Liefeng Bo

Cihan Xiao, Yiwen Shao, Chenxing Li et al.

Audio and omni-modal large language models exhibit impressive cross-modal reasoning capabilities. However, applying standard reinforcement learning post-training algorithms to these models exposes a critical structural vulnerability: methods like GRPO apply uniform policy gradients across all tokens, ignoring their unequal dependence on the non-text source modality. This exacerbates late-stage modality collapse during extended chain-of-thought generation, where models progressively abandon the primary source signal in favor of compressed textual priors, leading to confident but ungrounded hallucinations. To address this, we introduce Modality-Aware Policy Optimization (MAPO), a novel dual-branch reinforcement learning framework. First, MAPO dynamically concentrates the policy gradient on modality-critical tokens using a modality relevance mask, which is derived from the cross-modal differential entropy between an audio-ablated reference and the multimodal policy. Second, it integrates an auxiliary attention loss branch that applies a targeted, temporally scaled penalty to the model's internal attention distributions. This ensures the model actively sustains cross-modal grounding deep into the reasoning trace. Evaluations on complex audio reasoning benchmarks demonstrate that MAPO substantially improves long-horizon reasoning fidelity and multimodal instruction following, achieving highly competitive performance and setting new state-of-the-art results on several key benchmarks among open-weight models. By relying strictly on native statistical signals rather than domain-specific inductive biases, MAPO offers a promising foundation for mitigating epistemic collapse across diverse multimodal systems.

Lingzhi Li, Zhen Shen, Zhongshu Wang et al.

Approximating radiance fields with volumetric grids is one of promising directions for improving NeRF, represented by methods like Plenoxels and DVGO, which achieve super-fast training convergence and real-time rendering. However, these methods typically require a tremendous storage overhead, costing up to hundreds of megabytes of disk space and runtime memory for a single scene. We address this issue in this paper by introducing a simple yet effective framework, called vector quantized radiance fields (VQRF), for compressing these volume-grid-based radiance fields. We first present a robust and adaptive metric for estimating redundancy in grid models and performing voxel pruning by better exploring intermediate outputs of volumetric rendering. A trainable vector quantization is further proposed to improve the compactness of grid models. In combination with an efficient joint tuning strategy and post-processing, our method can achieve a compression ratio of 100$\times$ by reducing the overall model size to 1 MB with negligible loss on visual quality. Extensive experiments demonstrate that the proposed framework is capable of achieving unrivaled performance and well generalization across multiple methods with distinct volumetric structures, facilitating the wide use of volumetric radiance fields methods in real-world applications. Code Available at \url{https://github.com/AlgoHunt/VQRF}

Zhongshu Wang, Lingzhi Li, Zhen Shen et al.

In this paper, we present a novel and effective framework, named 4K-NeRF, to pursue high fidelity view synthesis on the challenging scenarios of ultra high resolutions, building on the methodology of neural radiance fields (NeRF). The rendering procedure of NeRF-based methods typically relies on a pixel-wise manner in which rays (or pixels) are treated independently on both training and inference phases, limiting its representational ability on describing subtle details, especially when lifting to a extremely high resolution. We address the issue by exploring ray correlation to enhance high-frequency details recovery. Particularly, we use the 3D-aware encoder to model geometric information effectively in a lower resolution space and recover fine details through the 3D-aware decoder, conditioned on ray features and depths estimated by the encoder. Joint training with patch-based sampling further facilitates our method incorporating the supervision from perception oriented regularization beyond pixel-wise loss. Benefiting from the use of geometry-aware local context, our method can significantly boost rendering quality on high-frequency details compared with modern NeRF methods, and achieve the state-of-the-art visual quality on 4K ultra-high-resolution scenarios. Code Available at \url{https://github.com/frozoul/4K-NeRF}

Qihang Fang, Yafei Song, Keqiang Li et al.

A radiance field is an effective representation of 3D scenes, which has been widely adopted in novel-view synthesis and 3D reconstruction. It is still an open and challenging problem to evaluate the geometry, i.e., the density field, as the ground-truth is almost impossible to obtain. One alternative indirect solution is to transform the density field into a point-cloud and compute its Chamfer Distance with the scanned ground-truth. However, many widely-used datasets have no point-cloud ground-truth since the scanning process along with the equipment is expensive and complicated. To this end, we propose a novel metric, named Inverse Mean Residual Color (IMRC), which can evaluate the geometry only with the observation images. Our key insight is that the better the geometry, the lower-frequency the computed color field. From this insight, given a reconstructed density field and observation images, we design a closed-form method to approximate the color field with low-frequency spherical harmonics, and compute the inverse mean residual color. Then the higher the IMRC, the better the geometry. Qualitative and quantitative experimental results verify the effectiveness of our proposed IMRC metric. We also benchmark several state-of-the-art methods using IMRC to promote future related research. Our code is available at https://github.com/qihangGH/IMRC.

