain massive human sequences by playing the video game with automatically annotated 3D ground truths. Specifically, we contribute GTA-Human, a large-scale 3D human dataset generated with the GTA-V game engine, featuring a highly diverse set of subjects, actions, and scenarios. More importantly, we study the use of game-playing data and obtain five major insights. First, game-playing data is surprisingly effective. A simple frame-based baseline trained on GTA-Human outperforms more sophisticated methods by a large margin. For video-based methods, GTA-Human is even on par with the in-domain training set. Second, we discover that synthetic data provides critical complements to the real data that is typically collected indoor. We highlight that our investigation into domain gap provides explanations for our data mixture strategies that are simple yet useful, which offers new insights to the research community. Third, the scale of the dataset matters. The performance boost is closely related to the additional data available. A systematic study on multiple key factors (such as camera angle and body pose) reveals that the model performance is sensitive to data density. Fourth, the effectiveness of GTA-Human is also attributed to the rich collection of strong supervision labels (SMPL parameters), which are otherwise expensive to acquire in real datasets. Fifth, the benefits of synthetic data extend to larger models such as deeper convolutional neural networks (CNNs) and Transformers, for which a significant impact is also observed. We hope our work could pave the way for scaling up 3D human recovery to the real world.

With the rapid advances in autonomous driving, it becomes critical to equip its sensing system with more holistic 3D perception. However, widely explored tasks like 3D detection or point cloud semantic segmentation focus on parsing either the objects or scenes. In this work, we propose to address the challenging task of LiDAR-based Panoptic Segmentation, which aims to parse both objects and scenes in a unified manner. In particular, we propose Dynamic Shifting Network (DS-Net), which serves as an effective panoptic segmentation framework in the point cloud realm. DS-Net features a dynamic shifting module for complex LiDAR point cloud distributions. We present an efficient learnable clustering module, dynamic shifting, which adapts kernel functions for different instances. To further explore the temporal information, we extend the single-scan processing framework to its temporal version, 4D-DS-Net, for the task of 4D Panoptic Segmentation, where the same instance across multiple frames should be given the same ID prediction. Instead of naively appending a tracking module to DS-Net, we propose to solve the 4D panoptic segmentation in a more unified way. Specifically, 4D-DS-Net first constructs 4D data volume by aligning consecutive LiDAR scans, upon which the temporally unified instance clustering is performed to obtain the final results. Extensive experiments on two large-scale autonomous driving LiDAR datasets, SemanticKITTI and Panoptic nuScenes, are conducted to demonstrate the effectiveness and superior performance of the proposed solution.

Facial editing is to manipulate the facial attributes of a given face image. Nowadays, with the development of generative models, users can easily generate 2D and 3D facial images with high fidelity and 3D-aware consistency. However, existing works are incapable of delivering a continuous and fine-grained editing mode (e.g., editing a slightly smiling face to a big laughing one) with natural interactions with users. In this work, we propose Talk-to-Edit, an interactive facial editing framework that performs fine-grained attribute manipulation through dialog between the user and the system. Our key insight is to model a continual "semantic field" in the GAN latent space. 1) Unlike previous works that regard the editing as traversing straight lines in the latent space, here the fine-grained editing is formulated as finding a curving trajectory that respects fine-grained attribute landscape on the semantic field. 2) The curvature at each step is location-specific and determined by the input image as well as the users' language requests. 3) To engage the users in a meaningful dialog, our system generates language feedback by considering both the user request and the current state of the semantic field. We demonstrate the effectiveness of our proposed framework on both 2D and 3D-aware generative models. We term the semantic field for the 3D-aware models as "tri-plane" flow, as it corresponds to the changes not only in the color space but also in the density space. We also contribute CelebA-Dialog, a visual-language facial editing dataset to facilitate large-scale study. Specifically, each image has manually annotated fine-grained attribute annotations as well as template-based textual descriptions in natural language. Extensive quantitative and qualitative experiments demonstrate the superiority of our framework in terms of 1) the smoothness of fine-grained editing, 2) the identity/attribute preservation, and 3) the visual photorealism and dialog fluency. Notably, the user study validates that our overall system is consistently favored by around 80% of the participants.

Visual segmentation seeks to partition images, video frames, or point clouds into multiple segments or groups. This technique has numerous real-world applications, such as autonomous driving, image editing, robot sensing, and medical analysis. Over the past decade, deep learning-based methods have made remarkable strides in this area. Recently, transformers, a type of neural network based on self-attention originally designed for natural language processing, have considerably surpassed previous convolutional or recurrent approaches in various vision processing tasks. Specifically, vision transformers offer robust, unified, and even simpler solutions for various segmentation tasks. This survey provides a thorough overview of transformer-based visual segmentation, summarizing recent advancements. We first review the background, encompassing problem definitions, datasets, and prior convolutional methods. Next, we summarize a meta-architecture that unifies all recent transformer-based approaches. Based on this meta-architecture, we examine various method designs, including modifications to the meta-architecture and associated applications. We also present several specific subfields, including 3D point cloud segmentation, foundation model tuning, domain-aware segmentation, efficient segmentation, and medical segmentation. Additionally, we compile and re-evaluate the reviewed methods on several well-established datasets. Finally, we identify open challenges in this field and propose directions for future research.

Real world data often exhibits a long-tailed and open-ended (i.e., with unseen classes) distribution. A practical recognition system must balance between majority (head) and minority (tail) classes, generalize across the distribution, and acknowledge novelty upon the instances of unseen classes (open classes). We define Open Long-Tailed Recognition++ (OLTR++) as learning from such naturally distributed data and optimizing for the classification accuracy over a balanced test set which includes both known and open classes. OLTR++ handles imbalanced classification, few-shot learning, open-set recognition, and active learning in one integrated algorithm, whereas existing classification approaches often focus only on one or two aspects and deliver poorly over the entire spectrum. The key challenges are: 1) how to share visual knowledge between head and tail classes, 2) how to reduce confusion between tail and open classes, and 3) how to actively explore open classes with learned knowledge. Our algorithm, OLTR++, maps images to a feature space such that visual concepts can relate to each other through a memory association mechanism and a learned metric (dynamic meta-embedding) that both respects the closed world classification of seen classes and acknowledges the novelty of open classes. Additionally, we propose an active learning scheme based on visual memory, which learns to recognize open classes in a data-efficient manner for future expansions. On three large-scale open long-tailed datasets we curated from ImageNet (object-centric), Places (scene-centric), and MS1M (face-centric) data, as well as three standard benchmarks (CIFAR-10-LT, CIFAR-100-LT, and iNaturalist-18), our approach, as a unified framework, consistently demonstrates competitive performance. Notably, our approach also shows strong potential for the active exploration of open classes and the fairness analysis of minority groups.

With the prevalent use of LiDAR sensors in autonomous driving, 3D point cloud object tracking has received increasing attention. In a point cloud sequence, 3D object tracking aims to predict the location and orientation of an object in consecutive frames. Motivated by the success of transformers, we propose Point Tracking TRansformer (PTTR), which efficiently predicts high-quality 3D tracking results in a coarse-to-fine manner with the help of transformer operations. PTTR consists of three novel designs. 1) Instead of random sampling, we design Relation-Aware Sampling to preserve relevant points to the given template during subsampling. 2) We propose a Point Relation Transformer for effective feature aggregation and feature matching between the template and search region. 3) Based on the coarse tracking results, we employ a novel Prediction Refinement Module to obtain the final refined prediction through local feature pooling. In addition, motivated by the favorable properties of the Bird's-Eye View (BEV) of point clouds in capturing object motion, we further design a more advanced framework named PTTR++, which incorporates both the point-wise view and BEV representation to exploit their complementary effect in generating high-quality tracking results. PTTR++ substantially boosts the tracking performance on top of PTTR with low computational overhead. Extensive experiments over multiple datasets show that our proposed approaches achieve superior 3D tracking accuracy and efficiency.

Misinformation has become a pressing issue. Fake media, in both visual and textual forms, is widespread on the web. Whilevarious deepfake detection and text fake news detection methods have been proposed, they are only designed for single-modality forgery based on binary classification, let alone analyzing and reasoning subtle forgery traces across different modalities. In this paper, we highlight a new research problem for multi-modal fake media, namelyDetectingandGroundingMulti-ModalMediaManipulation (DGM4).DGM4aims to not only detect the authenticity of multi-modal media, but also ground the manipulated content, which requires deeper reasoning of multi-modal media manipulation. To support a large-scale investigation, we construct the first DGM4 dataset. Moreover, we propose a novelHierArchicalMulti-modalManipulation rEasoning tRansformer (HAMMER) to fully capture the fine-grained interaction between different modalities. HAMMER performs 1) manipulation-aware contrastive learning between two uni-modal encoders as shallow manipulation reasoning, and 2) modality-aware cross-attention by multi-modal aggregator as deep manipulation reasoning. Dedicated manipulation detection and grounding heads are integrated from shallow to deep levels based on the interacted multi-modal information. To exploit more fine-grained contrastive learning for cross-modal semantic alignment, we further integrate Manipulation-Aware Contrastive Loss with Local View and construct a more advanced modelHAMMER++Finally, we build an extensive benchmark and set up rigorous evaluation metrics for this new research problem. Comprehensive experiments demonstrate the superiority ofHAMMERandHAMMER++; several valuable observations are also revealed to facilitate future research in multi-modal media manipulation..

Neural Radiance Field (NeRF) has achieved substantial progress in novel view synthesis given multi-view images. Recently, some works have attempted to train a NeRF from a single image with 3D priors. They mainly focus on a limited field of view with a few occlusions, which greatly limits their scalability to real-world 360-degree panoramic scenarios with large-size occlusions. In this paper, we present PERF, a 360-degree novel view synthesis framework that trains a panoramic neural radiance field from a single panorama. Notably, PERF allows 3D roaming in a complex scene without expensive and tedious image collection. To achieve this goal, we propose a novel collaborative RGBD inpainting method and a progressive inpainting-and-erasing method to lift up a 360-degree 2D scene to a 3D scene. Specifically, we first predict a panoramic depth map as initialization given a single panorama and reconstruct visible 3D regions with volume rendering. Then we introduce a collaborative RGBD inpainting approach into a NeRF for completing RGB images and depth maps from random views, which is derived from an RGB Stable Diffusion model and a monocular depth estimator. Finally, we introduce an inpainting-and-erasing strategy to avoid inconsistent geometry between a newly-sampled view and reference views. The two components are integrated into the learning of NeRFs in a unified optimization framework and achieve promising results. Extensive experiments on Replica and a new dataset PERF-in-the-wild demonstrate the superiority of our PERF over state-of-the-art methods. Our PERF can be widely used for real-world applications, such as panorama-to-3D, text-to-3D, and 3D scene stylization applications.

Panoptic Scene Graph (PSG) is a challenging task in Scene Graph Generation (SGG) that aims to create a more comprehensive scene graph representation using panoptic segmentation instead of boxes. Compared to SGG, PSG has several challenging problems: pixel-level segment outputs and full relationship exploration (It also considers thing and stuff relation). Thus, current PSG methods have limited performance, which hinders downstream tasks or applications. This work aims to design a novel and strong baseline for PSG. To achieve that, we first conduct an in-depth analysis to identify the bottleneck of the current PSG models, finding that inter-object pair-wise recall is a crucial factor that was ignored by previous PSG methods. Based on this and the recent query-based frameworks, we present a novel framework: Pair then Relation (Pair-Net), which uses a Pair Proposal Network (PPN) to learn and filter sparse pair-wise relationships between subjects and objects. Moreover, we also observed the sparse nature of object pairs for both. Motivated by this, we design a lightweight Matrix Learner within the PPN, which directly learns pair-wised relationships for pair proposal generation. Through extensive ablation and analysis, our approach significantly improves upon leveraging the segmenter solid baseline. Notably, our method achieves over 10% absolute gains compared to our baseline, PSGFormer.

Human motion modeling is important for many modern graphics applications, which typically require professional skills. In order to remove the skill barriers for laymen, recent motion generation methods can directly generate human motions conditioned on natural languages. However, it remains challenging to achieve diverse and fine-grained motion generation with various text inputs. To address this problem, we propose MotionDiffuse, one of the first diffusion model-based text-driven motion generation frameworks, which demonstrates several desired properties over existing methods. 1) Probabilistic Mapping. Instead of a deterministic language-motion mapping, MotionDiffuse generates motions through a series of denoising steps in which variations are injected. 2) Realistic Synthesis. MotionDiffuse excels at modeling complicated data distribution and generating vivid motion sequences. 3) Multi-Level Manipulation. MotionDiffuse responds to fine-grained instructions on body parts, and arbitrary-length motion synthesis with time-varied text prompts. Our experiments show MotionDiffuse outperforms existing SoTA methods by convincing margins on text-driven motion generation and action-conditioned motion generation. A qualitative analysis further demonstrates MotionDiffuse's controllability for comprehensive motion generation.

Deep models, e.g., CNNs and Vision Transformers, have achieved impressive achievements in many vision tasks in the closed world. However, novel classes emerge from time to time in our ever-changing world, requiring a learning system to acquire new knowledge continually. Class-Incremental Learning (CIL) enables the learner to incorporate the knowledge of new classes incrementally and build a universal classifier among all seen classes. Correspondingly, when directly training the model with new class instances, a fatal problem occurs - the model tends to catastrophically forget the characteristics of former ones, and its performance drastically degrades. There have been numerous efforts to tackle catastrophic forgetting in the machine learning community. In this paper, we survey comprehensively recent advances in class-incremental learning and summarize these methods from several aspects. We also provide a rigorous and unified evaluation of 17 methods in benchmark image classification tasks to find out the characteristics of different algorithms empirically. Furthermore, we notice that the current comparison protocol ignores the influence of memory budget in model storage, which may result in unfair comparison and biased results. Hence, we advocate fair comparison by aligning the memory budget in evaluation, as well as several memory-agnostic performance measures.

Video-to-Video synthesis (Vid2Vid) gains remarkable performance in generating a photo-realistic video from a sequence of semantic maps, such as segmentation, sketch and pose. However, this pipeline is heavily limited to high computational cost and long inference latency, mainly attributed to two essential factors: 1) network architecture parameters, 2) sequential data stream. Recently, the parameters of image-based generative models have been significantly reduced via more efficient network architectures. Existing methods mainly focus on slimming network architectures but ignore the size of the sequential data stream. Moreover, due to the lack of temporal coherence, image-based compression is not sufficient for the compression of the video task. In this paper, we present a spatial-temporal hybrid distillation compression framework, Fast-Vid2Vid++, which focuses on knowledge distillation of the teacher network and the data stream of generative models on both space and time. Fast-Vid2Vid++ makes the first attempt at time dimension to transfer hierarchical features and time coherence knowledge to reduce computational resources and accelerate inference. Specifically, we compress the data stream spatially and reduce the temporal redundancy. We distill the knowledge of the hierarchical features and the final response from the teacher network to the student network in high-resolution and full-time domains. We transfer the long-term dependencies of the features and video frames to the student model. After the proposed spatial-temporal hybrid knowledge distillation (Spatial-Temporal-HKD), our model can synthesize high-resolution key-frames using the low-resolution data stream. Finally, Fast-Vid2Vid++ interpolates intermediate frames by motion compensation with slight latency and generates full-length sequences with motion-aware inference (MAI). On standard benchmarks, Fast-Vid2Vid++ achieves a real-time performance of 30-59 FPS and saves 28-35× computational cost on a single V100 GPU.

Continual learning (CL) desires that the neural network sequentially perform learning tasks from a dynamic data stream without forgetting learned knowledge. To overcome forgetting, a line of work relies on gradient projection to minimize the influence between gradients during optimization. This article focuses on a challenging problem concerning CL:When, how, and where to implement gradient projection to promote CL.Tackling this problem can be divided into two perspectives, namely the gradient direction (when and how) and the area of gradient conflict (where). First, we propose a plug-and-play method UniGrad to tackle the inconsistency of conflicting and nonconflicting gradients during optimization in CL. Second, we explore the interaction mechanism of gradient projection and loss landscape in CL, and further propose a pluggable method UniGrad-FS to improve the CL performance. In short, this work expects to overcome forgetting through an efficient gradient projection at the area where the gradient conflicts are less intense. In essence, the proposed method is a general and pluggable method that can be used in any gradient-based optimizer. For evaluation, we plug UniGrad and UniGrad-FS into two top-performing baselines (WA and MEMO). Our method shows clear improvements, i.e., boosting WA and MEMO by +2.09% and 1.72% in the 20-step of the CIFAR100 benchmark. In addition, we observe performance enhancement on all settings of CIFAR100 and Tiny-ImageNet datasets. Extensive experiments demonstrate the simplicity and effectiveness of the proposed method.

Predicting interactions between proteins is one of the most important yet challenging problems in structural bioinformatics. Intrinsically, potential function sites in protein surfaces are determined by both geometric and chemical features. However, existing works only consider handcrafted or individually learned chemical features from the atom type and extract geometric features independently. Here, we identify two key properties of effective protein surface learning: 1) relationship among atoms: atoms are linked with each other by covalent bonds to form biomolecules instead of appearing alone, leading to the significance of modeling the relationship among atoms in chemical feature learning. 2) hierarchical feature interaction: the neighboring residue effect validates the significance of hierarchical feature interaction among atoms and between surface points and atoms (or residues). In this paper, we present a principled framework based on deep learning techniques, namely Hierarchical Chemical and Geometric Feature Interaction Network (HCGNet), for protein surface analysis by bridging chemical and geometric features with hierarchical interactions. Extensive experiments demonstrate that our method outperforms the prior state-of-the-art method by 2.3% in site prediction task and 3.2 available at https://github.com/lyqun/HCGNet.

While progress has been made in the field of portrait reenactment, the problem of how to produce high-fidelity and robust videos remains. Recent studies normally find it challenging to handle rarely seen target poses due to the limitation of source data. This paper proposes the Video Portrait via Non-local Quantization Modeling (VPNQ) framework, which produces pose- and disturbance-robust reenactable video portraits. Our key insight is to learn position-invariant quantized local patch representations and build a mapping between simple driving signals and local textures with non-local spatial-temporal modeling. Specifically, instead of learning a universal quantized codebook, we identify that a personalized one can be trained to preserve desired position-invariant local details better. Then, a simple representation of projected landmarks can be used as sufficient driving signals to avoid 3D rendering. Following, we employ a carefully designed Spatio-Temporal Transformer to predict reasonable and temporally consistent quantized tokens from the driving signal. The predicted codes can be decoded back to robust and high-quality videos. Comprehensive experiments have been conducted to validate the effectiveness of our approach.

This paper presents a novel grid-based NeRF called F2- NeRF (Fast-Free-NeRF) for novel view synthesis, which enables arbitrary input camera trajectories and only costs a few minutes for training. Existing fast grid-based NeRF training frameworks, like Instant-NGP, Plenoxels, DVGO, or TensoRF, are mainly designed for bounded scenes and rely on space warping to handle unbounded scenes. Existing two widely-used space-warping methods are only designed for the forward-facing trajectory or the 360° object-centric trajectory but cannot process arbitrary trajectories. In this paper, we delve deep into the mechanism of space warping to handle unbounded scenes. Based on our analysis, we further propose a novel space-warping method called perspective warping, which allows us to handle arbitrary trajectories in the grid-based NeRF framework. Extensive experiments demonstrate that F2-NeRF is able to use the same perspective warping to render high-quality images on two standard datasets and a new free trajectory dataset collected by us. Project page: totoro97.github.io/projects/f2-nerf.

Despite recent advances in syncing lip movements with any audio waves, current methods still struggle to balance generation quality and the model's generalization ability. Previous studies either require long-term data for training or produce a similar movement pattern on all subjects with low quality. In this paper, we propose StyleSync, an effective framework that enables high-fidelity lip synchronization. We identify that a style-based generator would sufficiently enable such a charming property on both one-shot and few-shot scenarios. Specifically, we design a mask-guided spatial information encoding module that preserves the details of the given face. The mouth shapes are accurately modified by audio through modulated convolutions. Moreover, our design also enables personalized lip-sync by introducing style space and generator refinement on only limited frames. Thus the identity and talking style of a target person could be accurately preserved. Extensive experiments demonstrate the effectiveness of our method in producing high-fidelity results on a variety of scenes. Resources can be found at https:/hangz-nju-cuhk.github.io/projects/StyleSync.

Diffusion models arise as a powerful generative tool recently. Despite the great progress, existing diffusion models mainly focus on uni-modal control, i.e., the diffusion process is driven by only one modality of condition. To further unleash the users' creativity, it is desirable for the model to be controllable by multiple modalities simultaneously, e.g. generating and editing faces by describing the age (text-driven) while drawing the face shape (mask-driven). In this work, we present Collaborative Diffusion, where pre-trained uni-modal diffusion models collaborate to achieve multi-modal face generation and editing without re-training. Our key insight is that diffusion models driven by different modalities are inherently complementary regarding the latent denoising steps, where bilateral connections can be established upon. Specifically, we propose dynamic diffuser, a meta-network that adaptively hallucinates multimodal denoising steps by predicting the spatial-temporal influence functions for each pre-trained uni-modal model. Collaborative Diffusion not only collaborates generation capabilities from uni-modal diffusion models, but also integrates multiple uni-modal manipulations to perform multi-modal editing. Extensive qualitative and quantitative experiments demonstrate the superiority of our framework in both image quality and condition consistency.

Densely annotating LiDAR point clouds is costly, which often restrains the scalability of fully-supervised learning methods. In this work, we study the underexplored semi-supervised learning (SSL) in LiDAR semantic segmentation. Our core idea is to leverage the strong spatial cues of LiDAR point clouds to better exploit unlabeled data. We propose LaserMix to mix laser beams from different LiDAR scans and then encourage the model to make consistent and confident predictions before and after mixing. Our framework has three appealing properties. 1) Generic: LaserMix is agnostic to LiDAR representations (e.g., range view and voxel), and hence our SSL framework can be universally applied. 2) Statistically grounded: We provide a detailed analysis to theoretically explain the applicability of the proposed framework. 3) Effective: Comprehensive experimental analysis on popular LiDAR segmentation datasets (nuScenes, SemanticKITTI, and ScribbleKITTI) demonstrates our effectiveness and superiority. Notably, we achieve competitive results over fully-supervised counterparts with 2× to 5× fewer labels and improve the supervised-only baseline significantly by relatively 10.8%. We hope this concise yet high-performing framework could facilitate future research in semi-supervised LiDAR segmentation. Code is publicly available11https://github.com/ldkong1205/LaserMix..

Misinformation has become a pressing issue. Fake media, in both visual and textual forms, is widespread on the web. While various deepfake detection and text fake news detection methods have been proposed, they are only designed for single-modality forgery based on binary classification, let alone analyzing and reasoning subtle forgery traces across different modalities. In this paper, we high-light a new research problem for multi-modal fake media, namely Detecting and Grounding Multi-Modal Media Manipulation (DGM4). DGM4aims to not only detect the authenticity of multi-modal media, but also ground the manipulated content (i.e., image bounding boxes and text tokens), which requires deeper reasoning of multi-modal media manipulation. To support a large-scale investigation, we construct the first DGM4dataset, where image-text pairs are manipulated by various approaches, with rich annotation of diverse manipulations. Moreover, we propose a novel HierArchical Multi-modal Manipulation rEasoning tRansformer (HAMMER) to fully capture the fine-grained interaction between different modalities. HAMMER performs 1) manipulation-aware contrastive learning between two uni-modal encoders as shallow manipulation reasoning, and 2) modality-aware cross-attention by multi-modal aggregator as deep manipulation reasoning. Dedicated manipulation detection and grounding heads are integrated from shallow to deep levels based on the interacted multi-modal information. Finally, we build an extensive bench-mark and set up rigorous evaluation metrics for this new research problem. Comprehensive experiments demonstrate the superiority of our model; several valuable observations are also revealed to facilitate future research in multi-modal media manipulation.

Recent advances in modeling 3D objects mostly rely on synthetic datasets due to the lack of large-scale real-scanned 3D databases. To facilitate the development of 3D perception, reconstruction, and generation in the real world, we propose OmniObject3D, a large vocabulary 3D object dataset with massive high-quality real-scanned 3D objects. OmniObject3D has several appealing properties: 1) Large Vocabulary: It comprises 6,000 scanned objects in 190 daily categories, sharing common classes with popular 2D datasets (e.g., ImageNet and LVIS), benefiting the pursuit of generalizable 3D representations. 2) Rich Annotations: Each 3D object is captured with both 2D and 3D sensors, providing textured meshes, point clouds, multi-view rendered images, and multiple real-captured videos. 3) Realistic Scans: The professional scanners support high-quality object scans with precise shapes and realistic appearances. With the vast exploration space offered by OmniObject3D, we carefully set up four evaluation tracks: a) robust 3D perception, b) novel-view synthesis, c) neural surface reconstruction, and d) 3D object generation. Extensive studies are performed on these four benchmarks, revealing new observations, challenges, and opportunities for future research in realistic 3D vision.

Towards building comprehensive real-world visual perception systems, we propose and study a new problem called panoptic scene graph generation (PVSG). PVSG is related to the existing video scene graph generation (VidSGG) problem, which focuses on temporal interactions between humans and objects localized with bounding boxes in videos. However, the limitation of bounding boxes in detecting non-rigid objects and backgrounds often causes VidSGG systems to miss key details that are crucial for comprehensive video understanding. In contrast, PVSG requires nodes in scene graphs to be grounded by more precise, pixel-level segmentation masks, which facilitate holistic scene understanding. To advance research in this new area, we contribute a high-quality PVSG dataset, which consists of 400 videos (289 third-person + 111 egocentric videos) with totally 150K frames labeled with panoptic segmentation masks as well as fine, temporal scene graphs. We also provide a variety of baseline methods and share useful design practices for future work.

Animating virtual avatars to make co-speech gestures facilitates various applications in human-machine interaction. The existing methods mainly rely on generative adversarial networks (GANs), which typically suffer from notorious mode collapse and unstable training, thus making it difficult to learn accurate audio-gesture joint distributions. In this work, we propose a novel diffusion-based framework, named Diffusion Co-Speech Gesture (DiffGesture), to effectively capture the cross-modal audio-to-gesture associations and preserve temporal coherence for high-fidelity audio-driven co-speech gesture generation. Specifically, we first establish the diffusion-conditional generation process on clips of skeleton sequences and audio to enable the whole framework. Then, a novel Diffusion Audio-Gesture Transformer is devised to better attend to the information from multiple modalities and model the long-term temporal dependency. Moreover, to eliminate temporal inconsistency, we propose an effective Diffusion Gesture Stabilizer with an annealed noise sampling strategy. Benefiting from the architectural advantages of diffusion models, we further incorporate implicit classifier-free guidance to trade off between diversity and gesture quality. Extensive experiments demonstrate that DiffGesture achieves state-of-the-art performance, which renders coherent gestures with better mode coverage and stronger audio correlations. Code is available at https://github.com/Advocate99/DiffGesture.

Text-driven content creation has evolved to be a transformative technique that revolutionizes creativity. Here we study the task of text-driven human video generation, where a video sequence is synthesized from texts describing the appearance and motions of a target performer. Compared to general text-driven video generation, human-centric video generation requires maintaining the appearance of synthesized human while performing complex motions. In this work, we present Text2Performer to generate vivid human videos with articulated motions from texts. Text2Performer has two novel designs: 1) decomposed human representation and 2) diffusion-based motion sampler. First, we decompose the VQVAE latent space into human appearance and pose representation in an unsupervised manner by utilizing the nature of human videos. In this way, the appearance is well maintained along the generated frames. Then, we propose continuous VQ-diffuser to sample a sequence of pose embeddings. Unlike existing VQ-based methods that operate in the discrete space, continuous VQ-diffuser directly outputs the continuous pose embeddings for better motion modeling. Finally, motion-aware masking strategy is designed to mask the pose embeddings spatial-temporally to enhance the temporal coherence. Moreover, to facilitate the task of text-driven human video generation, we contribute a Fashion-Text2Video dataset with manually annotated action labels and text descriptions. Extensive experiments demonstrate that Text2Performer generates high-quality human videos (up to 512 × 256 resolution) with diverse appearances and flexible motions. Our project page is https://yumingj.github.io/projects/Text2Performer.html

Realistic human-centric rendering plays a key role in both computer vision and computer graphics. Rapid progress has been made in the algorithm aspect over the years, yet existing human-centric rendering datasets and benchmarks are rather impoverished in terms of diversity (e.g., outfit's fabric/material, body's interaction with objects, and motion sequences), which are crucial for rendering effect. Researchers are usually constrained to explore and evaluate a small set of rendering problems on current datasets, while real-world applications require methods to be robust across different scenarios. In this work, we present DNA-Rendering, a large-scale, high-fidelity repository of human performance data for neural actor rendering. DNA-Rendering presents several appealing attributes. First, our dataset contains over 1500 human subjects, 5000 motion sequences, and 67.5M frames' data volume. Upon the massive collections, we provide human subjects with grand categories of pose actions, body shapes, clothing, accessories, hairdos, and object intersection, which ranges the geometry and appearance variances from everyday life to professional occasions. Second, we provide rich assets for each subject – 2D/3D human body keypoints, foreground masks, SMPLX models, cloth/accessory materials, multi-view images, and videos. These assets boost the current method's accuracy on downstream rendering tasks. Third, we construct a professional multi-view system to capture data, which contains 60 synchronous cameras with max 4096 × 3000 resolution, 15 fps speed, and stern camera calibration steps, ensuring high-quality resources for task training and evaluation.Along with the dataset, we provide a large-scale and quantitative benchmark in full-scale, with multiple tasks to evaluate the existing progress of novel view synthesis, novel pose animation synthesis, and novel identity rendering methods. In this manuscript, we describe our DNA-Rendering effort as a revealing of new observations, challenges, and future directions to human-centric rendering. The dataset, code, and benchmarks will be publicly available at https://dna-rendering.github.io/.

In this paper, we define and study a new Cloth2Body problem which has a goal of generating 3d human body meshes from a 2D clothing image. Unlike the existing human mesh recovery problem, Cloth2Body needs to address new and emerging challenges raised by the partial observation of the input and the high diversity of the output. Indeed, there are three specific challenges. First, how to locate and pose human bodies into the clothes. Second, how to effectively estimate body shapes out of various clothing types. Finally, how to generate diverse and plausible results from a 2D clothing image. To this end, we propose an end-to-end framework that can accurately estimate 3D body mesh parameterized by pose and shape from a 2D clothing image. Along this line, we first utilize Kinematics-aware Pose Estimation to estimate body pose parameters. 3D skeleton is employed as a proxy followed by an inverse kinematics module to boost the estimation accuracy. We additionally design an adaptive depth trick to align the re-projected 3D mesh better with 2D clothing image by disentangling the effects of object size and camera extrinsic. Next, we propose Physics-informed Shape Estimation to estimate body shape parameters. 3D shape parameters are predicted based on partial body measurements estimated from RGB image, which not only improves pixel-wise human-cloth alignment, but also enables flexible user editing. Finally, we design Evolution-based pose generation method, a skeleton transplanting method inspired by genetic algorithms to generate diverse reasonable poses during inference. As shown by experimental results on both synthetic and real-world data, the proposed framework achieves state-of-the-art performance and can effectively recover natural and diverse 3D body meshes from 2D images that align well with clothing.

Human generation has achieved significant progress. Nonetheless, existing methods still struggle to synthesize specific regions such as faces and hands. We argue that the main reason is rooted in the training data. A holistic human dataset inevitably has insufficient and low-resolution information on local parts. Therefore, we propose to use multisource datasets with various resolution images to jointly learn a high-resolution human generative model. However, multi-source data inherently a) contains different parts that do not spatially align into a coherent human, and b) comes with different scales. To tackle these challenges, we propose an end-to-end framework, UnitedHuman, that empowers continuous GAN with the ability to effectively utilize multi-source data for high-resolution human generation. Specifically, 1) we design a Multi-Source Spatial Transformer that spatially aligns multi-source images to full-body space with a human parametric model. 2) Next, a continuous GAN is proposed with global-structural guidance and CutMix consistency. Patches from different datasets are then sampled and transformed to supervise the training of this scale-invariant generative model. Extensive experiments demonstrate that our model jointly learned from multi-source data achieves superior quality than those learned from a holistic dataset. Project page: https://unitedhuman.github.io/.

Existing Human NeRF methods for reconstructing 3D humans typically rely on multiple 2D images from multi-view cameras or monocular videos captured from fixed camera views. However, in real-world scenarios, human images are often captured from random camera angles, presenting challenges for high-quality 3D human reconstruction. In this paper, we propose SHERF, the first generalizable Human NeRF model for recovering animatable 3D humans from a single input image. SHERF extracts and encodes 3D human representations in canonical space, enabling rendering and animation from free views and poses. To achieve high-fidelity novel view and pose synthesis, the encoded 3D human representations should capture both global appearance and local fine-grained textures. To this end, we propose a bank of 3D-aware hierarchical features, including global, point-level, and pixel-aligned features, to facilitate informative encoding. Global features enhance the information extracted from the single input image and complement the information missing from the partial 2D observation. Point-level features provide strong clues of 3D human structure, while pixel-aligned features preserve more fine-grained details. To effectively integrate the 3D-aware hierarchical feature bank, we design a feature fusion transformer. Extensive experiments on THuman, RenderPeople, ZJU_MoCap, and HuMMan datasets demonstrate that SHERF achieves state-of-the-art performance, with better generalizability for novel view and pose synthesis. Our code is available at https://github.com/skhu101/SHERF.

LiDAR segmentation is crucial for autonomous driving perception. Recent trends favor point- or voxel-based methods as they often yield better performance than the traditional range view representation. In this work, we unveil several key factors in building powerful range view models. We observe that the "many-to-one" mapping, semantic incoherence, and shape deformation are possible impediments against effective learning from range view projections. We present RangeFormer – a full-cycle framework comprising novel designs across network architecture, data augmentation, and post-processing – that better handles the learning and processing of LiDAR point clouds from the range view. We further introduce a Scalable Training from Range view (STR) strategy that trains on arbitrary low-resolution 2D range images, while still maintaining satisfactory 3D segmentation accuracy. We show that, for the first time, a range view method is able to surpass the point, voxel, and multi-view fusion counterparts in the competing LiDAR semantic and panoptic segmentation benchmarks, i.e., SemanticKITTI, nuScenes, and ScribbleKITTI.

The robustness of 3D perception systems under natural corruptions from environments and sensors is pivotal for safety-critical applications. Existing large-scale 3D perception datasets often contain data that are meticulously cleaned. Such configurations, however, cannot reflect the reliability of perception models during the deployment stage. In this work, we present Robo3D, the first comprehensive benchmark heading toward probing the robustness of 3D detectors and segmentors under out-of-distribution scenarios against natural corruptions that occur in real-world environments. Specifically, we consider eight corruption types stemming from severe weather conditions, external disturbances, and internal sensor failure. We uncover that, although promising results have been progressively achieved on standard benchmarks, state-of-the-art 3D perception models are at risk of being vulnerable to corruptions. We draw key observations on the use of data representations, augmentation schemes, and training strategies, that could severely affect the model's performance. To pursue better robustness, we propose a density-insensitive training framework along with a simple flexible voxelization strategy to enhance the model resiliency. We hope our benchmark and approach could inspire future research in designing more robust and reliable 3D perception models. Our robustness benchmark suite is publicly available1.

We aim to address a significant but understudied problem in the anime industry, namely the inbetweening of cartoon line drawings. Inbetweening involves generating intermediate frames between two black-and-white line drawings and is a time-consuming and expensive process that can benefit from automation. However, existing frame interpolation methods that rely on matching and warping whole raster images are unsuitable for line inbetweening and often produce blurring artifacts that damage the intricate line structures. To preserve the precision and detail of the line drawings, we propose a new approach, AnimeInbet, which geometrizes raster line drawings into graphs of endpoints and reframes the inbetweening task as a graph fusion problem with vertex repositioning. Our method can effectively capture the sparsity and unique structure of line drawings while preserving the details during inbetweening. This is made possible via our novel modules, i.e., vertex geometric embedding, a vertex correspondence Transformer, an effective mechanism for vertex repositioning and a visibility predictor. To train our method, we introduce Mixamo-Line240, a new dataset of line drawings with ground truth vectorization and matching labels. Our experiments demonstrate that AnimeInbet synthesizes high-quality, clean, and complete intermediate line drawings, outperforming existing methods quantitatively and qualitatively, especially in cases with large motions. Data and code are available at https://github.com/lisiyao21/AnimeInbet.

Neural Radiance Field (NeRF) significantly degrades when only a limited number of views are available. To complement the lack of 3D information, depth-based models, such as DSNeRF and MonoSDF, explicitly assume the availability of accurate depth maps of multiple views. They linearly scale the accurate depth maps as supervision to guide the predicted depth of few-shot NeRFs. However, accurate depth maps are difficult and expensive to capture due to wide-range depth distances in the wild. This work presents a new Sparse-view NeRF (SparseNeRF) framework that exploits depth priors from real-world inaccurate observations. The inaccurate depth observations are either from pre-trained depth models or coarse depth maps of consumer-level depth sensors. Since coarse depth maps are not strictly scaled to the ground-truth depth maps, we propose a simple yet effective constraint, a local depth ranking method, on NeRFs such that the expected depth ranking of the NeRF is consistent with that of the coarse depth maps in local patches. To preserve the spatial continuity of the estimated depth of NeRF, we further propose a spatial continuity constraint to encourage the consistency of the expected depth continuity of NeRF with coarse depth maps. Surprisingly, with simple depth ranking constraints, SparseNeRF outperforms all state-of-the-art few-shot NeRF methods (including depth-based models) on standard LLFF and DTU datasets. Moreover, we collect a new dataset NVS-RGBD that contains real-world depth maps from Azure Kinect, ZED 2, and iPhone 13 Pro. Extensive experiments on NVS-RGBD dataset also validate the superiority and generaliz-ability of SparseNeRF. Code and dataset are available at https://sparsenerf.github.io/.