Weichuang Li, Longhao Zhang, Dong Wang et al.

Talking head generation aims to generate faces that maintain the identity information of the source image and imitate the motion of the driving image. Most pioneering methods rely primarily on 2D representations and thus will inevitably suffer from face distortion when large head rotations are encountered. Recent works instead employ explicit 3D structural representations or implicit neural rendering to improve performance under large pose changes. Nevertheless, the fidelity of identity and expression is not so desirable, especially for novel-view synthesis. In this paper, we propose HiDe-NeRF, which achieves high-fidelity and free-view talking-head synthesis. Drawing on the recently proposed Deformable Neural Radiance Fields, HiDe-NeRF represents the 3D dynamic scene into a canonical appearance field and an implicit deformation field, where the former comprises the canonical source face and the latter models the driving pose and expression. In particular, we improve fidelity from two aspects: (i) to enhance identity expressiveness, we design a generalized appearance module that leverages multi-scale volume features to preserve face shape and details; (ii) to improve expression preciseness, we propose a lightweight deformation module that explicitly decouples the pose and expression to enable precise expression modeling. Extensive experiments demonstrate that our proposed approach can generate better results than previous works. Project page: https://www.waytron.net/hidenerf/

Shuyuan Tu, Qi Tian, Zihan Yang et al.

Current open-source diffusion models struggle to generate stable and synchronized audio-visual content, particularly in scenarios demanding complex semantic reasoning. The root cause is that existing methods rely on coarse text embeddings from off-the-shelf encoders to guide audio-video denoising, which discards fine-grained semantics and, critically, lacks a shared long-horizon plan, leading to uncoordinated denoising trajectories and fragile cross-modal alignment. We propose Baton, the first framework that introduces explicit semantic planning into joint video-audio generation. Our key insight is that complementing coarse text guidance with semantically rich, modality-aware planned tokens, jointly reasoned and mutually aligned before denoising, can simultaneously restore fine-grained semantic detail and establish a shared blueprint that coordinates both audio and video denoising trajectories. Concretely, Baton first introduces the VA-Planner, a multimodal language model equipped with dual semantic alignment towers, where learnable queries cross-attend to both video and audio features to produce a pair of semantically aligned video and audio planned tokens as keyframe-level blueprints. These planned tokens are injected into the diffusion backbone via cross-attention layers, providing temporally grounded guidance complementary to coarse text embeddings. Since planned tokens do not share one-to-one spatial-temporal correspondence with diffusion latents, we further propose Relative Semantic RoPE, a relative positional encoding that maps planned tokens and latents into a shared spatial-temporal coordinate frame, enabling each latent to accurately attend to its positionally corresponding semantic cues. Experiments on benchmarks show the effectiveness of Baton both qualitatively and quantitatively.

Kaihang Pan, Qi Tian, Jianwei Zhang et al.

While proprietary systems such as Seedance-2.0 have achieved remarkable success in omni-capable video generation, open-source alternatives significantly lag behind. Most academic models remain heavily fragmented, and the few existing efforts toward unified video generation still struggle to seamlessly integrate diverse tasks within a single framework. To bridge this gap, we propose OmniWeaving, an omni-level video generation model featuring powerful multimodal composition and reasoning-informed capabilities. By leveraging a massive-scale pretraining dataset that encompasses diverse compositional and reasoning-augmented scenarios, OmniWeaving learns to temporally bind interleaved text, multi-image, and video inputs while acting as an intelligent agent to infer complex user intentions for sophisticated video creation. Furthermore, we introduce IntelligentVBench, the first comprehensive benchmark designed to rigorously assess next-level intelligent unified video generation. Extensive experiments demonstrate that OmniWeaving achieves SoTA performance among open-source unified models. The codes and model will be made publicly available soon. Project Page: https://omniweaving.github.io.

Weihao Yuan, Weichao Shen, Yisheng He et al.

Motion generation from discrete quantization offers many advantages over continuous regression, but at the cost of inevitable approximation errors. Previous methods usually quantize the entire body pose into one code, which not only faces the difficulty in encoding all joints within one vector but also loses the spatial relationship between different joints. Differently, in this work we quantize each individual joint into one vector, which i) simplifies the quantization process as the complexity associated with a single joint is markedly lower than that of the entire pose; ii) maintains a spatial-temporal structure that preserves both the spatial relationships among joints and the temporal movement patterns; iii) yields a 2D token map, which enables the application of various 2D operations widely used in 2D images. Grounded in the 2D motion quantization, we build a spatial-temporal modeling framework, where 2D joint VQVAE, temporal-spatial 2D masking technique, and spatial-temporal 2D attention are proposed to take advantage of spatial-temporal signals among the 2D tokens. Extensive experiments demonstrate that our method significantly outperforms previous methods across different datasets, with a 26.6% decrease of FID on HumanML3D and a 29.9% decrease on KIT-ML. Project page: https://aigc3d.github.io/mogents.

Lingteng Qiu, Guanying Chen, Xiaodong Gu et al.

Lifting 2D diffusion for 3D generation is a challenging problem due to the lack of geometric prior and the complex entanglement of materials and lighting in natural images. Existing methods have shown promise by first creating the geometry through score-distillation sampling (SDS) applied to rendered surface normals, followed by appearance modeling. However, relying on a 2D RGB diffusion model to optimize surface normals is suboptimal due to the distribution discrepancy between natural images and normals maps, leading to instability in optimization. In this paper, recognizing that the normal and depth information effectively describe scene geometry and be automatically estimated from images, we propose to learn a generalizable Normal-Depth diffusion model for 3D generation. We achieve this by training on the large-scale LAION dataset together with the generalizable image-to-depth and normal prior models. In an attempt to alleviate the mixed illumination effects in the generated materials, we introduce an albedo diffusion model to impose data-driven constraints on the albedo component. Our experiments show that when integrated into existing text-to-3D pipelines, our models significantly enhance the detail richness, achieving state-of-the-art results. Our project page is https://aigc3d.github.io/richdreamer/.

Akash Gupta, Jingen Liu, Liefeng Bo et al.

While action anticipation has garnered a lot of research interest recently, most of the works focus on anticipating future action directly through observed visual cues only. In this work, we take a step back to analyze how the human capability to anticipate the future can be transferred to machine learning algorithms. To incorporate this ability in intelligent systems a question worth pondering upon is how exactly do we anticipate? Is it by anticipating future actions from past experiences? Or is it by simulating possible scenarios based on cues from the present? A recent study on human psychology explains that, in anticipating an occurrence, the human brain counts on both systems. In this work, we study the impact of each system for the task of action anticipation and introduce a paradigm to integrate them in a learning framework. We believe that intelligent systems designed by leveraging the psychological anticipation models will do a more nuanced job at the task of human action prediction. Furthermore, we introduce cyclic transformation in the temporal dimension in feature and semantic label space to instill the human ability of reasoning of past actions based on the predicted future. Experiments on Epic-Kitchen, Breakfast, and 50Salads dataset demonstrate that the action anticipation model learned using a combination of the two systems along with the cycle transformation performs favorably against various state-of-the-art approaches.

Li Hu, Xin Gao, Peng Zhang et al.

Character Animation aims to generating character videos from still images through driving signals. Currently, diffusion models have become the mainstream in visual generation research, owing to their robust generative capabilities. However, challenges persist in the realm of image-to-video, especially in character animation, where temporally maintaining consistency with detailed information from character remains a formidable problem. In this paper, we leverage the power of diffusion models and propose a novel framework tailored for character animation. To preserve consistency of intricate appearance features from reference image, we design ReferenceNet to merge detail features via spatial attention. To ensure controllability and continuity, we introduce an efficient pose guider to direct character's movements and employ an effective temporal modeling approach to ensure smooth inter-frame transitions between video frames. By expanding the training data, our approach can animate arbitrary characters, yielding superior results in character animation compared to other image-to-video methods. Furthermore, we evaluate our method on benchmarks for fashion video and human dance synthesis, achieving state-of-the-art results.

Daiheng Gao, Xu Chen, Xindi Zhang et al.