Synthetic data has emerged as a promising source for 3D human research as it offers low-cost access to large-scale human datasets. To advance the diversity and annotation quality of human models, we introduce a new synthetic dataset, SynBody, with three appealing features: 1) a clothed parametric human model that can generate a diverse range of subjects; 2) the layered human representation that naturally offers high-quality 3D annotations to support multiple tasks; 3) a scalable system for producing realistic data to facilitate real-world tasks. The dataset comprises 1.2M images with corresponding accurate 3D annotations, covering 10,000 human body models, 1,187 actions, and various viewpoints. The dataset includes two subsets for human pose and shape estimation as well as human neural rendering. Extensive experiments on SynBody indicate that it substantially enhances both SMPL and SMPL-X estimation. Furthermore, the incorporation of layered annotations offers a valuable training resource for investigating the Human Neural Radiance Fields(NeRF).

Recent advances in face manipulation using StyleGAN have produced impressive results. However, StyleGAN is inherently limited to cropped aligned faces at a fixed image resolution it is pre-trained on. In this paper, we propose a simple and effective solution to this limitation by using dilated convolutions to rescale the receptive fields of shallow layers in StyleGAN, without altering any model parameters. This allows fixed-size small features at shallow layers to be extended into larger ones that can accommodate variable resolutions, making them more robust in characterizing unaligned faces. To enable real face inversion and manipulation, we introduce a corresponding encoder that provides the first-layer feature of the extended StyleGAN in addition to the latent style code. We validate the effectiveness of our method using unaligned face inputs of various resolutions in a diverse set of face manipulation tasks, including facial attribute editing, super-resolution, sketch/mask-to-face translation, and face toonification. Project page https://www.mmlab-ntu.com/project/styleganex

3D human motion generation is crucial for creative industry. Recent advances rely on generative models with domain knowledge for text-driven motion generation, leading to substantial progress in capturing common motions. However, the performance on more diverse motions remains unsatisfactory. In this work, we propose ReMoDiffuse, a diffusion-model-based motion generation framework that integrates a retrieval mechanism to refine the denoising process. ReMoDiffuse enhances the generalizability and diversity of text-driven motion generation with three key designs: 1) Hybrid Retrieval finds appropriate references from the database in terms of both semantic and kinematic similarities. 2) Semantic-Modulated Transformer selectively absorbs retrieval knowledge, adapting to the difference between retrieved samples and the target motion sequence. 3) Condition Mixture better utilizes the retrieval database during inference, overcoming the scale sensitivity in classifier-free guidance. Extensive experiments demonstrate that ReMoDiffuse outperforms state-of-the-art methods by balancing both text-motion consistency and motion quality, especially for more diverse motion generation. Project page: https://mingyuan-zhang.github.io/projects/ReMoDiffuse.html

In this paper, we address the challenging problem of 3D toonification, which involves transferring the style of an artistic domain onto a target 3D face with stylized geometry and texture. Although fine-tuning a pre-trained 3D GAN on the artistic domain can produce reasonable performance, this strategy has limitations in the 3D domain. In particular, fine-tuning can deteriorate the original GAN latent space, which affects subsequent semantic editing, and requires independent optimization and storage for each new style, limiting flexibility and efficient deployment. To overcome these challenges, we propose DeformToon3d, an effective toonification framework tailored for hierarchical 3D GAN. Our approach decomposes 3D toonification into subproblems of geometry and texture stylization to better preserve the original latent space. Specifically, we devise a novel StyleField that predicts conditional 3D deformation to align a real-space NeRF to the style space for geometry stylization. Thanks to the StyleField formulation, which already handles geometry stylization well, texture stylization can be achieved conveniently via adaptive style mixing that injects information of the artistic domain into the decoder of the pre-trained 3D GAN. Due to the unique design, our method enables flexible style degree control and shape-texture-specific style swap. Furthermore, we achieve efficient training without any real-world 2D-3D training pairs but proxy samples synthesized from off-the-shelf 2D toonification models. Code is released at https://github.com/junzhezhang/DeformToon3D.

Network binarization emerges as one of the most promising compression approaches offering extraordinary computation and memory savings by minimizing the bit-width. However, recent research has shown that applying existing binarization algorithms to diverse tasks, architectures, and hardware in realistic scenarios is still not straightforward. Common challenges of binarization, such as accuracy degradation and efficiency limitation, suggest that its attributes are not fully understood. To close this gap, we present BiBench, a rigorously designed benchmark with in-depth analysis for network binarization. We first carefully scrutinize the requirements of binarization in the actual production and define evaluation tracks and metrics for a comprehensive and fair investigation. Then, we evaluate and analyze a series of milestone binarization algorithms that function at the operator level and with extensive influence. Our benchmark reveals that 1) the binarized operator has a crucial impact on the performance and deployability of binarized networks; 2) the accuracy of binarization varies significantly across different learning tasks and neural architectures; 3) binarization has demonstrated promising efficiency potential on edge devices despite the limited hardware support. The results and analysis also lead to a promising paradigm for accurate and efficient binarization. We believe that BiBench will contribute to the broader adoption of binarization and serve as a foundation for future research. The code for our BiBench is released https://github.com/htqin/BiBench .

Human-machine interaction (HMI) is much important in factories, and most HMI ways are contact which arise safety and health issues. To avoid such problems, contactless HMI ways such as in-air hand gesture recognition (HGR) via Wi-Fi or radar are widely studied by both academia and industry. However, these RF-based methods are not very appropriate for the industry because of the electromagnetic interference. As for visible light sensing, it is free of electromagnetic radiation and can reuse the existing devices, e.g., lamps on machines, hence utilizing visible light to realize HGR is a good solution for HMI in factories. The current visible-light-enabled HGR (VL-HGR) methods using deep learning algorithms are all supervised, which increases the cost of manual labeling and further hinders the industrial applications of VL-HGR. To this end, we propose SemiGest, a semi-supervised learning (SSL) method for VL-HGR, to facilitate the applications of VL-HGR in industry. The system prototype is built on a table lamp to mimic the lamp on a machine emitting lights at four distinct carrier frequencies, and the lights reflected by hands are collected by a receiver. SemiGest utilizes the variation and correlation of the lights to realize HGR with an SSL algorithm using only a small amount of labeled data and lots of unlabeled data. Furthermore, the SSL algorithm is designed not only for the visible light data but also can be generalized to other time-series data in the industry. To confirm the effectiveness and robustness of SemiGest, we perform various experiments to show the potential for practical implementation in the industry.

We present PrimDiffusion, the first diffusion-based framework for 3D human generation. Devising diffusion models for 3D human generation is difficult due to the intensive computational cost of 3D representations and the articulated topology of 3D humans. To tackle these challenges, our key insight is operating the denoising diffusion process directly on a set of volumetric primitives, which models the human body as a number of small volumes with radiance and kinematic information. This volumetric primitives representation marries the capacity of volumetric representations with the efficiency of primitive-based rendering. Our PrimDiffusion framework has three appealing properties: **1)** compact and expressive parameter space for the diffusion model, **2)** flexible representation that incorporates human prior, and **3)** decoder-free rendering for efficient novel-view and novel-pose synthesis. Extensive experiments validate that PrimDiffusion outperforms state-of-the-art methods in 3D human generation. Notably, compared to GAN-based methods, our PrimDiffusion supports real-time rendering of high-quality 3D humans at a resolution of $512\times512$ once the denoising process is done. We also demonstrate the flexibility of our framework on training-free conditional generation such as texture transfer and 3D inpainting.

Expressive human pose and shape estimation (EHPS) unifies body, hands, and face motion capture with numerous applications. Despite encouraging progress, current state-of-the-art methods still depend largely on a confined set of training datasets. In this work, we investigate scaling up EHPS towards the first generalist foundation model (dubbed SMPLer-X), with up to ViT-Huge as the backbone and training with up to 4.5M instances from diverse data sources. With big data and the large model, SMPLer-X exhibits strong performance across diverse test benchmarks and excellent transferability to even unseen environments. 1) For the data scaling, we perform a systematic investigation on 32 EHPS datasets, including a wide range of scenarios that a model trained on any single dataset cannot handle. More importantly, capitalizing on insights obtained from the extensive benchmarking process, we optimize our training scheme and select datasets that lead to a significant leap in EHPS capabilities. 2) For the model scaling, we take advantage of vision transformers to study the scaling law of model sizes in EHPS. Moreover, our finetuning strategy turn SMPLer-X into specialist models, allowing them to achieve further performance boosts. Notably, our foundation model SMPLer-X consistently delivers state-of-the-art results on seven benchmarks such as AGORA (107.2 mm NMVE), UBody (57.4 mm PVE), EgoBody (63.6 mm PVE), and EHF (62.3 mm PVE without finetuning).

Visual reasoning requires multimodal perception and commonsense cognition of the world. Recently, multiple vision-language models (VLMs) have been proposed with excellent commonsense reasoning ability in various domains. However, how to harness the collective power of these complementary VLMs is rarely explored. Existing methods like ensemble still struggle to aggregate these models with the desired higher-order communications. In this work, we propose Cola, a novel paradigm that coordinates multiple VLMs for visual reasoning. Our key insight is that a large language model (LLM) can efficiently coordinate multiple VLMs by facilitating natural language communication that leverages their distinct and complementary capabilities. Extensive experiments demonstrate that our instruction tuning variant, Cola-FT, achieves state-of-the-art performance on visual question answering (VQA), outside knowledge VQA, visual entailment, and visual spatial reasoning tasks. Moreover, we show that our in-context learning variant, Cola-Zero, exhibits competitive performance in zero and few-shot settings, without finetuning. Through systematic ablation studies and visualizations, we validate that a coordinator LLM indeed comprehends the instruction prompts as well as the separate functionalities of VLMs; it then coordinates them to enable impressive visual reasoning capabilities.

Recent advancements in vision foundation models (VFMs) have opened up new possibilities for versatile and efficient visual perception. In this work, we introduce Seal, a novel framework that harnesses VFMs for segmenting diverse automotive point cloud sequences. Seal exhibits three appealing properties: i) Scalability: VFMs are directly distilled into point clouds, obviating the need for annotations in either 2D or 3D during pretraining. ii) Consistency: Spatial and temporal relationships are enforced at both the camera-to-LiDAR and point-to-segment regularization stages, facilitating cross-modal representation learning. iii) Generalizability: Seal enables knowledge transfer in an off-the-shelf manner to downstream tasks involving diverse point clouds, including those from real/synthetic, low/high-resolution, large/small-scale, and clean/corrupted datasets. Extensive experiments conducted on eleven different point cloud datasets showcase the effectiveness and superiority of Seal. Notably, Seal achieves a remarkable 45.0% mIoU on nuScenes after linear probing, surpassing random initialization by 36.9% mIoU and outperforming prior arts by 6.1% mIoU. Moreover, Seal demonstrates significant performance gains over existing methods across 20 different few-shot fine-tuning tasks on all eleven tested point cloud datasets. The code is available at this link.

Synthesizing high-fidelity head avatars is a central problem for computer vision and graphics. While head avatar synthesis algorithms have advanced rapidly, the best ones still face great obstacles in real-world scenarios. One of the vital causes is the inadequate datasets -- 1) current public datasets can only support researchers to explore high-fidelity head avatars in one or two task directions; 2) these datasets usually contain digital head assets with limited data volume, and narrow distribution over different attributes, such as expressions, ages, and accessories. In this paper, we present RenderMe-360, a comprehensive 4D human head dataset to drive advance in head avatar algorithms across different scenarios. It contains massive data assets, with 243+ million complete head frames and over 800k video sequences from 500 different identities captured by multi-view cameras at 30 FPS. It is a large-scale digital library for head avatars with three key attributes: 1) High Fidelity: all subjects are captured in 360 degrees via 60 synchronized, high-resolution 2K cameras. 2) High Diversity: The collected subjects vary from different ages, eras, ethnicities, and cultures, providing abundant materials with distinctive styles in appearance and geometry. Moreover, each subject is asked to perform various dynamic motions, such as expressions and head rotations, which further extend the richness of assets. 3) Rich Annotations: the dataset provides annotations with different granularities: cameras' parameters, background matting, scan, 2D/3D facial landmarks, FLAME fitting, and text description. Based on the dataset, we build a comprehensive benchmark for head avatar research, with 16 state-of-the-art methods performed on five main tasks: novel view synthesis, novel expression synthesis, hair rendering, hair editing, and talking head generation. Our experiments uncover the strengths and flaws of state-of-the-art methods. RenderMe-360 opens the door for future exploration in modern head avatars. All of the data, code, and models will be publicly available at https://renderme-360.github.io/.

Whole-body pose and shape estimation aims to jointly predict different behaviors (e.g., pose, hand gesture, facial expression) of the entire human body from a monocular image. Existing methods often exhibit suboptimal performance due to the complexity of in-the-wild scenarios. We argue that the prediction accuracy of these models is significantly affected by the quality of the _bounding box_, e.g., scale, alignment. The natural discrepancy between the ideal bounding box annotations and model detection results is particularly detrimental to the performance of whole-body pose and shape estimation. In this paper, we propose a novel framework to enhance the robustness of whole-body pose and shape estimation. Our framework incorporates three new modules to address the above challenges from three perspectives: (1) a **Localization Module** enhances the model's awareness of the subject's location and semantics within the image space; (2) a **Contrastive Feature Extraction Module** encourages the model to be invariant to robust augmentations by incorporating a contrastive loss and positive samples; (3) a **Pixel Alignment Module** ensures the reprojected mesh from the predicted camera and body model parameters are more accurate and pixel-aligned. We perform comprehensive experiments to demonstrate the effectiveness of our proposed framework on body, hands, face and whole-body benchmarks.

Generating animation of physics-based characters with intuitive control has long been a desirable task with numerous applications. However, generating physically simulated animations that reflect high-level human instructions remains a difficult problem due to the complexity of physical environments and the richness of human language. In this paper, we present $\textbf{InsActor}$, a principled generative framework that leverages recent advancements in diffusion-based human motion models to produce instruction-driven animations of physics-based characters. Our framework empowers InsActor to capture complex relationships between high-level human instructions and character motions by employing diffusion policies for flexibly conditioned motion planning. To overcome invalid states and infeasible state transitions in planned motions, InsActor discovers low-level skills and maps plans to latent skill sequences in a compact latent space. Extensive experiments demonstrate that InsActor achieves state-of-the-art results on various tasks, including instruction-driven motion generation and instruction-driven waypoint heading. Notably, the ability of InsActor to generate physically simulated animations using high-level human instructions makes it a valuable tool, particularly in executing long-horizon tasks with a rich set of instructions. Our project page is available at [jiawei-ren.github.io/projects/insactor/index.html](https://jiawei-ren.github.io/projects/insactor/index.html)

We are living in a three-dimensional space while moving forward through a fourth dimension: time. To allow artificial intelligence to develop a comprehensive understanding of such a 4D environment, we introduce **4D Panoptic Scene Graph (PSG-4D)**, a new representation that bridges the raw visual data perceived in a dynamic 4D world and high-level visual understanding. Specifically, PSG-4D abstracts rich 4D sensory data into nodes, which represent entities with precise location and status information, and edges, which capture the temporal relations. To facilitate research in this new area, we build a richly annotated PSG-4D dataset consisting of 3K RGB-D videos with a total of 1M frames, each of which is labeled with 4D panoptic segmentation masks as well as fine-grained, dynamic scene graphs. To solve PSG-4D, we propose PSG4DFormer, a Transformer-based model that can predict panoptic segmentation masks, track masks along the time axis, and generate the corresponding scene graphs via a relation component. Extensive experiments on the new dataset show that our method can serve as a strong baseline for future research on PSG-4D. In the end, we provide a real-world application example to demonstrate how we can achieve dynamic scene understanding by integrating a large language model into our PSG-4D system.

Text-driven motion generation has achieved substantial progress with the emergence of diffusion models. However, existing methods still struggle to generate complex motion sequences that correspond to fine-grained descriptions, depicting detailed and accurate spatio-temporal actions.This lack of fine controllability limits the usage of motion generation to a larger audience. To tackle these challenges, we present FineMoGen, a diffusion-based motion generation and editing framework that can synthesize fine-grained motions, with spatial-temporal composition to the user instructions. Specifically, FineMoGen builds upon diffusion model with a novel transformer architecture dubbed Spatio-Temporal Mixture Attention SAMI. SAMI optimizes the generation of the global attention template from two perspectives: 1) explicitly modeling the constraints of spatio-temporal composition; and 2) utilizing sparsely-activated mixture-of-experts to adaptively extract fine-grained features. To facilitate a large-scale study on this new fine-grained motion generation task, we contribute the HuMMan-MoGen dataset, which consists of 2,968 videos and 102,336 fine-grained spatio-temporal descriptions. Extensive experiments validate that FineMoGen exhibits superior motion generation quality over state-of-the-art methods. Notably, FineMoGen further enables zero-shot motion editing capabilities with the aid of modern large language models (LLM), which faithfully manipulates motion sequences with fine-grained instructions.

Large vision models with billions of parameters and trained on broad data have great potential in numerous downstream applications. However, these models are typically difficult to adapt due to their large parameter size and sometimes lack of accesss to their weights---entities able to develop large vision models often provide APIs only. In this paper, we study how to better utilize large vision models through the lens of in-context learning, a concept that has been well-known in natural language processing but has only been studied very recently in computer vision. In-context learning refers to the ability to perform inference on tasks never seen during training by simply conditioning on in-context examples (i.e., input-output pairs) without updating any internal model parameters. To demystify in-context learning in computer vision, we conduct an extensive research and identify a critical problem: downstream performance is highly sensitivie to the choice of visual in-context examples. To address this problem, we propose a prompt retrieval framework specifically for large vision models, allowing the selection of in-context examples to be fully automated. Concretely, we provide two implementations: (i) an unsupervised prompt retrieval method based on nearest example search using an off-the-shelf model, and (ii) a supervised prompt retrieval method, which trains a neural network to choose examples that directly maximize in-context learning performance. Both methods do not require access to the internal weights of large vision models. Our results demonstrate that our methods can bring non-trivial improvements to visual in-context learning in comparison to the commonly-used random selection. Code and models will be released.