Fabricating and designing 3D garments has become extremely demanding with the increasing need for synthesizing realistic dressed persons for a variety of applications, e.g. 3D virtual try-on, digitalization of 2D clothes into 3D apparel, and cloth animation. It thus necessitates a simple and straightforward pipeline to obtain high-quality texture from simple input, such as 2D reference images. Since traditional warping-based texture generation methods require a significant number of control points to be manually selected for each type of garment, which can be a time-consuming and tedious process. We propose a novel method, called Cloth2Tex, which eliminates the human burden in this process. Cloth2Tex is a self-supervised method that generates texture maps with reasonable layout and structural consistency. Another key feature of Cloth2Tex is that it can be used to support high-fidelity texture inpainting. This is done by combining Cloth2Tex with a prevailing latent diffusion model. We evaluate our approach both qualitatively and quantitatively and demonstrate that Cloth2Tex can generate high-quality texture maps and achieve the best visual effects in comparison to other methods. Project page: tomguluson92.github.io/projects/cloth2tex/

Bin Lin, Zongjian Li, Yuwei Niu et al.

The field of image generation is currently bifurcated into autoregressive (AR) models operating on discrete tokens and diffusion models utilizing continuous latents. This divide, rooted in the distinction between VQ-VAEs and VAEs, hinders unified modeling and fair benchmarking. Finite Scalar Quantization (FSQ) offers a theoretical bridge, yet vanilla FSQ suffers from a critical flaw: its equal-interval quantization can cause activation collapse. This mismatch forces a trade-off between reconstruction fidelity and information efficiency. In this work, we resolve this dilemma by simply replacing the activation function in original FSQ with a distribution-matching mapping to enforce a uniform prior. Termed iFSQ, this simple strategy requires just one line of code yet mathematically guarantees both optimal bin utilization and reconstruction precision. Leveraging iFSQ as a controlled benchmark, we uncover two key insights: (1) The optimal equilibrium between discrete and continuous representations lies at approximately 4 bits per dimension. (2) Under identical reconstruction constraints, AR models exhibit rapid initial convergence, whereas diffusion models achieve a superior performance ceiling, suggesting that strict sequential ordering may limit the upper bounds of generation quality. Finally, we extend our analysis by adapting Representation Alignment (REPA) to AR models, yielding LlamaGen-REPA. Codes is available at https://github.com/Tencent-Hunyuan/iFSQ

Yifang Men, Yuan Yao, Miaomiao Cui et al.

Character video synthesis aims to produce realistic videos of animatable characters within lifelike scenes. As a fundamental problem in the computer vision and graphics community, 3D works typically require multi-view captures for per-case training, which severely limits their applicability of modeling arbitrary characters in a short time. Recent 2D methods break this limitation via pre-trained diffusion models, but they struggle for pose generality and scene interaction. To this end, we propose MIMO, a novel framework which can not only synthesize character videos with controllable attributes (i.e., character, motion and scene) provided by simple user inputs, but also simultaneously achieve advanced scalability to arbitrary characters, generality to novel 3D motions, and applicability to interactive real-world scenes in a unified framework. The core idea is to encode the 2D video to compact spatial codes, considering the inherent 3D nature of video occurrence. Concretely, we lift the 2D frame pixels into 3D using monocular depth estimators, and decompose the video clip to three spatial components (i.e., main human, underlying scene, and floating occlusion) in hierarchical layers based on the 3D depth. These components are further encoded to canonical identity code, structured motion code and full scene code, which are utilized as control signals of synthesis process. The design of spatial decomposed modeling enables flexible user control, complex motion expression, as well as 3D-aware synthesis for scene interactions. Experimental results demonstrate effectiveness and robustness of the proposed method.

Ke Sun, Jian Cao, Qi Wang et al.

Virtual Try-On (VTON) has become a transformative technology, empowering users to experiment with fashion without ever having to physically try on clothing. However, existing methods often struggle with generating high-fidelity and detail-consistent results. While diffusion models, such as Stable Diffusion series, have shown their capability in creating high-quality and photorealistic images, they encounter formidable challenges in conditional generation scenarios like VTON. Specifically, these models struggle to maintain a balance between control and consistency when generating images for virtual clothing trials. OutfitAnyone addresses these limitations by leveraging a two-stream conditional diffusion model, enabling it to adeptly handle garment deformation for more lifelike results. It distinguishes itself with scalability-modulating factors such as pose, body shape and broad applicability, extending from anime to in-the-wild images. OutfitAnyone's performance in diverse scenarios underscores its utility and readiness for real-world deployment. For more details and animated results, please see \url{https://humanaigc.github.io/outfit-anyone/}.

Jade Zou, Tao Huang, Weijie Kong et al.

Reinforcement learning (RL) has become an effective way to improve prompt alignment and perceptual quality in diffusion and flow-matching generators. A critical step for applying online RL to flow matching is turning the deterministic sampling trajectory into a stochastic policy, typically by replacing the reverse-time Ordinary Differential Equation (ODE) with a Stochastic Differential Equation (SDE). The stochastic sampler, controlling the exploration behavior and denoising dynamics, is thus part of the policy, and its design can significantly affect the reward optimization performance. We break down the sampler design into two interdependent components: choosing the right amount of stochastic exploration, and discretizing the resulting SDE faithfully at the small step counts used in RL. To address the first component, we analyze the inherent tension between exploration and stability in denoising and derive an SDE schedule that balances the two. Turning to the discretization challenge, we use a toy example to show that existing samplers can deviate from the flow-matching process, either by introducing excessive discretization noise or by relying on heuristic rules that do not guarantee convergence to the data distribution. To address these issues, we propose Precise, a new stochastic sampler that balances effective exploration with stability. Crucially, Precise keeps the denoising trajectory SDE-consistent through a novel approximation that freezes the clean-latent posterior mean, resolving the excess noise issue in standard samplers. Extensive experiments demonstrate that this formulation leads to significantly faster and more stable reward optimization via reinforcement learning, achieving state-of-the-art alignment scores (e.g., PickScore, HPSv2.1) while requiring 13.1-53.2% less wall-clock training time to match the best in-domain performance of prior samplers.

Jiaxing Zhao, Qize Yang, Yixing Peng et al.

In human-centric scenes, the ability to simultaneously understand visual and auditory information is crucial. While recent omni models can process multiple modalities, they generally lack effectiveness in human-centric scenes due to the absence of large-scale, specialized datasets and non-targeted architectures. In this work, we developed HumanOmni, the industry's first human-centric Omni-multimodal large language model. We constructed a dataset containing over 2.4 million human-centric video clips with detailed captions and more than 14 million instructions, facilitating the understanding of diverse human-centric scenes. HumanOmni includes three specialized branches for understanding different types of scenes. It adaptively fuses features from these branches based on user instructions, significantly enhancing visual understanding in scenes centered around individuals. Moreover, HumanOmni integrates audio features to ensure a comprehensive understanding of environments and individuals. Our experiments validate HumanOmni's advanced capabilities in handling human-centric scenes across a variety of tasks, including emotion recognition, facial expression description, and action understanding. Our model will be open-sourced to facilitate further development and collaboration within both academia and industry.

Junjie He, Yifeng Geng, Liefeng Bo

This paper presents UniPortrait, an innovative human image personalization framework that unifies single- and multi-ID customization with high face fidelity, extensive facial editability, free-form input description, and diverse layout generation. UniPortrait consists of only two plug-and-play modules: an ID embedding module and an ID routing module. The ID embedding module extracts versatile editable facial features with a decoupling strategy for each ID and embeds them into the context space of diffusion models. The ID routing module then combines and distributes these embeddings adaptively to their respective regions within the synthesized image, achieving the customization of single and multiple IDs. With a carefully designed two-stage training scheme, UniPortrait achieves superior performance in both single- and multi-ID customization. Quantitative and qualitative experiments demonstrate the advantages of our method over existing approaches as well as its good scalability, e.g., the universal compatibility with existing generative control tools. The project page is at https://aigcdesigngroup.github.io/UniPortrait-Page/ .

You Qin, Linqing Wang, Hao Fei et al.