3D content creation from a single image is a long-standing yet highly desirable task. Recent advances introduce 2D diffusion priors, yielding reasonable results. However, existing methods are not hyper-realistic enough for post-generation usage, as users cannot view, render and edit the resulting 3D content from a full range. To address these challenges, we introduce HyperDreamer with several key designs and appealing properties: 1) Full-range viewable: 360° mesh modeling with high-resolution textures enables the creation of visually compelling 3D models from a full range of observation points. 2) Full-range renderable: Fine-grained semantic segmentation and data-driven priors are incorporated as guidance to learn reasonable albedo, roughness, and specular properties of the materials, enabling semantic-aware arbitrary material estimation. 3) Full-range editable: For a generated model or their own data, users can interactively select any region via a few clicks and efficiently edit the texture with text-based guidance. Extensive experiments demonstrate the effectiveness of HyperDreamer in modeling region-aware materials with high-resolution textures and enabling user-friendly editing. We believe that HyperDreamer holds promise for advancing 3D content creation and finding applications in various domains.

Large text-to-image diffusion models have exhibited impressive proficiency in generating high-quality images. However, when applying these models to video domain, ensuring temporal consistency across video frames remains a formidable challenge. This paper proposes a novel zero-shot text-guided video-to-video translation framework to adapt image models to videos. The framework includes two parts: key frame translation and full video translation. The first part uses an adapted diffusion model to generate key frames, with hierarchical cross-frame constraints applied to enforce coherence in shapes, textures and colors. The second part propagates the key frames to other frames with temporal-aware patch matching and frame blending. Our framework achieves global style and local texture temporal consistency at a low cost (without re-training or optimization). The adaptation is compatible with existing image diffusion techniques, allowing our framework to take advantage of them, such as customizing a specific subject with LoRA, and introducing extra spatial guidance with ControlNet. Extensive experimental results demonstrate the effectiveness of our proposed framework over existing methods in rendering high-quality and temporally-coherent videos. Code is available at our project page: https://www.mmlab-ntu.com/project/rerender/

Visual tracking has made significant improvements in the past few decades. Most existing state-of-the-art trackers 1) merely aim for performance in ideal conditions while overlooking the real-world conditions; 2) adopt the tracking-by-detection paradigm, neglecting rich temporal contexts; 3) only integrate the temporal information into the template, where temporal contexts among consecutive frames are far from being fully utilized. To handle those problems, we propose a two-level framework (TCTrack) that can exploit temporal contexts efficiently. Based on it, we propose a stronger version for real-world visual tracking, i.e., TCTrack++. It boils down to two levels: features and similarity maps. Specifically, for feature extraction, we propose an attention-based temporally adaptive convolution to enhance the spatial features using temporal information, which is achieved by dynamically calibrating the convolution weights. For similarity map refinement, we introduce an adaptive temporal transformer to encode the temporal knowledge efficiently and decode it for the accurate refinement of the similarity map. To further improve the performance, we additionally introduce a curriculum learning strategy. Also, we adopt online evaluation to measure performance in real-world conditions. Exhaustive experiments on 8 well-known benchmarks demonstrate the superiority of TCTrack++. Real-world tests directly verify that TCTrack++ can be readily used in real-world applications.

In this work, we present SceneDreamer, an unconditional generative model for unbounded 3D scenes, which synthesizes large-scale 3D landscapes from random noise. Our framework is learned from in-the-wild 2D image collections only, without any 3D annotations. At the core of SceneDreamer is a principled learning paradigm comprising: 1) an efficient yet expressive 3D scene representation, 2) a generative scene parameterization, and 3) an effective renderer that can leverage the knowledge from 2D images. Our approach begins with an efficient bird's-eye-view (BEV) representation generated from simplex noise, which includes a height field for surface elevation and a semantic field for detailed scene semantics. This BEV scene representation enables: 1) representing a 3D scene with quadratic complexity, 2) disentangled geometry and semantics, and 3) efficient training. Moreover, we propose a novel generative neural hash grid to parameterize the latent space based on 3D positions and scene semantics, aiming to encode generalizable features across various scenes. Lastly, a neural volumetric renderer, learned from 2D image collections through adversarial training, is employed to produce photorealistic images. Extensive experiments demonstrate the effectiveness of SceneDreamer and superiority over state-of-the-art methods in generating vivid yet diverse unbounded 3D worlds.

Reference-based Super-Resolution (Ref-SR) has recently emerged as a promising paradigm to enhance a low-resolution (LR) input image or video by introducing an additional high-resolution (HR) reference image. Existing Ref-SR methods mostly rely on implicit correspondence matching to borrow HR textures from reference images to compensate for the information loss in input images. However, performing local transfer is difficult because of two gaps between input and reference images: the transformation gap (e.g.scale and rotation) and the resolution gap (e.g.HR and LR). To tackle these challenges, we propose$C^{2}$-Matching in this work, which performs explicit robust matching crossing transformation and resolution. 1) To bridge the transformation gap, we propose a contrastive correspondence network, which learns transformation-robust correspondences using augmented views of the input image. 2) To address the resolution gap, we adopt teacher-student correlation distillation, which distills knowledge from the easier HR-HR matching to guide the more ambiguous LR-HR matching. 3) Finally, we design a dynamic aggregation module to address the potential misalignment issue between input images and reference images. In addition, to faithfully evaluate the performance of Reference-based Image Super-Resolution (Ref Image SR) under a realistic setting, we contribute the Webly-Referenced SR (WR-SR) dataset, mimicking the practical usage scenario. We also extend$C^{2}$-Matching to Reference-based Video Super-Resolution (Ref VSR) task, where an image taken in a similar scene serves as the HR reference image. Extensive experiments demonstrate that our proposed$C^{2}$-Matching significantly outperforms state of the arts by up to 0.7dB on the standard CUFED5 benchmark and also boosts the performance of video super-resolution by incorporating the$C^{2}$-Matching component into Video SR pipelines. Notably,$C^{2}$-Matching also shows great generalizability on WR-SR dataset as well as robustness across large scale and rotation transformations. Codes and datasets are available athttps://github.com/yumingj/C2-Matching.

Real-scanned point clouds are often incomplete due to viewpoint, occlusion, and noise, which hampers 3D geometric modeling and perception. Existing point cloud completion methods tend to generate global shape skeletons and hence lack fine local details. Furthermore, they mostly learn a deterministic partial-to-complete mapping, but overlook structural relations in man-made objects. To tackle these challenges, this paper proposes a variational framework, Variational Relational point Completion network (VRCNet) with two appealing properties: 1) Probabilistic Modeling. In particular, we propose a dual-path architecture to enable principled probabilistic modeling across partial and complete clouds. One path consumes complete point clouds for reconstruction by learning a point VAE. The other path generates complete shapes for partial point clouds, whose embedded distribution is guided by distribution obtained from the reconstruction path during training. 2) Relational Enhancement. Specifically, we carefully design point self-attention kernel and point selective kernel module to exploit relational point features, which refines local shape details conditioned on the coarse completion. In addition, we contribute multi-view partial point cloud datasets (MVP and MVP-40 dataset) containing over 200,000 high-quality scans, which render partial 3D shapes from 26 uniformly distributed camera poses for each 3D CAD model. Extensive experiments demonstrate that VRCNet outperforms state-of-the-art methods on all standard point cloud completion benchmarks. Notably, VRCNet shows great generalizability and robustness on real-world point cloud scans. Moreover, we can achieve robust 3D classification for partial point clouds with the help of VRCNet, which can highly increase classification accuracy.

Our proposed music-to-dance framework, Bailando++, addresses the challenges of driving 3D characters to dance in a way that follows the constraints of choreography norms and maintains temporal coherency with different music genres. Bailando++ consists of two components: a choreographic memory that learns to summarize meaningful dancing units from 3D pose sequences, and an actor-critic Generative Pre-trained Transformer (GPT) that composes these units into a fluent dance coherent to the music. In particular, to synchronize the diverse motion tempos and music beats, we introduce an actor-critic-based reinforcement learning scheme to the GPT with a novel beat-align reward function. Additionally, we consider learning human dance poses in the rotation domain to avoid body distortions incompatible with human morphology, and introduce a musical contextual encoding to allow the motion GPT to grasp longer-term patterns of music. Our experiments on the standard benchmark show that Bailando++ achieves state-of-the-art performance both qualitatively and quantitatively, with the added benefit of the unsupervised discovery of human-interpretable dancing-style poses in the choreographic memory.

Recent advances in deep learning have witnessed many successful unsupervised image-to-image translation models that learn correspondences between two visual domains without paired data. However, it is still a great challenge to build robust mappings between various domains especially for those with drastic visual discrepancies. In this paper, we introduce a novel versatile framework, Generative Prior-guided UNsupervised Image-to-image Translation (GP-UNIT), that improves the quality, applicability and controllability of the existing translation models. The key idea of GP-UNIT is to distill the generative prior from pre-trained class-conditional GANs to build coarse-level cross-domain correspondences, and to apply the learned prior to adversarial translations to excavate fine-level correspondences. With the learned multi-level content correspondences, GP-UNIT is able to perform valid translations between both close domains and distant domains. For close domains, GP-UNIT can be conditioned on a parameter to determine the intensity of the content correspondences during translation, allowing users to balance between content and style consistency. For distant domains, semi-supervised learning is explored to guide GP-UNIT to discover accurate semantic correspondences that are hard to learn solely from the appearance. We validate the superiority of GP-UNIT over state-of-the-art translation models in robust, high-quality and diversified translations between various domains through extensive experiments.

Generalization to out-of-distribution (OOD) data is a capability natural to humans yet challenging for machines to reproduce. This is because most learning algorithms strongly rely on the i.i.d. assumption on source/target data, which is often violated in practice due to domain shift. Domain generalization (DG) aims to achieve OOD generalization by using only source data for model learning. Over the last ten years, research in DG has made great progress, leading to a broad spectrum of methodologies, e.g., those based on domain alignment, meta-learning, data augmentation, or ensemble learning, to name a few; DG has also been studied in various application areas including computer vision, speech recognition, natural language processing, medical imaging, and reinforcement learning. In this paper, for the first time a comprehensive literature review in DG is provided to summarize the developments over the past decade. Specifically, we first cover the background by formally defining DG and relating it to other relevant fields like domain adaptation and transfer learning. Then, we conduct a thorough review into existing methods and theories. Finally, we conclude this survey with insights and discussions on future research directions.

The task of audiovisual sound source localization has been well studied under constrained scenes, where the audio recordings are clean. However, in real world scenarios, audios are usually contaminated by off screen sound and background noise. They will interfere with the procedure of identifying desired sources and building visual sound connections, making previous studies nonapplicable. In this work, we propose the Interference Eraser (IEr) framework, which tackles the problem of audiovisual sound source localization in the wild. The key idea is to eliminate the interference by redefining and carving discriminative audio representations. Specifically, we observe that the previous practice of learning only a single audio representation is insufficient due to the additive nature of audio signals. We thus extend the audio representation with our Audio Instance Identifier module, which clearly distinguishes sounding instances when audio signals of different volumes are unevenly mixed. Then we erase the influence of the audible but off screen sounds and the silent but visible objects by a Cross modal Referrer module with cross modality distillation. Quantitative and qualitative evaluations demonstrate that our framework achieves superior results on sound localization tasks, especially under real world scenarios.

Multiple modalities can provide rich semantic information; and exploiting such information will normally lead to better performance compared with the single-modality counterpart. However, it is not easy to devise an effective cross-modal fusion structure due to the variations of feature dimensions and semantics, especially when the inputs even come from different sensors, as in the field of audio-visual learning. In this work, we propose SepFusion, a novel framework that can smoothly produce optimal fusion structures for visual-sound separation. The framework is composed of two components, namely the model generator and the evaluator. To construct the generator, we devise a lightweight architecture space that can adapt to different input modalities. In this way, we can easily obtain audio-visual fusion structures according to our demands. For the evaluator, we adopt the idea of neural architecture search to select superior networks effectively. This automatic process can significantly save human efforts while achieving competitive performances. Moreover, since our SepFusion provides a series of strong models, we can utilize the model family for broader applications, such as further promoting performance via model assembly, or providing suitable architectures for the separation of certain instrument classes. These potential applications further enhance the competitiveness of our approach.

Temporal contexts among consecutive frames are far from being fully utilized in existing visual trackers. In this work, we present TCTrack11https://github.com/vision4robotics/TCTrack, a comprehensive framework to fully exploit temporal contexts for aerial tracking. The temporal contexts are incorporated at two levels: the extraction of features and the refinement of similarity maps. Specifically, for feature extraction, an online temporally adaptive convolution is proposed to enhance the spatial features using temporal information, which is achieved by dynamically calibrating the convolution weights according to the previous frames. For similarity map refinement, we propose an adaptive temporal transformer, which first effectively encodes temporal knowledge in a memory-efficient way, before the temporal knowledge is decoded for accurate adjustment of the similarity map. TCTrack is effective and efficient: evaluation on four aerial tracking benchmarks shows its impressive performance; real-world UAV tests show its high speed of over 27 FPS on NVIDIA Jetson AGX Xavier.

Human-centric perception plays a vital role in vision and graphics. But their data annotations are prohibitively expensive. Therefore, it is desirable to have a versatile pretrain model that serves as a foundation for data-efficient downstream tasks transfer. To this end, we propose the Human-Centric Multi-Modal Contrastive Learning framework HCMoCo that leverages the multi-modal nature of human data (e.g. RGB, depth, 2D keypoints) for effective representation learning. The objective comes with two main challenges: dense pre-train for multi-modality data, efficient usage of sparse human priors. To tackle the challenges, we design the novel Dense Intra-sample Contrastive Learning and Sparse Structure-aware Contrastive Learning targets by hierarchically learning a modal-invariant latent space featured with continuous and ordinal feature distribution and structure-aware semantic consistency. HCMoCo provides pre-train for different modalities by combining heterogeneous datasets, which allows efficient usage of existing task-specific human data. Extensive experiments on four downstream tasks of different modalities demonstrate the effectiveness of HCMoCo, especially under data-efficient settings (7.16% and 12% improvement on DensePose Estimation and Human Parsing). Moreover, we demonstrate the versatility of HCMoCo by exploring cross-modality super-vision and missing-modality inference, validating its strong ability in cross-modal association and reasoning. Codes are available at https://github.com/hongfz16/HCMoCo.

Data imbalance exists ubiquitously in real-world visual regressions, e.g., age estimation and pose estimation, hurting the model's generalizability and fairness. Thus, im-balanced regression gains increasing research attention recently. Compared to imbalanced classification, imbalanced regression focuses on continuous labels, which can be boundless and high-dimensional and hence more challenging. In this work, we identify that the widely used Mean Square Error (MSE) loss function can be ineffective in imbalanced regression. We revisit MSE from a statistical view and propose a novel loss function, Balanced MSE, to accommodate the imbalanced training label distribution. We further design multiple implementations of Balanced MSE to tackle different real-world scenarios, particularly including the one that requires no prior knowledge about the training label distribution. Moreover, to the best of our knowledge, Balanced MSE is the first general solution to high-dimensional imbalanced regression in modern context. Extensive experiments on both synthetic and three real-world benchmarks demonstrate the effectiveness of Balanced MSE. Code and models are available at github.com/jiawei-ren/BalancedMSE.

Driving 3D characters to dance following a piece of music is highly challenging due to the spatial constraints applied to poses by choreography norms. In addition, the generated dance sequence also needs to maintain temporal coherency with different music genres. To tackle these challenges, we propose a novel music-to-dance framework, Bailando, with two powerful components: 1) a choreographic memory that learns to summarize meaningful dancing units from 3D pose sequence to a quantized codebook, 2) an actor-critic Generative Pre-trained Transformer (GPT) that composes these units to a fluent dance coherent to the music. With the learned choreographic memory, dance generation is realized on the quantized units that meet high choreography standards, such that the generated dancing sequences are confined within the spatial constraints. To achieve synchronized alignment between diverse motion tempos and music beats, we introduce an actor-critic-based reinforcement learning scheme to the GPT with a newly-designed beat-align reward function. Extensive experiments on the standard benchmark demonstrate that our proposed framework achieves state-of-the-art performance both qualitatively and quantitatively. Notably, the learned choreographic memory is shown to discover human-interpretable dancing-style poses in an unsupervised manner. Code and video demo are available at https://github.com/lisiyao21/Bailando/

Recent studies on StyleGAN show high performance on artistic portrait generation by transfer learning with limited data. In this paper, we explore more challenging exemplar-based high-resolution portrait style transfer by introducing a novel DualStyleGAN with flexible control of dual styles of the original face domain and the extended artistic portrait domain. Different from StyleGAN, DualStyleGAN provides a natural way of style transfer by characterizing the content and style of a portrait with an intrinsic style path and a new extrinsic style path, respectively. The del-icately designed extrinsic style path enables our model to modulate both the color and complex structural styles hierarchically to precisely pastiche the style example. Furthermore, a novel progressive fine-tuning scheme is introduced to smoothly transform the generative space of the model to the target domain, even with the above modifications on the network architecture. Experiments demonstrate the superiority of DualStyleGAN over state-of-the-art methods in high-quality portrait style transfer and flexible stylecontrol. Code is available at https://github.com/williamyang1991/DualStyleGAN.

Unsupervised image-to-image translation aims to learn the translation between two visual domains without paired data. Despite the recent progress in image translation models, it remains challenging to build mappings between complex domains with drastic visual discrepancies. In this work, we present a novel framework, Generative Priorguided UNsupervised Image-to-image Translation (GP-UNIT), to improve the overall quality and applicability of the translation algorithm. Our key insight is to leverage the generative prior from pre-trained class-conditional GANs (e.g., BigGAN) to learn rich content correspondences across various domains. We propose a novel coarse-to-fine scheme: we first distill the generative prior to capture a robust coarse-level content representation that can link objects at an abstract semantic level, based on which finelevel content features are adaptively learned for more accurate multi-level content correspondences. Extensive experiments demonstrate the superiority of our versatile framework over state-of-the-art methods in robust, high-quality and diversified translations, even for challenging and distant domains. Code is available at https://github.com/williamyang1991/GP-UNIT.

Virtual try-on aims to transfer a target clothing image onto a reference person. Though great progress has been achieved, the functioning zone of existing works is still limited to standard clothes (e.g., plain shirt without complex laces or ripped effect), while the vast complexity and variety of non-standard clothes (e.g., off-shoulder shirt, word-shoulder dress) are largely ignored. In this work, we propose a principled framework, Re-current Tri-Level Transform (RT-VTON), that performs full-range virtual try-on on both standard and non-standard clothes. We have two key insights towards the framework design: 1) Semantics transfer requires a gradual feature transform on three different levels of clothing representations, namely clothes code, pose code and parsing code. 2) Geometry transfer requires a regularized image deformation between rigidity and flexibility. Firstly, we predict the semantics of the “after-try-on” person by recurrently refining the tri-level feature codes using local gated attention and non-local correspondence learning. Next, we design a semi-rigid deformation to align the clothing image and the predicted semantics, which preserves local warping similarity. Finally, a canonical try-on synthesizer fuses all the processed information to generate the clothed person image. Extensive experiments on conventional benchmarks along with user studies demonstrate that our framework achieves state-of-the-art performance both quantitatively and qualitatively. Notably, RT-VTON shows compelling results on a wide range of non-standard clothes. Project page: https://lzqhardworker.github.io/RT-VTON/.