The post-training pipeline for diffusion models currently has two stages: supervised fine-tuning (SFT) on curated data and reinforcement learning (RL) with reward models. A fundamental gap separates them. SFT optimizes the denoiser only on ground-truth states sampled from the forward noising process; once inference deviates from these ideal states, subsequent denoising relies on out-of-distribution generalization rather than learned correction, exhibiting the same exposure bias that afflicts autoregressive models, but accumulated along the denoising trajectory instead of the token sequence. RL can in principle address this mismatch, yet its terminal reward signal is sparse, suffers from credit-assignment difficulty, and risks reward hacking. We propose SOAR (Self-Correction for Optimal Alignment and Refinement), a bias-correction post-training method that fills this gap. Starting from a real sample, SOAR performs a single stop-gradient rollout with the current model, re-noises the resulting off-trajectory state, and supervises the model to steer back toward the original clean target. The method is on-policy, reward-free, and provides dense per-timestep supervision with no credit-assignment problem. On SD3.5-Medium, SOAR improves GenEval from 0.70 to 0.78 and OCR from 0.64 to 0.67 over SFT, while simultaneously raising all model-based preference scores. In controlled reward-specific experiments, SOAR surpasses Flow-GRPO in final metric value on both aesthetic and text-image alignment tasks, despite having no access to a reward model. Since SOAR's base loss subsumes the standard SFT objective, it can directly replace SFT as a stronger first post-training stage after pretraining, while remaining fully compatible with subsequent RL alignment.

Yuxiang Tuo, Yifeng Geng, Liefeng Bo

As the text-to-image (T2I) domain progresses, generating text that seamlessly integrates with visual content has garnered significant attention. However, even with accurate text generation, the inability to control font and color can greatly limit certain applications, and this issue remains insufficiently addressed. This paper introduces AnyText2, a novel method that enables precise control over multilingual text attributes in natural scene image generation and editing. Our approach consists of two main components. First, we propose a WriteNet+AttnX architecture that injects text rendering capabilities into a pre-trained T2I model. Compared to its predecessor, AnyText, our new approach not only enhances image realism but also achieves a 19.8% increase in inference speed. Second, we explore techniques for extracting fonts and colors from scene images and develop a Text Embedding Module that encodes these text attributes separately as conditions. As an extension of AnyText, this method allows for customization of attributes for each line of text, leading to improvements of 3.3% and 9.3% in text accuracy for Chinese and English, respectively. Through comprehensive experiments, we demonstrate the state-of-the-art performance of our method. The code and model will be made open-source in https://github.com/tyxsspa/AnyText2.

Xiangyue Liu, Zijian Zhang, Miles Yang et al.

Empowering Large Multimodal Models (LMMs) with image generation often leads to catastrophic forgetting in understanding tasks due to severe gradient conflicts. While existing paradigms like Mixture-of-Transformers (MoT) mitigate this conflict through structural isolation, they fundamentally sever cross-modal synergy and suffer from capacity fragmentation. In this work, we present Symbiotic-MoE, a unified pre-training framework that resolves task interference within a native multimodal Mixture-of-Experts (MoE) Transformers architecture with zero-parameter overhead. We first identify that standard MoE tuning leads to routing collapse, where generative gradients dominate expert utilization. To address this, we introduce Modality-Aware Expert Disentanglement, which partitions experts into task-specific groups while utilizing shared experts as a multimodal semantic bridge. Crucially, this design allows shared experts to absorb fine-grained visual semantics from generative tasks to enrich textual representations. To optimize this, we propose a Progressive Training Strategy featuring differential learning rates and early-stage gradient shielding. This mechanism not only shields pre-trained knowledge from early volatility but eventually transforms generative signals into constructive feedback for understanding. Extensive experiments demonstrate that Symbiotic-MoE achieves rapid generative convergence while unlocking cross-modal synergy, boosting inherent understanding with remarkable gains on MMLU and OCRBench.

Xuerui Qiu, Yutao Cui, Guozhen Zhang et al.

Unified Multimodal Models struggle to bridge the fundamental gap between the abstract representations needed for visual understanding and the detailed primitives required for generation. Existing approaches typically compromise by employing decoupled encoders, stacking representation encoder atop VAEs, or utilizing discrete quantization. However, these methods often disrupt information coherence and lead to optimization conflicts. To this end, we introduce HYDRA-TOK, a representation-harmonized pure ViT in the insight that visual modeling should evolve from generation to understanding. HYDRA-TOK reformulates the standard backbone into a progressive learner that transitions from a Gen-ViT, which captures structure-preserving primitives, to a Sem-ViT for semantic encoding. Crucially, this transition is mediated by a Generation-Semantic Bottleneck (GSB), which compresses features into a low-dimensional space to filter noise for robust synthesis, then restores dimensionality to empower complex semantic comprehension. Built upon this foundation, we present HYDRA, a native unified framework integrating perception and generation within a single parameter space. Extensive experiments establish HYDRA as a new state-of-the-art. It sets a benchmark in visual reconstruction (rFID 0.08) and achieves top-tier generation performance on GenEval (0.86), DPG-Bench (86.4), and WISE (0.53), while simultaneously outperforming previous native UMMs by an average of 10.0 points across eight challenging understanding benchmarks.