Vision Transformers (ViTs) have achieved impressive performance on various vision tasks, yet their generalization under distribution shifts (DS) is rarely understood. In this work, we comprehensively study the out-of-distribution (OOD) generalization of ViTs. For systematic investigation, we first present a taxonomy of DS. We then perform extensive evaluations of ViT variants under different DS and compare their generalization with Convolutional Neural Network (CNN) models. Important observations are obtained: 1) ViTs learn weaker biases on backgrounds and textures, while they are equipped with stronger inductive biases towards shapes and structures, which is more consistent with human cognitive traits. Therefore, ViTs generalize better than CNNs under DS. With the same or less amount of parameters, ViTs are ahead of corresponding CNNs by more than 5% in top-1 accuracy under most types of DS. 2) As the model scale increases, ViTs strengthen these biases and thus gradually narrow the in-distribution and OOD performance gap. To further improve the generalization of ViTs, we design the Generalization-Enhanced ViTs (GE-ViTs) from the perspectives of adversarial learning, information theory, and self-supervised learning. By comprehensively investigating these GE-ViTs and comparing with their corresponding CNN models, we observe: 1) For the enhanced model, larger ViTs still benefit more for the OOD generalization. 2) GE-ViTs are more sensitive to the hyper-parameters than their corresponding CNN models. We design a smoother learning strategy to achieve a stable training process and obtain performance improvements on OOD data by 4% from vanilla ViTs. We hope our comprehensive study could shed light on the design of more generalizable learning architectures. Codes and datasets are released in https://github.com/Phoenix1153/ViT_OOD_generalization.

With the rise of powerful pre-trained vision-language models like CLIP, it becomes essential to investigate ways to adapt these models to downstream datasets. A recently proposed method named Context Optimization (CoOp) introduces the concept of prompt learning—a recent trend in NLP—to the vision domain for adapting pre-trained vision-language models. Specifically, CoOp turns context words in a prompt into a set of learnable vectors and, with only a few labeled images for learning, can achieve huge improvements over intensively-tuned manual prompts. In our study we identify a critical problem of CoOp: the learned context is not generalizable to wider unseen classes within the same dataset, suggesting that CoOp overfits base classes observed during training. To address the problem, we propose Conditional Context Optimization (CoCoOp), which extends CoOp by further learning a lightweight neural network to generate for each image an input-conditional token (vector). Compared to CoOp's static prompts, our dynamic prompts adapt to each instance and are thus less sensitive to class shift. Extensive experiments show that CoCoOp generalizes much better than CoOp to unseen classes, even showing promising transferability beyond a single dataset; and yields stronger domain generalization performance as well. Code is available at https://github.com/KaiyangZhou/CoOp.

Though impressive performance has been achieved in specific visual realms (e.g. faces, dogs, and places), an omni-vision representation generalizing to many natural visual domains is highly desirable. But, existing benchmarks are biased and inefficient to evaluate the omni-vision representation—these benchmarks either only include several specific realms, or cover most realms at the expense of subsuming numerous datasets that have extensive realm overlapping. In this paper, we propose Omni-Realm Benchmark (OmniBenchmark). It includes 21 realm-wise datasets with 7,372 concepts and 1,074,346 images. Without semantic overlapping, these datasets cover most visual realms comprehensively and meanwhile efficiently. In addition, we propose a new supervised contrastive learning framework, namely Relational Contrastive learning (ReCo), for a better omni-vision representation. Beyond pulling two instances from the same concept closer—the typical supervised contrastive learning framework—ReCo also pulls two instances from the same semantic realm closer, encoding the semantic relation between concepts, facilitating omni-vision representation learning. We benchmark ReCo and other advances in omni-vision representation studies that are different in architectures (from CNNs to transformers) and in learning paradigms (from supervised learning to self-supervised learning) on OmniBenchmark. We illustrate the superior of ReCo to other supervised contrastive learning methods, and reveal multiple practical observations to facilitate future research. The code and models are available at https://zhangyuanhan-ai.github.io/OmniBenchmark .

3D perception, especially point cloud classification, has achieved substantial progress. However, in real-world deployment, point cloud corruptions are inevitable due to the scene complexity, sensor inaccuracy, and processing imprecision. In this work, we aim to rigorously benchmark and analyze point cloud classification under corruptions. To conduct a systematic investigation, we first provide a taxonomy of common 3D corruptions and identify the atomic corruptions. Then, we perform a comprehensive evaluation on a wide range of representative point cloud models to understand their robustness and generalizability. Our benchmark results show that although point cloud classification performance improves over time, the state-of-the-art methods are on the verge of being less robust. Based on the obtained observations, we propose several effective techniques to enhance point cloud classifier robustness. We hope our comprehensive benchmark, in-depth analysis, and proposed techniques could spark future research in robust 3D perception.

Optical camera communication (OCC) enabled by LED and embedded cameras has attracted extensive attention, thanks to its rich spectrum availability and ready deployability. However, the close interactions between OCC and the indoor spaces have created two major challenges. On one hand, the stripe pattern incurred by OCC may greatly damage the accuracy of image-based object recognition. On the other hand, the patterns inherent to indoor spaces can significantly degrade the decoding performance of reflected OCC. To this end, we propose CORE-Lens as a pipeline to make the mutual interference transparent to existing OR and OCC algorithms. Essentially, CORE-Lens treats the two challenges as two sides of a signal mixture issue: the signals transmitted by OCC get mixed with background images so well that their features become entangled. Consequently, CORE-Lens exploits the idea of disentangled representation learning to separate the mixed signals in the feature space: while the GAN-reconstructed clean background images are used to perform object recognition, OCC decoding is conducted on the residual of the original image after subtracting the reconstructed background. Our extensive experiments on evaluating the real-life performance of CORE-Lens evidently demonstrate its superiority over conventional approaches.

Visual correspondence of 2D animation is the core of many applications and deserves careful study. Existing correspondence datasets for 2D cartoon suffer from simple frame composition and monotonic movements, making them insufficient to simulate real animations. In this work, we present a new 2D animation visual correspondence dataset, AnimeRun, by converting open source 3D movies to full scenes in 2D style, including simultaneous moving background and interactions of multiple subjects. Statistics show that our proposed dataset not only resembles real anime more in image composition, but also possesses richer and more complex motion patterns compared to existing datasets. With this dataset, we establish a comprehensive benchmark by evaluating several existing optical flow and segment matching methods, and analyze shortcomings of these methods on animation data. Data are available at https://lisiyao21.github.io/projects/AnimeRun.

Co-speech gesture is crucial for human-machine interaction and digital entertainment. While previous works mostly map speech audio to human skeletons (e.g., 2D keypoints), directly generating speakers' gestures in the image domain remains unsolved. In this work, we formally define and study this challenging problem of audio-driven co-speech gesture video generation, i.e., using a unified framework to generate speaker image sequence driven by speech audio. Our key insight is that the co-speech gestures can be decomposed into common motion patterns and subtle rhythmic dynamics. To this end, we propose a novel framework, Audio-driveN Gesture vIdeo gEneration (ANGIE), to effectively capture the reusable co-speech gesture patterns as well as fine-grained rhythmic movements. To achieve high-fidelity image sequence generation, we leverage an unsupervised motion representation instead of a structural human body prior (e.g., 2D skeletons). Specifically, 1) we propose a vector quantized motion extractor (VQ-Motion Extractor) to summarize common co-speech gesture patterns from implicit motion representation to codebooks. 2) Moreover, a co-speech gesture GPT with motion refinement (Co-Speech GPT) is devised to complement the subtle prosodic motion details. Extensive experiments demonstrate that our framework renders realistic and vivid co-speech gesture video. Demo video and more resources can be found in: https://alvinliu0.github.io/projects/ANGIE

3D human pose and shape estimation (a.k.a. ``human mesh recovery'') has achieved substantial progress. Researchers mainly focus on the development of novel algorithms, while less attention has been paid to other critical factors involved. This could lead to less optimal baselines, hindering the fair and faithful evaluations of newly designed methodologies. To address this problem, this work presents the \textit{first} comprehensive benchmarking study from three under-explored perspectives beyond algorithms. \emph{1) Datasets.} An analysis on 31 datasets reveals the distinct impacts of data samples: datasets featuring critical attributes (\emph{i.e.} diverse poses, shapes, camera characteristics, backbone features) are more effective. Strategical selection and combination of high-quality datasets can yield a significant boost to the model performance. \emph{2) Backbones.} Experiments with 10 backbones, ranging from CNNs to transformers, show the knowledge learnt from a proximity task is readily transferable to human mesh recovery. \emph{3) Training strategies.} Proper augmentation techniques and loss designs are crucial. With the above findings, we achieve a PA-MPJPE of 47.3 (mm) on the 3DPW test set with a relatively simple model. More importantly, we provide strong baselines for fair comparisons of algorithms, and recommendations for building effective training configurations in the future. Codebase is available at \url{https://github.com/smplbody/hmr-benchmarks}.

Out-of-distribution (OOD) detection is vital to safety-critical machine learning applications and has thus been extensively studied, with a plethora of methods developed in the literature. However, the field currently lacks a unified, strictly formulated, and comprehensive benchmark, which often results in unfair comparisons and inconclusive results. From the problem setting perspective, OOD detection is closely related to neighboring fields including anomaly detection (AD), open set recognition (OSR), and model uncertainty, since methods developed for one domain are often applicable to each other. To help the community to improve the evaluation and advance, we build a unified, well-structured codebase called OpenOOD, which implements over 30 methods developed in relevant fields and provides a comprehensive benchmark under the recently proposed generalized OOD detection framework. With a comprehensive comparison of these methods, we are gratified that the field has progressed significantly over the past few years, where both preprocessing methods and the orthogonal post-hoc methods show strong potential.

Previous studies have explored generating accurately lip-synced talking faces for arbitrary targets given audio conditions. However, most of them deform or generate the whole facial area, leading to non-realistic results. In this work, we delve into the formulation of altering only the mouth shapes of the target person. This requires masking a large percentage of the original image and seamlessly inpainting it with the aid of audio and reference frames. To this end, we propose the Audio-Visual Context-Aware Transformer (AV-CAT) framework, which produces accurate lip-sync with photo-realistic quality by predicting the masked mouth shapes. Our key insight is to exploit desired contextual information provided in audio and visual modalities thoroughly with delicately designed Transformers. Specifically, we propose a convolution-Transformer hybrid backbone and design an attention-based fusion strategy for filling the masked parts. It uniformly attends to the textural information on the unmasked regions and the reference frame. Then the semantic audio information is involved in enhancing the self-attention computation. Additionally, a refinement network with audio injection improves both image and lip-sync quality. Extensive experiments validate that our model can generate high-fidelity lip-synced results for arbitrary subjects.

Feature reassembly, i.e. feature downsampling and upsampling, is a key operation in a number of modern convolutional network architectures, e.g., residual networks and feature pyramids. Its design is critical for dense prediction tasks such as object detection and semantic/instance segmentation. In this work, we propose unified Content-Aware ReAssembly of FEatures (CARAFE++), a universal, lightweight, and highly effective operator to fulfill this goal. CARAFE++ has several appealing properties: (1) Unlike conventional methods such as pooling and interpolation that only exploit sub-pixel neighborhood, CARAFE++ aggregates contextual information within a large receptive field. (2) Instead of using a fixed kernel for all samples (e.g. convolution and deconvolution), CARAFE++ generates adaptive kernels on-the-fly to enable instance-specific content-aware handling. (3) CARAFE++ introduces little computational overhead and can be readily integrated into modern network architectures. We conduct comprehensive evaluations on standard benchmarks in object detection, instance/semantic segmentation, and image inpainting. CARAFE++ shows consistent and substantial gains on mainstream methods across all the tasks with negligible computational overhead. It shows great potential to serve as a strong building block for modern deep networks.

Deep neural networks have achieved remarkable progress in single-image 3D human reconstruction. However, existing methods still fall short in predicting rare poses. The reason is that most of the current models perform regression based on a single human prototype, which is similar to common poses while far from the rare poses. In this work, we 1) identify and analyze this learning obstacle and 2) propose a prototype memory-augmented network, PM-Net, that effectively improves performances of predicting rare poses. The core of our framework is a memory module that learns and stores a set of 3D human prototypes capturing local distributions for either common poses or rare poses. With this formulation, the regression starts from a better initialization, which is relatively easier to converge. Extensive experiments on several widely employed datasets demonstrate the proposed framework's effectiveness compared to other state-of-the-art methods. Notably, our approach significantly improves the models' performances on rare poses while generating comparable results on other samples.

High-quality HDRIs (High Dynamic Range Images), typically HDR panoramas, are one of the most popular ways to create photorealistic lighting and 360-degree reflections of 3D scenes in graphics. Given the difficulty of capturing HDRIs, a versatile and controllable generative model is highly desired, where layman users can intuitively control the generation process. However, existing state-of-the-art methods still struggle to synthesize high-quality panoramas for complex scenes. In this work, we propose a zero-shot text-driven framework, Text2Light , to generate 4K+ resolution HDRIs without paired training data. Given a free-form text as the description of the scene, we synthesize the corresponding HDRI with two dedicated steps: 1 ) text-driven panorama generation in low dynamic range (LDR) and low resolution (LR), and 2 ) super-resolution inverse tone mapping to scale up the LDR panorama both in resolution and dynamic range. Specifically, to achieve zero-shot text-driven panorama generation, we first build dual codebooks as the discrete representation for diverse environmental textures. Then, driven by the pre-trained Contrastive Language-Image Pre-training (CLIP) model, a text-conditioned global sampler learns to sample holistic semantics from the global codebook according to the input text. Furthermore, a structure-aware local sampler learns to synthesize LDR panoramas patch-by-patch, guided by holistic semantics. To achieve super-resolution inverse tone mapping, we derive a continuous representation of 360-degree imaging from the LDR panorama as a set of structured latent codes anchored to the sphere. This continuous representation enables a versatile module to upscale the resolution and dynamic range simultaneously. Extensive experiments demonstrate the superior capability of Text2Light in generating high-quality HDR panoramas. In addition, we show the feasibility of our work in realistic rendering and immersive VR.

3D avatar creation plays a crucial role in the digital age. However, the whole production process is prohibitively time-consuming and labor-intensive. To democratize this technology to a larger audience, we propose AvatarCLIP, a zero-shot text-driven framework for 3D avatar generation and animation. Unlike professional software that requires expert knowledge, AvatarCLIP empowers layman users to customize a 3D avatar with the desired shape and texture, and drive the avatar with the described motions using solely natural languages. Our key insight is to take advantage of the powerful vision-language model CLIP for supervising neural human generation, in terms of 3D geometry, texture and animation. Specifically, driven by natural language descriptions, we initialize 3D human geometry generation with a shape VAE network. Based on the generated 3D human shapes, a volume rendering model is utilized to further facilitate geometry sculpting and texture generation. Moreover, by leveraging the priors learned in the motion VAE, a CLIP-guided reference-based motion synthesis method is proposed for the animation of the generated 3D avatar. Extensive qualitative and quantitative experiments validate the effectiveness and generalizability of AvatarCLIP on a wide range of avatars. Remarkably, AvatarCLIP can generate unseen 3D avatars with novel animations, achieving superior zero-shot capability. Codes are available at https://github.com/hongfz16/AvatarCLIP.

Generating high-quality and diverse human images is an important yet challenging task in vision and graphics. However, existing generative models often fall short under the high diversity of clothing shapes and textures. Furthermore, the generation process is even desired to be intuitively controllable for layman users. In this work, we present a text-driven controllable framework, Text2Human, for a high-quality and diverse human generation. We synthesize full-body human images starting from a given human pose with two dedicated steps. 1) With some texts describing the shapes of clothes, the given human pose is first translated to a human parsing map. 2) The final human image is then generated by providing the system with more attributes about the textures of clothes. Specifically, to model the diversity of clothing textures, we build a hierarchical texture-aware codebook that stores multi-scale neural representations for each type of texture. The codebook at the coarse level includes the structural representations of textures, while the codebook at the fine level focuses on the details of textures. To make use of the learned hierarchical codebook to synthesize desired images, a diffusion-based transformer sampler with mixture of experts is firstly employed to sample indices from the coarsest level of the codebook, which then is used to predict the indices of the codebook at finer levels. The predicted indices at different levels are translated to human images by the decoder learned accompanied with hierarchical codebooks. The use of mixture-of-experts allows for the generated image conditioned on the fine-grained text input. The prediction for finer level indices refines the quality of clothing textures. Extensive quantitative and qualitative evaluations demonstrate that our proposed Text2Human framework can generate more diverse and realistic human images compared to state-of-the-art methods. Our project page is https://yumingj.github.io/projects/Text2Human.html. Code and pretrained models are available at https://github.com/yumingj/Text2Human.

Generating high-quality artistic portrait videos is an important and desirable task in computer graphics and vision. Although a series of successful portrait image toonification models built upon the powerful StyleGAN have been proposed, these image-oriented methods have obvious limitations when applied to videos, such as the fixed frame size, the requirement of face alignment, missing non-facial details and temporal inconsistency. In this work, we investigate the challenging controllable high-resolution portrait video style transfer by introducing a novel VToonify framework. Specifically, VToonify leverages the mid- and high-resolution layers of StyleGAN to render high-quality artistic portraits based on the multi-scale content features extracted by an encoder to better preserve the frame details. The resulting fully convolutional architecture accepts non-aligned faces in videos of variable size as input, contributing to complete face regions with natural motions in the output. Our framework is compatible with existing StyleGAN-based image toonification models to extend them to video toonification, and inherits appealing features of these models for flexible style control on color and intensity. This work presents two instantiations of VToonify built upon Toonify and DualStyleGAN for collection-based and exemplar-based portrait video style transfer, respectively. Extensive experimental results demonstrate the effectiveness of our proposed VToonify framework over existing methods in generating high-quality and temporally-coherent artistic portrait videos with flexible style controls. Code and pretrained models are available at our project page: www.mmlab-ntu.com/project/vtoonify/.

3D interacting hand reconstruction is essential to facilitate human-machine interaction and human behaviors understanding. Previous works in this field either rely on auxiliary inputs such as depth images or they can only handle a single hand if monocular single RGB images are used. Single-hand methods tend to generate collided hand meshes, when applied to closely interacting hands, since they cannot model the interactions between two hands explicitly. In this paper, we make the first attempt to reconstruct 3D interacting hands from monocular single RGB images. Our method can generate 3D hand meshes with both precise 3D poses and minimal collisions. This is made possible via a two-stage framework. Specifically, the first stage adopts a convolutional neural network to generate coarse predictions that tolerate collisions but encourage pose-accurate hand meshes. The second stage progressively ameliorates the collisions through a series of factorized refinements while retaining the preciseness of 3D poses. We carefully investigate potential implementations for the factorized refinement, considering the trade-off between efficiency and accuracy. Extensive quantitative and qualitative results on large-scale datasets such as InterHand2.6M demonstrate the effectiveness of the proposed approach.

Second language (L2) English learners often find it difficult to improve their pronunciations due to the lack of expressive and personalized corrective feedback. In this paper, we present Pronunciation Teacher (PTeacher), a Computer-Aided Pronunciation Training (CAPT) system that provides personalized exaggerated audio-visual corrective feedback for mispronunciations. Though the effectiveness of exaggerated feedback has been demonstrated, it is still unclear how to define the appropriate degrees of exaggeration when interacting with individual learners. To fill in this gap, we interview 100 L2 English learners and 22 professional native teachers to understand their needs and experiences. Three critical metrics are proposed for both learners and teachers to identify the best exaggeration levels in both audio and visual modalities. Additionally, we incorporate the personalized dynamic feedback mechanism given the English proficiency of learners. Based on the obtained insights, a comprehensive interactive pronunciation training course is designed to help L2 learners rectify mispronunciations in a more perceptible, understandable, and discriminative manner. Extensive user studies demonstrate that our system significantly promotes the learners’ learning efficiency.

Reference-based Super-Resolution (Ref-SR) has recently emerged as a promising paradigm to enhance a low-resolution (LR) input image by introducing an additional high-resolution (HR) reference image. Existing Ref-SR methods mostly rely on implicit correspondence matching to borrow HR textures from reference images to compensate for the information loss in input images. However, performing local transfer is difficult because of two gaps between input and reference images: the transformation gap (e.g. scale and rotation) and the resolution gap (e.g. HR and LR). To tackle these challenges, we propose C2-Matching in this work, which produces explicit robust matching crossing transformation and resolution. 1) For the transformation gap, we propose a contrastive correspondence network, which learns transformation-robust correspondences using augmented views of the input image. 2) For the resolution gap, we adopt a teacher-student correlation distillation, which distills knowledge from the easier HR-HR matching to guide the more ambiguous LR-HR matching. 3) Finally, we design a dynamic aggregation module to address the potential misalignment issue. In addition, to faithfully evaluate the performance of Ref-SR under a realistic setting, we contribute the Webly-Referenced SR (WR-SR) dataset, mimicking the practical usage scenario. Extensive experiments demonstrate that our proposed C2-Matching significantly outperforms state of the arts by over 1dB on the standard CUFED5 benchmark. Notably, it also shows great generalizability on WR-SR dataset as well as robustness across large scale and rotation transformations1.

The rapid progress of photorealistic synthesis techniques have reached at a critical point where the boundary between real and manipulated images starts to blur. Thus, benchmarking and advancing digital forgery analysis have become a pressing issue. However, existing face forgery datasets either have limited diversity or only support coarse-grained analysis.To counter this emerging threat, we construct the ForgeryNet dataset, an extremely large face forgery dataset with unified annotations in image- and video-level data across four tasks: 1) Image Forgery Classification, including two-way (real/fake), three-way (real/fake with identity-replaced forgery approaches/fake with identity-remained forgery approaches), and n-way (real and 15 respective forgery approaches) classification. 2) Spatial Forgery Localization, which segments the manipulated area of fake images compared to their corresponding real images. 3) Video Forgery Classification, which re-defines the video-level forgery classification with manipulated frames in random positions. This task is important because attackers in real world are free to manipulate any target frame. and 4) Temporal Forgery Localization, to localize the temporal segments which are manipulated. ForgeryNet is by far the largest publicly available deep face forgery dataset in terms of data-scale (2.9 million images, 221,247 videos), manipulations (7 image-level approaches, 8 video-level approaches), perturbations (36 independent and more mixed perturbations) and annotations (6.3 million classification labels, 2.9 million manipulated area annotations and 221,247 temporal forgery segment labels). We perform extensive benchmarking and studies of existing face forensics methods and obtain several valuable observations. We hope that the scale, quality, and variety of our ForgeryNet dataset will foster further research and innovation in the area of face forgery classification, as well as spatial and temporal forgery localization etc.

With the rapid advances of autonomous driving, it becomes critical to equip its sensing system with more holistic 3D perception. However, existing works focus on parsing either the objects (e.g. cars and pedestrians) or scenes (e.g. trees and buildings) from the LiDAR sensor. In this work, we address the task of LiDAR-based panoptic segmentation, which aims to parse both objects and scenes in a unified manner. As one of the first endeavors towards this new challenging task, we propose the Dynamic Shifting Network (DS-Net), which serves as an effective panoptic segmentation framework in the point cloud realm. In particular, DS-Net has three appealing properties: 1) strong backbone design. DS-Net adopts the cylinder convolution that is specifically designed for LiDAR point clouds. The extracted features are shared by the semantic branch and the instance branch which operates in a bottom-up clustering style. 2) Dynamic Shifting for complex point distributions. We observe that commonly-used clustering algorithms like BFS or DBSCAN are incapable of handling complex autonomous driving scenes with non-uniform point cloud distributions and varying instance sizes. Thus, we present an efficient learnable clustering module, dynamic shifting, which adapts kernel functions on-the-fly for different instances. 3) Consensus-driven Fusion. Finally, consensus-driven fusion is used to deal with the disagreement between semantic and instance predictions. To comprehensively evaluate the performance of LiDAR-based panoptic segmentation, we construct and curate benchmarks from two large-scale autonomous driving LiDAR datasets, SemanticKITTI and nuScenes. Extensive experiments demonstrate that our proposed DS-Net achieves superior accuracies over current state-of-the-art methods. Notably, we achieve 1st place on the public leaderboard of SemanticKITTI, outperforming 2nd place by 2.6% in terms of the PQ metric1.

In the animation industry, cartoon videos are usually produced at low frame rate since hand drawing of such frames is costly and time-consuming. Therefore, it is desirable to develop computational models that can automatically interpolate the in-between animation frames. However, existing video interpolation methods fail to produce satisfying results on animation data. Compared to natural videos, animation videos possess two unique characteristics that make frame interpolation difficult: 1) cartoons comprise lines and smooth color pieces. The smooth areas lack textures and make it difficult to estimate accurate motions on animation videos. 2) cartoons express stories via exaggeration. Some of the motions are non-linear and extremely large. In this work, we formally define and study the animation video interpolation problem for the first time. To address the aforementioned challenges, we propose an effective framework, AnimeInterp, with two dedicated modules in a coarse-to-fine manner. Specifically, 1) Segment-Guided Matching resolves the "lack of textures" challenge by exploiting global matching among color pieces that are piece-wise coherent. 2) Recurrent Flow Refinement resolves the "non-linear and extremely large motion" challenge by recur-rent predictions using a transformer-like architecture. To facilitate comprehensive training and evaluations, we build a large-scale animation triplet dataset, ATD-12K, which comprises 12,000 triplets with rich annotations. Extensive experiments demonstrate that our approach outperforms existing state-of-the-art interpolation methods for animation videos. Notably, AnimeInterp shows favorable perceptual quality and robustness for animation scenarios in the wild. The proposed dataset and code are available at https://github.com/lisiyao21/AnimeInterp/.

Real-scanned point clouds are often incomplete due to viewpoint, occlusion, and noise. Existing point cloud completion methods tend to generate global shape skeletons and hence lack fine local details. Furthermore, they mostly learn a deterministic partial-to-complete mapping, but overlook structural relations in man-made objects. To tackle these challenges, this paper proposes a variational framework, Variational Relational point Completion network (VRC-Net) with two appealing properties: 1) Probabilistic Modeling. In particular, we propose a dual-path architecture to enable principled probabilistic modeling across partial and complete clouds. One path consumes complete point clouds for reconstruction by learning a point VAE. The other path generates complete shapes for partial point clouds, whose embedded distribution is guided by distribution obtained from the reconstruction path during training. 2) Relational Enhancement. Specifically, we carefully design point self-attention kernel and point selective kernel module to exploit relational point features, which refines local shape de tails conditioned on the coarse completion. In addition, we contribute a multi-view partial point cloud dataset (MVP dataset) containing over 100,000 high-quality scans, which renders partial 3D shapes from 26 uniformly distributed camera poses for each 3D CAD model. Extensive experiments demonstrate that VRCNet outperforms state-of-the-art methods on all standard point cloud completion benchmarks. Notably, VRCNet shows great generalizability and robustness on real-world point cloud scans.

Unsupervised feature learning has made great strides with contrastive learning based on instance discrimination and invariant mapping, as benchmarked on curated class-balanced datasets. However, natural data could be highly correlated and long-tail distributed. Natural between-instance similarity conflicts with the presumed instance distinction, causing unstable training and poor performance.Our idea is to discover and integrate between-instance similarity into contrastive learning, not directly by instance grouping, but by cross-level discrimination (CLD) between instances and local instance groups. While invariant mapping of each instance is imposed by attraction within its augmented views, between-instance similarity could emerge from common repulsion against instance groups.Our batch-wise and cross-view comparisons also greatly improve the positive/negative sample ratio of contrastive learning and achieve better invariant mapping. To effect both grouping and discrimination objectives, we impose them on features separately derived from a shared representation. In addition, we propose normalized projection heads and unsupervised hyper-parameter tuning for the first time.Our extensive experimentation demonstrates that CLD is a lean and powerful add-on to existing methods such as NPID, MoCo, InfoMin, and BYOL on highly correlated, long-tail, or balanced datasets. It not only achieves new state-of-the-art on self-supervision, semi-supervision, and transfer learning benchmarks, but also beats MoCo v2 and SimCLR on every reported performance attained with a much larger compute. CLD effectively brings unsupervised learning closer to natural data and real-world applications. Our code is publicly available at: https://github.com/frank-xwang/CLD-UnsupervisedLearning.

Instance segmentation has witnessed a remarkable progress on class-balanced benchmarks. However, they fail to perform as accurately in real-world scenarios, where the category distribution of objects naturally comes with a long tail. Instances of head classes dominate a long-tailed dataset and they serve as negative samples of tail categories. The overwhelming gradients of negative samples on tail classes lead to a biased learning process for classifiers. Consequently, objects of tail categories are more likely to be misclassified as backgrounds or head categories. To tackle this problem, we propose Seesaw Loss to dynamically re-balance gradients of positive and negative samples for each category, with two complementary factors, i.e., mitigation factor and compensation factor. The mitigation factor reduces punishments to tail categories w.r.t. the ratio of cumulative training instances between different categories. Meanwhile, the compensation factor increases the penalty of misclassified instances to avoid false positives of tail categories. We conduct extensive experiments on Seesaw Loss with mainstream frameworks and different data sampling strategies. With a simple end-to-end training pipeline, Seesaw Loss obtains significant gains over Cross-Entropy Loss, and achieves state-of-the-art performance on LVIS dataset without bells and whistles. Code is available at https://github.com/open-mmlab/mmdetection.

Adversarial robustness has attracted extensive studies recently by revealing the vulnerability and intrinsic characteristics of deep networks. However, existing works on adversarial robustness mainly focus on balanced datasets, while real-world data usually exhibits a long-tailed distribution. To push adversarial robustness towards more realistic scenarios, in this work we investigate the adversarial vulnerability as well as defense under long-tailed distributions. In particular, we first reveal the negative impacts induced by imbalanced data on both recognition performance and adversarial robustness, uncovering the intrinsic challenges of this problem. We then perform a systematic study on existing long-tailed recognition methods in conjunction with the adversarial training framework. Several valuable observations are obtained: 1) natural accuracy is relatively easy to improve, 2) fake gain of robust accuracy exists under unreliable evaluation, and 3) boundary error limits the promotion of robustness. Inspired by these observations, we propose a clean yet effective framework, RoBal, which consists of two dedicated modules, a scale-invariant classifier and data re-balancing via both margin engineering at training stage and boundary adjustment during inference. Extensive experiments demonstrate the superiority of our approach over other state-of-the-art defense methods. To our best knowledge, we are the first to tackle adversarial robustness under long-tailed distributions, which we believe would be a significant step towards real-world robustness. Our code is available at: https://github.com/wutong16/Adversarial_Long-Tail.

Stereophonic audio, especially binaural audio, plays an essential role in immersive viewing environments. Recent research has explored generating visually guided stereophonic audios supervised by multi-channel audio collections. However, due to the requirement of professional recording devices, existing datasets are limited in scale and variety, which impedes the generalization of supervised methods in real-world scenarios. In this work, we propose PseudoBinaural, an effective pipeline that is free of binaural recordings. The key insight is to carefully build pseudo visual-stereo pairs with mono data for training. Specifically, we leverage spherical harmonic decomposition and head-related impulse response (HRIR) to identify the relationship between spatial locations and received binaural audios. Then in the visual modality, corresponding visual cues of the mono data are manually placed at sound source positions to form the pairs. Compared to fully-supervised paradigms, our binaural-recording-free pipeline shows great stability in cross-dataset evaluation and achieves comparable performance under subjective preference. Moreover, combined with binaural recordings, our method is able to further boost the performance of binaural audio generation under supervised settings1.

While accurate lip synchronization has been achieved for arbitrary-subject audio-driven talking face generation, the problem of how to efficiently drive the head pose remains. Previous methods rely on pre-estimated structural information such as landmarks and 3D parameters, aiming to generate personalized rhythmic movements. However, the inaccuracy of such estimated information under extreme conditions would lead to degradation problems. In this paper, we propose a clean yet effective framework to generate pose-controllable talking faces. We operate on non-aligned raw face images, using only a single photo as an identity reference. The key is to modularize audio-visual representations by devising an implicit low-dimension pose code. Substantially, both speech content and head pose information lie in a joint non-identity embedding space. While speech content information can be defined by learning the intrinsic synchronization between audio-visual modalities, we identify that a pose code will be complementarily learned in a modulated convolution-based reconstruction framework.Extensive experiments show that our method generates accurately lip-synced talking faces whose poses are controllable by other videos. Moreover, our model has multiple advanced capabilities including extreme view robustness and talking face frontalization.1

Facial editing is an important task in vision and graphics with numerous applications. However, existing works are incapable to deliver a continuous and fine-grained editing mode (e.g., editing a slightly smiling face to a big laughing one) with natural interactions with users. In this work, we propose Talk-to-Edit, an interactive facial editing framework that performs fine-grained attribute manipulation through dialog between the user and the system. Our key insight is to model a continual "semantic field" in the GAN latent space. 1) Unlike previous works that regard the editing as traversing straight lines in the latent space, here the fine-grained editing is formulated as finding a curving trajectory that respects fine-grained attribute landscape on the semantic field. 2) The curvature at each step is location-specific and determined by the input image as well as the users’ language requests. 3) To engage the users in a meaningful dialog, our system generates language feedback by considering both the user request and the current state of the semantic field.We also contribute CelebA-Dialog, a visual-language facial editing dataset to facilitate large-scale study. Specifically, each image has manually annotated fine-grained attribute annotations as well as template-based textual descriptions in natural language. Extensive quantitative and qualitative experiments demonstrate the superiority of our framework in terms of 1) the smoothness of fine-grained editing, 2) the identity/attribute preservation, and 3) the visual photorealism and dialog fluency. Notably, user study validates that our overall system is consistently favored by around 80% of the participants. Our project page is https://www.mmlab-ntu.com/project/talkedit/.

Deep learning-based 3D object detection has achieved unprecedented success with the advent of large-scale autonomous driving datasets. However, drastic performance degradation remains a critical challenge for cross-domain deployment. In addition, existing 3D domain adaptive detection methods often assume prior access to the target domain annotations, which is rarely feasible in the real world. To address this challenge, we study a more realistic setting, unsupervised 3D domain adaptive detection, which only utilizes source domain annotations. 1) We first comprehensively investigate the major underlying factors of the domain gap in 3D detection. Our key insight is that geometric mismatch is the key factor of domain shift. 2) Then, we propose a novel and unified framework, Multi-Level Consistency Network (MLC-Net), which employs a teacher-student paradigm to generate adaptive and reliable pseudo-targets. MLC-Net exploits point-, instance- and neural statistics-level consistency to facilitate cross-domain transfer. Extensive experiments demonstrate that MLC-Net out-performs existing state-of-the-art methods (including those using additional target domain information) on standard benchmarks. Notably, our approach is detector-agnostic, which achieves consistent gains on both single- and two-stage 3D detectors. Code will be released.

Confidence calibration is of great importance to the reliability of decisions made by machine learning systems. However, discriminative classifiers based on deep neural networks are often criticized for producing overconfident predictions that fail to reflect the true correctness likelihood of classification accuracy. We argue that such an inability to model uncertainty is mainly caused by the closed-world nature in softmax: a model trained by the cross-entropy loss will be forced to classify input into one of K pre-defined categories with high probability. To address this problem, we for the first time propose a novel K+1-way softmax formulation, which incorporates the modeling of open-world uncertainty as the extra dimension. To unify the learning of the original K-way classification task and the extra dimension that models uncertainty, we 1) propose a novel energy-based objective function, and moreover, 2) theoretically prove that optimizing such an objective essentially forces the extra dimension to capture the marginal data distribution. Extensive experiments show that our approach, Energy-based Open-World Softmax (EOW-Softmax), is superior to existing state-of-the-art methods in improving confidence calibration.

City modeling is the foundation for computational urban planning, navigation, and entertainment. In this work, we present the first generative model of city blocks named BlockPlanner, and showcase its ability to synthesize valid city blocks with varying land lots configurations. We propose a novel vectorized city block representation utilizing a ring topology and a two-tier graph to capture the global and local structures of a city block. Each land lot is abstracted into a vector representation covering both its 3D geometry and land use semantics. Such vectorized representation enables us to deploy a lightweight network to capture the underlying distribution of land lots configurations in a city block. To enforce intrinsic spatial constraints of a valid city block, a set of effective loss functions are imposed to shape rational results. We contribute a pilot city block dataset to demonstrate the effectiveness and efficiency of our representation and framework over the state-of-the-art. Notably, our BlockPlanner is also able to edit and manipulate city blocks, enabling several useful applications, e.g., topology refinement and footprint generation.

Current out-of-distribution (OOD) detection benchmarks are commonly built by defining one dataset as in-distribution (ID) and all others as OOD. However, these benchmarks unfortunately introduce some unwanted and impractical goals, e.g., to perfectly distinguish CIFAR dogs from ImageNet dogs, even though they have the same semantics and negligible covariate shifts. These unrealistic goals will result in an extremely narrow range of model capabilities, greatly limiting their use in real applications. To overcome these drawbacks, we re-design the benchmarks and propose the semantically coherent out-of-distribution detection (SC-OOD). On the SC-OOD benchmarks, existing methods suffer from large performance degradation, suggesting that they are extremely sensitive to low-level discrepancy between data sources while ignoring their inherent semantics. To develop an effective SC-OOD detection approach, we leverage an external unlabeled set and design a concise framework featured by unsupervised dual grouping (UDG) for the joint modeling of ID and OOD data. The proposed UDG can not only enrich the semantic knowledge of the model by exploiting unlabeled data in an unsupervised manner, but also distinguish ID/OOD samples to enhance ID classification and OOD detection tasks simultaneously. Extensive experiments demonstrate that our approach achieves the state-of-the-art performance on SC-OOD benchmarks. Code and benchmarks are provided on our project page: https://jingkang50.github.io/projects/scood.

Motivated by the success of Transformers in natural language processing (NLP) tasks, there emerge some attempts (e.g., ViT and DeiT) to apply Transformers to the vision domain. However, pure Transformer architectures often require a large amount of training data or extra supervision to obtain comparable performance with convolutional neural networks (CNNs). To overcome these limitations, we analyze the potential drawbacks when directly borrowing Transformer architectures from NLP. Then we propose a new Convolution-enhanced image Transformer (CeiT) which combines the advantages of CNNs in extracting low-level features, strengthening locality, and the advantages of Transformers in establishing long-range dependencies. Three modifications are made to the original Transformer: 1) instead of the straightforward tokenization from raw input images, we design an Image-to-Tokens (I2T) module that extracts patches from generated low-level features; 2) the feed-froward network in each encoder block is replaced with a Locally-enhanced Feed-Forward (LeFF) layer that promotes the correlation among neighboring tokens in the spatial dimension; 3) a Layer-wise Class token Attention (LCA) is attached at the top of the Transformer that utilizes the multi-level representations.Experimental results on ImageNet and seven down-stream tasks show the effectiveness and generalization ability of CeiT compared with previous Transformers and state- of-the-art CNNs, without requiring a large amount of training data and extra CNN teachers. Besides, CeiT models demonstrate better convergence with 3× fewer training iterations, which can reduce the training cost significantly1.

A standard practice of deploying deep neural networks is to apply the same architecture to all the input instances. However, a fixed architecture may not be suitable for different data with high diversity. To boost the model capacity, existing methods usually employ larger convolutional kernels or deeper network layers, which incurs prohibitive computational costs. In this paper, we address this issue by proposing Differentiable Dynamic Wirings (DDW), which learns the instance-aware connectivity that creates different wiring patterns for different instances. 1) Specifically, the network is initialized as a complete directed acyclic graph, where the nodes represent convolutional blocks and the edges represent the connection paths. 2) We generate edge weights by a learnable module, Router, and select the edges whose weights are larger than a threshold, to adjust the connectivity of the neural network structure. 3) Instead of using the same path of the network, DDW aggregates features dynamically in each node, which allows the network to have more representation power.To facilitate effective training, we further represent the network connectivity of each sample as an adjacency matrix. The matrix is updated to aggregate features in the forward pass, cached in the memory, and used for gradient computing in the backward pass. We validate the effectiveness of our approach with several mainstream architectures, including MobileNetV2, ResNet, ResNeXt, and RegNet. Extensive experiments are performed on ImageNet classification and COCO object detection, which demonstrates the effectiveness and generalization ability of our approach.

What can we picture solely from a clip of speech? Previous research has shown the possibility of directly inferring the appearance of a person's face by listening to a voice. However, within human speech lies not only the biometric identity signal but also the identity-irrelevant information such as the talking content. Our goal is to extract as much information from a clip of speech as possible. In particular, we aim at not only inferring the face of a person but also animating it. Our key insight is to synchronize audio and visual representations from two perspectives in a style-based generative framework. Specifically, contrastive learning is leveraged to map both the identity and speech content information within the speech to visual representation spaces. Furthermore, the identity space is strengthened with class centroids. Through curriculum learning, the style-based generator is capable of automatically balancing the information from the two latent spaces. Extensive experiments show that our approach encourages better speech-identity correlation learning while generating vivid faces whose identities are consistent with given speech samples. Moreover, by leveraging the same model, these inferred faces can be driven to talk by the audio.

We present MMFashion, a comprehensive, flexible and user-friendly open-source visual fashion analysis toolbox based on PyTorch. This toolbox supports a wide spectrum of fashion analysis tasks, including Fashion Attribute Prediction, Fashion Recognition and Retrieval, Fashion Landmark Detection, Fashion Parsing and Segmentation and Fashion Compatibility and Recommendation. It covers almost all the mainstream tasks in fashion analysis community. MMFashion has several appealing properties. Firstly, MMFashion follows the principle of modular design. The framework is decomposed into different components so that it is easily extensible for diverse customized modules. In addition, detailed documentations, demo scripts and off-the-shelf models are available, which ease the burden of layman users to leverage the recent advances in deep learning-based fashion analysis. Our proposed MMFashion is currently the most complete platform for visual fashion analysis in deep learning era, with more functionalities to be added. This toolbox and the benchmark could serve the flourishing research community by providing a flexible toolkit to deploy existing models and develop new ideas and approaches. We welcome all contributions to this still-growing efforts towards open science: https://github.com/open-mmlab/mmfashion.

Object detection has achieved substantial progress in the last decade. However, detecting novel classes with only few samples remains challenging, since deep learning under low data regime usually leads to a degraded feature space. Existing works employ a holistic fine-tuning paradigm to tackle this problem, where the model is first pre-trained on all base classes with abundant samples, and then it is used to carve the novel class feature space. Nonetheless, this paradigm is still imperfect. Durning fine-tuning, a novel class may implicitly leverage the knowledge of multiple base classes to construct its feature space, which induces a scattered feature space, hence violating the inter-class separability. To overcome these obstacles, we propose a two-step fine-tuning framework, Few-shot object detection via Association and DIscrimination (FADI), which builds up a discriminative feature space for each novel class with two integral steps. 1) In the association step, in contrast to implicitly leveraging multiple base classes, we construct a compact novel class feature space via explicitly imitating a specific base class feature space. Specifically, we associate each novel class with a base class according to their semantic similarity. After that, the feature space of a novel class can readily imitate the well-trained feature space of the associated base class. 2) In the discrimination step, to ensure the separability between the novel classes and associated base classes, we disentangle the classification branches for base and novel classes. To further enlarge the inter-class separability between all classes, a set-specialized margin loss is imposed. Extensive experiments on standard Pascal VOC and MS-COCO datasets demonstrate that FADI achieves new state-of-the-art performance, significantly improving the baseline in any shot/split by +18.7. Notably, the advantage of FADI is most announced on extremely few-shot scenarios (e.g. 1- and 3- shot).

Learning to reconstruct 3D garments is important for dressing 3D human bodies of different shapes in different poses. Previous works typically rely on 2D images as input, which however suffer from the scale and pose ambiguities. To circumvent the problems caused by 2D images, we propose a principled framework, Garment4D, that uses 3D point cloud sequences of dressed humans for garment reconstruction. Garment4D has three dedicated steps: sequential garments registration, canonical garment estimation, and posed garment reconstruction. The main challenges are two-fold: 1) effective 3D feature learning for fine details, and 2) capture of garment dynamics caused by the interaction between garments and the human body, especially for loose garments like skirts. To unravel these problems, we introduce a novel Proposal-Guided Hierarchical Feature Network and Iterative Graph Convolution Network, which integrate both high-level semantic features and low-level geometric features for fine details reconstruction. Furthermore, we propose a Temporal Transformer for smooth garment motions capture. Unlike non-parametric methods, the reconstructed garment meshes by our method are separable from the human body and have strong interpretability, which is desirable for downstream tasks. As the first attempt at this task, high-quality reconstruction results are qualitatively and quantitatively illustrated through extensive experiments. Codes are available at https://github.com/hongfz16/Garment4D.

Chamfer Distance (CD) and Earth Mover’s Distance (EMD) are two broadly adopted metrics for measuring the similarity between two point sets. However, CD is usually insensitive to mismatched local density, and EMD is usually dominated by global distribution while overlooks the fidelity of detailed structures. Besides, their unbounded value range induces a heavy influence from the outliers. These defects prevent them from providing a consistent evaluation. To tackle these problems, we propose a new similarity measure named Density-aware Chamfer Distance (DCD). It is derived from CD and benefits from several desirable properties: 1) it can detect disparity of density distributions and is thus a more intensive measure of similarity compared to CD; 2) it is stricter with detailed structures and significantly more computationally efficient than EMD; 3) the bounded value range encourages a more stable and reasonable evaluation over the whole test set. We adopt DCD to evaluate the point cloud completion task, where experimental results show that DCD pays attention to both the overall structure and local geometric details and provides a more reliable evaluation even when CD and EMD contradict each other. We can also use DCD as the training loss, which outperforms the same model trained with CD loss on all three metrics. In addition, we propose a novel point discriminator module that estimates the priority for another guided down-sampling step, and it achieves noticeable improvements under DCD together with competitive results for both CD and EMD. We hope our work could pave the way for a more comprehensive and practical point cloud similarity evaluation. Our code will be available at https://github.com/wutong16/DensityawareChamfer_Distance.

Contrastive self-supervised learning has largely narrowed the gap to supervised pre-training on ImageNet. However, its success highly relies on the object-centric priors of ImageNet, i.e., different augmented views of the same image correspond to the same object. Such a heavily curated constraint becomes immediately infeasible when pre-trained on more complex scene images with many objects. To overcome this limitation, we introduce Object-level Representation Learning (ORL), a new self-supervised learning framework towards scene images. Our key insight is to leverage image-level self-supervised pre-training as the prior to discover object-level semantic correspondence, thus realizing object-level representation learning from scene images. Extensive experiments on COCO show that ORL significantly improves the performance of self-supervised learning on scene images, even surpassing supervised ImageNet pre-training on several downstream tasks. Furthermore, ORL improves the downstream performance when more unlabeled scene images are available, demonstrating its great potential of harnessing unlabeled data in the wild. We hope our approach can motivate future research on more general-purpose unsupervised representation learning from scene data.

Despite significant progress, controlled generation of complex images with interacting people remains difficult. Existing layout generation methods fall short of synthesizing realistic person instances; while pose-guided generation approaches focus on a single person and assume simple or known backgrounds. To tackle these limitations, we propose a new problem, Persons in Context Synthesis, which aims to synthesize diverse person instance(s) in consistent contexts, with user control over both. The context is specified by the bounding box object layout which lacks shape information, while pose of the person(s) by keypoints which are sparsely annotated. To handle the stark difference in input structures, we proposed two separate neural branches to attentively composite the respective (context/person) inputs into shared "compositional structural space", which encodes shape, location and appearance information for both context and person structures in a disentangled manner. This structural space is then decoded to the image space using multi-level feature modulation strategy, and learned in a self supervised manner from image collections and their corresponding inputs. Extensive experiments on two large-scale datasets (COCO-Stuff [6] and Visual Genome [19]) demonstrate that our framework outperforms state-of-theart methods w.r.t. synthesis quality.

Fashion recommendation systems are highly desired by customers to find visually-collocated fashion items, such as clothes, shoes, bags, etc. While existing methods demonstrate promising results, they remain lacking in flexibility and diversity, e.g. assuming a fixed number of items or favoring safe but boring recommendations. In this paper, we propose a novel fashion collocation framework,Neural Graph Filtering, that models a flexible set of fashion items via a graph neural network. Specifically, we consider the visual embeddings of each garment as a node in the graph, and describe the inter-garment relationship as the edge between nodes. By applying symmetric operations on the edge vectors, this framework allows varying numbers of inputs/outputs and is invariant to their ordering. We further include a style classifier augmented with focal loss to enable the collocation of significantly diverse styles, which are inherently imbalanced in the training set. To facilitate a comprehensive study on diverse fashion collocation, we reorganize Amazon Fashion dataset with carefully designed evaluation protocols. We evaluate the proposed approach on three popular benchmarks, the Polyvore dataset, the Polyvore-D dataset, and our reorganized Amazon Fashion dataset. Extensive experimental results show that our approach significantly outperforms the state-of-the-art methods with over 10% improvements on the standard AUC metric. More importantly, 82.5% of the users prefer our diverse-style recommendations over other alternatives in a real-world perception study.

A typical domain adaptation approach is to adapt models trained on the annotated data in a source domain (e.g., sunny weather) for achieving high performance on the test data in a target domain (e.g., rainy weather). Whether the target contains a single homogeneous domain or multiple heterogeneous domains, existing works always assume that there exist clear distinctions between the domains, which is often not true in practice (e.g., changes in weather). We study an open compound domain adaptation (OCDA) problem, in which the target is a compound of multiple homogeneous domains without domain labels, reflecting realistic data collection from mixed and novel situations. We propose a new approach based on two technical insights into OCDA: 1) a curriculum domain adaptation strategy to bootstrap generalization across domains in a data-driven self-organizing fashion and 2) a memory module to increase the model's agility towards novel domains. Our experiments on digit classification, facial expression recognition, semantic segmentation, and reinforcement learning demonstrate the effectiveness of our approach.

Recent advances in adversarial attacks uncover the intrinsic vulnerability of modern deep neural networks. Since then, extensive efforts have been devoted to enhancing the robustness of deep networks via specialized learning algorithms and loss functions. In this work, we take an architectural perspective and investigate the patterns of network architectures that are resilient to adversarial attacks. To obtain the large number of networks needed for this study, we adopt one-shot neural architecture search, training a large network for once and then finetuning the sub-networks sampled therefrom. The sampled architectures together with the accuracies they achieve provide a rich basis for our study. Our ''robust architecture Odyssey'' reveals several valuable observations: 1) densely connected patterns result in improved robustness; 2) under computational budget, adding convolution operations to direct connection edge is effective; 3) flow of solution procedure (FSP) matrix is a good indicator of network robustness. Based on these observations, we discover a family of robust architectures (RobNets). On various datasets, including CIFAR, SVHN, Tiny-ImageNet, and ImageNet, RobNets exhibit superior robustness performance to other widely used architectures. Notably, RobNets substantially improve the robust accuracy (~5% absolute gains) under both white-box and black-box attacks, even with fewer parameter numbers. Code is available at https://github.com/gmh14/RobNets.

Facial image manipulation has achieved great progress in recent years. However, previous methods either operate on a predefined set of face attributes or leave users little freedom to interactively manipulate images. To overcome these drawbacks, we propose a novel framework termed MaskGAN, enabling diverse and interactive face manipulation. Our key insight is that semantic masks serve as a suitable intermediate representation for flexible face manipulation with fidelity preservation. MaskGAN has two main components: 1) Dense Mapping Network (DMN) and 2) Editing Behavior Simulated Training (EBST). Specifically, DMN learns style mapping between a free-form user modified mask and a target image, enabling diverse generation results. EBST models the user editing behavior on the source mask, making the overall framework more robust to various manipulated inputs. Specifically, it introduces dual-editing consistency as the auxiliary supervision signal. To facilitate extensive studies, we construct a large-scale high-resolution face dataset with fine-grained mask annotations named CelebAMask-HQ. MaskGAN is comprehensively evaluated on two challenging tasks: attribute transfer and style copy, demonstrating superior performance over other state-of-the-art methods. The code, models, and dataset are available at https://github.com/switchablenorms/CelebAMask-HQ.

Natural scene understanding is a challenging task, particularly when encountering images of multiple objects that are partially occluded. This obstacle is given rise by varying object ordering and positioning. Existing scene understanding paradigms are able to parse only the visible parts, resulting in incomplete and unstructured scene interpretation. In this paper, we investigate the problem of scene de-occlusion, which aims to recover the underlying occlusion ordering and complete the invisible parts of occluded objects. We make the first attempt to address the problem through a novel and unified framework that recovers hidden scene structures without ordering and amodal annotations as supervisions. This is achieved via Partial Completion Network (PCNet)-mask (M) and -content (C), that learn to recover fractions of object masks and contents, respectively, in a self-supervised manner. Based on PCNet-M and PCNet-C, we devise a novel inference scheme to accomplish scene de-occlusion, via progressive ordering recovery, amodal completion and content completion. Extensive experiments on real-world scenes demonstrate the superior performance of our approach to other alternatives. Remarkably, our approach that is trained in a self-supervised manner achieves comparable results to fully-supervised methods. The proposed scene de-occlusion framework benefits many applications, including high-quality and controllable image manipulation and scene recomposition (see Fig. 1), as well as the conversion of existing modal mask annotations to amodal mask annotations.

Joint clustering and feature learning methods have shown remarkable performance in unsupervised representation learning. However, the training schedule alternating between feature clustering and network parameters update leads to unstable learning of visual representations. To overcome this challenge, we propose Online Deep Clustering (ODC) that performs clustering and network update simultaneously rather than alternatingly. Our key insight is that the cluster centroids should evolve steadily in keeping the classifier stably updated. Specifically, we design and maintain two dynamic memory modules, i.e., samples memory to store samples' labels and features, and centroids memory for centroids evolution. We break down the abrupt global clustering into steady memory update and batch-wise label re-assignment. The process is integrated into network update iterations. In this way, labels and the network evolve shoulder-to-shoulder rather than alternatingly. Extensive experiments demonstrate that ODC stabilizes the training process and boosts the performance effectively.

Though face rotation has achieved rapid progress in recent years, the lack of high-quality paired training data remains a great hurdle for existing methods. The current generative models heavily rely on datasets with multi-view images of the same person. Thus, their generated results are restricted by the scale and domain of the data source. To overcome these challenges, we propose a novel unsupervised framework that can synthesize photo-realistic rotated faces using only single-view image collections in the wild. Our key insight is that rotating faces in the 3D space back and forth, and re-rendering them to the 2D plane can serve as a strong self-supervision. We leverage the recent advances in 3D face modeling and high-resolution GAN to constitute our building blocks. Since the 3D rotation-and-render on faces can be applied to arbitrary angles without losing details, our approach is extremely suitable for in-the-wild scenarios (i.e. no paired data are available), where existing methods fall short. Extensive experiments demonstrate that our approach has superior synthesis quality as well as identity preservation over the state-of-the-art methods, across a wide range of poses and domains. Furthermore, we validate that our rotate-and-render framework naturally can act as an effective data augmentation engine for boosting modern face recognition systems even on strong baseline models.

We address the problem of instance-level facial attribute transfer without paired training data, e.g., faithfully transferring the exact mustache from a source face to a target face. This is a more challenging task than the conventional semantic-level attribute transfer, which only preserves the generic attribute style instead of instance-level traits. We propose the use of geometry-aware flow, which serves as a wellsuited representation for modeling the transformation between instance-level facial attributes. Specifically, we leverage the facial landmarks as the geometric guidance to learn the differentiable flows automatically, despite of the large pose gap existed. Geometry-aware flow is able to warp the source face attribute into the target face context and generate a warp-and-blend result. To compensate for the potential appearance gap between source and target faces, we propose a hallucination sub-network that produces an appearance residual to further refine the warp-and-blend result. Finally, a cycle-consistency framework consisting of both attribute transfer module and attribute removal module is designed, so that abundant unpaired face images can be used as training data. Extensive evaluations validate the capability of our approach in transferring instance-level facial attributes faithfully across large pose and appearance gaps. Thanks to the flow representation, our approach can readily be applied to generate realistic details on high-resolution images1.

Talking face generation aims to synthesize a sequence of face images that correspond to a clip of speech. This is a challenging task because face appearance variation and semantics of speech are coupled together in the subtle movements of the talking face regions. Existing works either construct specific face appearance model on specific subjects or model the transformation between lip motion and speech. In this work, we integrate both aspects and enable arbitrary-subject talking face generation by learning disentangled audio-visual representation. We find that the talking face sequence is actually a composition of both subject-related information and speech-related information. These two spaces are then explicitly disentangled through a novel associative-and-adversarial training process. This disentangled representation has an advantage where both audio and video can serve as inputs for generation. Extensive experiments show that the proposed approach generates realistic talking face sequences on arbitrary subjects with much clearer lip motion patterns than previous work. We also demonstrate the learned audio-visual representation is extremely useful for the tasks of automatic lip reading and audio-video retrieval.

Real world data often have a long-tailed and open-ended distribution. A practical recognition system must classify among majority and minority classes, generalize from a few known instances, and acknowledge novelty upon a never seen instance. We define Open Long-Tailed Recognition (OLTR) as learning from such naturally distributed data and optimizing the classification accuracy over a balanced test set which include head, tail, and open classes. OLTR must handle imbalanced classification, few-shot learning, and open-set recognition in one integrated algorithm, whereas existing classification approaches focus only on one aspect and deliver poorly over the entire class spectrum. The key challenges are how to share visual knowledge between head and tail classes and how to reduce confusion between tail and open classes. We develop an integrated OLTR algorithm that maps an image to a feature space such that visual concepts can easily relate to each other based on a learned metric that respects the closed-world classification while acknowledging the novelty of the open world. Our so-called dynamic meta-embedding combines a direct image feature and an associated memory feature, with the feature norm indicating the familiarity to known classes. On three large-scale OLTR datasets we curate from object-centric ImageNet, scene-centric Places, and face-centric MS1M data, our method consistently outperforms the state-of-the-art. Our code, datasets, and models enable future OLTR research and are publicly available at \url{https://liuziwei7.github.io/projects/LongTail.html}.

Cascade is a classic yet powerful architecture that has boosted performance on various tasks. However, how to introduce cascade to instance segmentation remains an open question. A simple combination of Cascade R-CNN and Mask R-CNN only brings limited gain. In exploring a more effective approach, we find that the key to a successful instance segmentation cascade is to fully leverage the reciprocal relationship between detection and segmentation. In this work, we propose a new framework, Hybrid Task Cascade (HTC), which differs in two important aspects: (1) instead of performing cascaded refinement on these two tasks separately, it interweaves them for a joint multi-stage processing; (2) it adopts a fully convolutional branch to provide spatial context, which can help distinguishing hard foreground from cluttered background. Overall, this framework can learn more discriminative features progressively while integrating complementary features together in each stage. Without bells and whistles, a single HTC obtains 38.4% and 1.5% improvement over a strong Cascade Mask R-CNN baseline on MSCOCO dataset. Moreover, our overall system achieves 48.6 mask AP on the test-challenge split, ranking 1st in the COCO 2018 Challenge Object Detection Task. Code is available at https://github.com/open-mmlab/mmdetection.

Intelligent agent naturally learns from motion. Various self-supervised algorithms have leveraged the motion cues to learn effective visual representations. The hurdle here is that motion is both ambiguous and complex, rendering previous works either suffer from degraded learning efficacy, or resort to strong assumptions on object motions. In this work, we design a new learning-from-motion paradigm to bridge these gaps. Instead of explicitly modeling the motion probabilities, we design the pretext task as a conditional motion propagation problem. Given an input image and several sparse flow guidance on it, our framework seeks to recover the full-image motion. Compared to other alternatives, our framework has several appealing properties: (1) Using sparse flow guidance during training resolves the inherent motion ambiguity, and thus easing feature learning. (2) Solving the pretext task of conditional motion propagation encourages the emergence of kinematically-sound representations that poss greater expressive power. Extensive experiments demonstrate that our framework learns structural and coherent features; and achieves state-of-the-art self-supervision performance on several downstream tasks including semantic segmentation, instance segmentation and human parsing. Furthermore, our framework is successfully extended to several useful applications such as semi-automatic pixel-level annotation.

Though much progress has been achieved in single-image 3D human recovery, estimating 3D model for in-the-wild images remains a formidable challenge. The reason lies in the fact that obtaining high-quality 3D annotations for in-the-wild images is an extremely hard task that consumes enormous amount of resources and manpower. To tackle this problem, previous methods adopt a hybrid training strategy that exploits multiple heterogeneous types of annotations including 3D and 2D while leaving the efficacy of each annotation not thoroughly investigated. In this work, we aim to perform a comprehensive study on cost and effectiveness trade-off between different annotations. Specifically, we focus on the challenging task of in-the-wild 3D human recovery from single images when paired 3D annotations are not fully available. Through extensive experiments, we obtain several observations: 1) 3D annotations are efficient, whereas traditional 2D annotations such as 2D keypoints and body part segmentation are less competent in guiding 3D human recovery. 2) Dense Correspondence such as DensePose is effective. When there are no paired in-the-wild 3D annotations available, the model exploiting dense correspondence can achieve 92% of the performance compared to a model trained with paired 3D data. We show that incorporating dense correspondence into in-the-wild 3D human recovery is promising and competitive due to its high efficiency and relatively low annotating cost. Our model trained with dense correspondence can serve as a strong reference for future research.

Feature upsampling is a key operation in a number of modern convolutional network architectures, e.g. feature pyramids. Its design is critical for dense prediction tasks such as object detection and semantic/instance segmentation. In this work, we propose Content-Aware ReAssembly of FEatures (CARAFE), a universal, lightweight and highly effective operator to fulfill this goal. CARAFE has several appealing properties: (1) Large field of view. Unlike previous works (e.g. bilinear interpolation) that only exploit subpixel neighborhood, CARAFE can aggregate contextual information within a large receptive field. (2) Content-aware handling. Instead of using a fixed kernel for all samples (e.g. deconvolution), CARAFE enables instance-specific content-aware handling, which generates adaptive kernels on-the-fly. (3) Lightweight and fast to compute. CARAFE introduces little computational overhead and can be readily integrated into modern network architectures. We conduct comprehensive evaluations on standard benchmarks in object detection, instance/semantic segmentation and inpainting. CARAFE shows consistent and substantial gains across all the tasks (1.2% AP, 1.3% AP, 1.8% mIoU, 1.1dB respectively) with negligible computational overhead. It has great potential to serve as a strong building block for future research. Code and models are available at https://github.com/open-mmlab/mmdetection.

Multi-modality perception is essential to develop interactive intelligence. In this work, we consider a new task of visual information-infused audio inpainting, i.e. synthesizing missing audio segments that correspond to their accompanying videos. We identify two key aspects for a successful inpainter: (1) It is desirable to operate on spectrograms instead of raw audios. Recent advances in deep semantic image inpainting could be leveraged to go beyond the limitations of traditional audio inpainting. (2) To synthesize visually indicated audio, a visual-audio joint feature space needs to be learned with synchronization of audio and video. To facilitate a large-scale study, we collect a new multi-modality instrument-playing dataset called MUSIC-Extra-Solo (MUSICES) by enriching MUSIC dataset [51]. Extensive experiments demonstrate that our framework is capable of inpainting realistic and varying audio segments with or without visual contexts. More importantly, our synthesized audio segments are coherent with their video counterparts, showing the effectiveness of our proposed Vision-Infused Audio Inpainter (VIAI). Code, models, dataset and video results are available at https://github.com/Hangz-nju-cuhk/ Vision-Infused-Audio-Inpainter-VIAI.

Point clouds provide a flexible geometric representation suitable for countless applications in computer graphics; they also comprise the raw output of most 3D data acquisition devices. While hand-designed features on point clouds have long been proposed in graphics and vision, however, the recent overwhelming success of convolutional neural networks (CNNs) for image analysis suggests the value of adapting insight from CNN to the point cloud world. Point clouds inherently lack topological information, so designing a model to recover topology can enrich the representation power of point clouds. To this end, we propose a new neural network module dubbed EdgeConv suitable for CNN-based high-level tasks on point clouds, including classification and segmentation. EdgeConv acts on graphs dynamically computed in each layer of the network. It is differentiable and can be plugged into existing architectures. Compared to existing modules operating in extrinsic space or treating each point independently, EdgeConv has several appealing properties: It incorporates local neighborhood information; it can be stacked applied to learn global shape properties; and in multi-layer systems affinity in feature space captures semantic characteristics over potentially long distances in the original embedding. We show the performance of our model on standard benchmarks, including ModelNet40, ShapeNetPart, and S3DIS.

Deep convolutional networks for semantic image segmentation typically require large-scale labeled data, e.g., ImageNet and MS COCO, for network pre-training. To reduce annotation efforts, self-supervised semantic segmentation is recently proposed to pre-train a network without any human-provided labels. The key of this new form of learning is to design a proxy task (e.g., image colorization), from which a discriminative loss can be formulated on unlabeled data. Many proxy tasks, however, lack the critical supervision signals that could induce discriminative representation for the target image segmentation task. Thus self-supervision’s performance is still far from that of supervised pre-training. In this study, we overcome this limitation by incorporating a "mix-and-match" (M&M) tuning stage in the self-supervision pipeline. The proposed approach is readily pluggable to many self-supervision methods and does not use more annotated samples than the original process. Yet, it is capable of boosting the performance of target image segmentation task to surpass fully-supervised pre-trained counterpart. The improvement is made possible by better harnessing the limited pixel-wise annotations in the target dataset. Specifically, we first introduce the "mix" stage, which sparsely samples and mixes patches from the target set to reflect rich and diverse local patch statistics of target images. A ‘match’ stage then forms a class-wise connected graph, which can be used to derive a strong triplet-based discriminative loss for finetuning the network. Our paradigm follows the standard practice in existing self-supervised studies and no extra data or label is required. With the proposed M&M approach, for the first time, a self-supervision method can achieve comparable or even better performance compared to its ImageNet pretrained counterpart on both PASCAL VOC2012 dataset and CityScapes dataset.

Semantic segmentation tasks can be well modeled by Markov Random Field (MRF). This paper addresses semantic segmentation by incorporating high-order relations and mixture of label contexts into MRF. Unlike previous works that optimized MRFs using iterative algorithm, we solve MRF by proposing a Convolutional Neural Network (CNN), namely Deep Parsing Network (DPN), which enables deterministic end-to-end computation in a single forward pass. Specifically, DPN extends a contemporary CNN to model unary terms and additional layers are devised to approximate the mean field (MF) algorithm for pairwise terms. It has several appealing properties. First, different from the recent works that required many iterations of MF during back-propagation, DPN is able to achieve high performance by approximating one iteration of MF. Second, DPN represents various types of pairwise terms, making many existing models as its special cases. Furthermore, pairwise terms in DPN provide a unified framework to encode rich contextual information in high-dimensional data, such as images and videos. Third, DPN makes MF easier to be parallelized and speeded up, thus enabling efficient inference. DPN is thoroughly evaluated on standard semantic image/video segmentation benchmarks, where a single DPN model yields state-of-the-art segmentation accuracies on PASCAL VOC 2012, Cityscapes dataset and CamVid dataset.

We propose a novel deep layer cascade (LC) method to improve the accuracy and speed of semantic segmentation. Unlike the conventional model cascade (MC) that is composed of multiple independent models, LC treats a single deep model as a cascade of several sub-models. Earlier sub-models are trained to handle easy and confident regions, and they progressively feed-forward harder regions to the next sub-model for processing. Convolutions are only calculated on these regions to reduce computations. The proposed method possesses several advantages. First, LC classifies most of the easy regions in the shallow stage and makes deeper stage focuses on a few hard regions. Such an adaptive and difficulty-aware learning improves segmentation performance. Second, LC accelerates both training and testing of deep network thanks to early decisions in the shallow stage. Third, in comparison to MC, LC is an end-to-end trainable framework, allowing joint learning of all sub-models. We evaluate our method on PASCAL VOC and Cityscapes datasets, achieving state-of-the-art performance and fast speed.

We address the problem of synthesizing new video frames in an existing video, either in-between existing frames (interpolation), or subsequent to them (extrapolation). This problem is challenging because video appearance and motion can be highly complex. Traditional optical-flow-based solutions often fail where flow estimation is challenging, while newer neural-network-based methods that hallucinate pixel values directly often produce blurry results. We combine the advantages of these two methods by training a deep network that learns to synthesize video frames by flowing pixel values from existing ones, which we call deep voxel flow. Our method requires no human supervision, and any video can be used as training data by dropping, and then learning to predict, existing frames. The technique is efficient, and can be applied at any video resolution. We demonstrate that our method produces results that both quantitatively and qualitatively improve upon the state-of-the-art.

Fashion landmarks are functional key points defined on clothes, such as corners of neckline, hemline, and cuff. They have been recently introduced [18]as an effective visual representation for fashion image understanding. However, detecting fashion landmarks are challenging due to background clutters, human poses, and scales. To remove the above variations, previous works usually assumed bounding boxes of clothes are provided in training and test as additional annotations, which are expensive to obtain and inapplicable in practice. This work addresses unconstrained fashion landmark detection, where clothing bounding boxes are not provided in both training and test. To this end, we present a novel Deep LAndmark Network (DLAN), where bounding boxes and landmarks are jointly estimated and trained iteratively in an end-to-end manner. DLAN contains two dedicated modules, including a Selective Dilated Convolution for handling scale discrepancies, and a Hierarchical Recurrent Spatial Transformer for handling background clutters. To evaluate DLAN, we present a large-scale fashion landmark dataset, namely Unconstrained Landmark Database (ULD), consisting of 30K images. Statistics show that ULD is more challenging than existing datasets in terms of image scales, background clutters, and human poses. Extensive experiments demonstrate the effectiveness of DLAN over the state-of-the-art methods. DLAN also exhibits excellent generalization across different clothing categories and modalities, making it extremely suitable for real-world fashion analysis.

The recent advanced face recognition systems werebuilt on large Deep Neural Networks (DNNs) or theirensembles, which have millions of parameters. However, the expensive computation of DNNs make theirdeployment difficult on mobile and embedded devices. This work addresses model compression for face recognition,where the learned knowledge of a large teachernetwork or its ensemble is utilized as supervisionto train a compact student network. Unlike previousworks that represent the knowledge by the soften labelprobabilities, which are difficult to fit, we represent theknowledge by using the neurons at the higher hiddenlayer, which preserve as much information as the label probabilities, but are more compact. By leveragingthe essential characteristics (domain knowledge) of thelearned face representation, a neuron selection methodis proposed to choose neurons that are most relevant toface recognition. Using the selected neurons as supervisionto mimic the single networks of DeepID2+ andDeepID3, which are the state-of-the-art face recognition systems, a compact student with simple network structure achieves better verification accuracy on LFW than its teachers, respectively. When using an ensemble of DeepID2+ as teacher, a mimicked student is able to outperform it and achieves 51.6 times compression ratio and 90 times speed-up in inference, making this cumbersome model applicable on portable devices.

Recent advances in clothes recognition have been driven by the construction of clothes datasets. Existing datasets are limited in the amount of annotations and are difficult to cope with the various challenges in real-world applications. In this work, we introduce DeepFashion1, a large-scale clothes dataset with comprehensive annotations. It contains over 800,000 images, which are richly annotated with massive attributes, clothing landmarks, and correspondence of images taken under different scenarios including store, street snapshot, and consumer. Such rich annotations enable the development of powerful algorithms in clothes recognition and facilitating future researches. To demonstrate the advantages of DeepFashion, we propose a new deep model, namely FashionNet, which learns clothing features by jointly predicting clothing attributes and landmarks. The estimated landmarks are then employed to pool or gate the learned features. It is optimized in an iterative manner. Extensive experiments demonstrate the effectiveness of FashionNet and the usefulness of DeepFashion.

This paper addresses semantic image segmentation by incorporating rich information into Markov Random Field (MRF), including high-order relations and mixture of label contexts. Unlike previous works that optimized MRFs using iterative algorithm, we solve MRF by proposing a Convolutional Neural Network (CNN), namely Deep Parsing Network (DPN), which enables deterministic end-to-end computation in a single forward pass. Specifically, DPN extends a contemporary CNN architecture to model unary terms and additional layers are carefully devised to approximate the mean field algorithm (MF) for pairwise terms. It has several appealing properties. First, different from the recent works that combined CNN and MRF, where many iterations of MF were required for each training image during back-propagation, DPN is able to achieve high performance by approximating one iteration of MF. Second, DPN represents various types of pairwise terms, making many existing works as its special cases. Third, DPN makes MF easier to be parallelized and speeded up in Graphical Processing Unit (GPU). DPN is thoroughly evaluated on the PASCAL VOC 2012 dataset, where a single DPN model yields a new state-of-the-art segmentation accuracy of 77.5%.

Predicting face attributes in the wild is challenging due to complex face variations. We propose a novel deep learning framework for attribute prediction in the wild. It cascades two CNNs, LNet and ANet, which are fine-tuned jointly with attribute tags, but pre-trained differently. LNet is pre-trained by massive general object categories for face localization, while ANet is pre-trained by massive face identities for attribute prediction. This framework not only outperforms the state-of-the-art with a large margin, but also reveals valuable facts on learning face representation. (1) It shows how the performances of face localization (LNet) and attribute prediction (ANet) can be improved by different pre-training strategies. (2) It reveals that although the filters of LNet are fine-tuned only with image-level attribute tags, their response maps over entire images have strong indication of face locations. This fact enables training LNet for face localization with only image-level annotations, but without face bounding boxes or landmarks, which are required by all attribute recognition works. (3) It also demonstrates that the high-level hidden neurons of ANet automatically discover semantic concepts after pre-training with massive face identities, and such concepts are significantly enriched after fine-tuning with attribute tags. Each attribute can be well explained with a sparse linear combination of these concepts.

This paper presents a fast denoising method that produces a clean image from a burst of noisy images. We accelerate alignment of the images by introducing a lightweight camera motion representation called homography flow . The aligned images are then fused to create a denoised output with rapid per-pixel operations in temporal and spatial domains. To handle scene motion during the capture, a mechanism of selecting consistent pixels for temporal fusion is proposed to "synthesize" a clean, ghost-free image, which can largely reduce the computation of tracking motion between frames. Combined with these efficient solutions, our method runs several orders of magnitude faster than previous work, while the denoising quality is comparable. A smartphone prototype demonstrates that our method is practical and works well on a large variety of real examples.