Qihang Fang, Yafei Song, Keqiang Li et al.

Neural radiance field (NeRF) enables the synthesis of cutting-edge realistic novel view images of a 3D scene. It includes density and color fields to model the shape and radiance of a scene, respectively. Supervised by the photometric loss in an end-to-end training manner, NeRF inherently suffers from the shape-radiance ambiguity problem, i.e., it can perfectly fit training views but does not guarantee decoupling the two fields correctly. To deal with this issue, existing works have incorporated prior knowledge to provide an independent supervision signal for the density field, including total variation loss, sparsity loss, distortion loss, etc. These losses are based on general assumptions about the density field, e.g., it should be smooth, sparse, or compact, which are not adaptive to a specific scene. In this paper, we propose a more adaptive method to reduce the shape-radiance ambiguity. The key is a rendering method that is only based on the density field. Specifically, we first estimate the color field based on the density field and posed images in a closed form. Then NeRF's rendering process can proceed. We address the problems in estimating the color field, including occlusion and non-uniformly distributed views. Afterward, it is applied to regularize NeRF's density field. As our regularization is guided by photometric loss, it is more adaptive compared to existing ones. Experimental results show that our method improves the density field of NeRF both qualitatively and quantitatively. Our code is available at https://github.com/qihangGH/Closed-form-color-field.

Yaqi Zhao, Wang Lin, Zijian Zhang et al.

Current unified multimodal models typically rely on discrete visual tokenizers to bridge the modality gap. However, discretization inevitably discards fine-grained semantic information, leading to suboptimal performance in visual understanding tasks. Conversely, directly modeling continuous semantic representations (e.g., CLIP, SigLIP) poses significant challenges in high-dimensional generative modeling, resulting in slow convergence and training instability. To resolve this dilemma, we introduce UniCom, a unified framework that harmonizes multimodal understanding and generation via compressed continuous representation. We empirically demonstrate that reducing channel dimension is significantly more effective than spatial downsampling for both reconstruction and generation. Accordingly, we design an attention-based semantic compressor to distill dense features into a compact unified representation. Furthermore, we validate that the transfusion architecture surpasses query-based designs in convergence and consistency. Experiments demonstrate that UniCom achieves state-of-the-art generation performance among unified models. Notably, by preserving rich semantic priors, it delivers exceptional controllability in image editing and maintains image consistency even without relying on VAE.

Kangneng Zhou, Daiheng Gao, Xuan Wang et al.

3D-aware portrait editing has a wide range of applications in multiple fields. However, current approaches are limited due that they can only perform mask-guided or text-based editing. Even by fusing the two procedures into a model, the editing quality and stability cannot be ensured. To address this limitation, we propose \textbf{MaTe3D}: mask-guided text-based 3D-aware portrait editing. In this framework, first, we introduce a new SDF-based 3D generator which learns local and global representations with proposed SDF and density consistency losses. This enhances masked-based editing in local areas; second, we present a novel distillation strategy: Conditional Distillation on Geometry and Texture (CDGT). Compared to exiting distillation strategies, it mitigates visual ambiguity and avoids mismatch between texture and geometry, thereby producing stable texture and convincing geometry while editing. Additionally, we create the CatMask-HQ dataset, a large-scale high-resolution cat face annotation for exploration of model generalization and expansion. We perform expensive experiments on both the FFHQ and CatMask-HQ datasets to demonstrate the editing quality and stability of the proposed method. Our method faithfully generates a 3D-aware edited face image based on a modified mask and a text prompt. Our code and models will be publicly released.

Mingzhe Zheng, Weijie Kong, Yue Wu et al.

Group Relative Policy Optimization (GRPO) methods for video generation like FlowGRPO remain far less reliable than their counterparts for language models and images. This gap arises because video generation has a complex solution space, and the ODE-to-SDE conversion used for exploration can inject excess noise, lowering rollout quality and making reward estimates less reliable, which destabilizes post-training alignment. To address this problem, we view the pre-trained model as defining a valid video data manifold and formulate the core problem as constraining exploration within the vicinity of this manifold, ensuring that rollout quality is preserved and reward estimates remain reliable. We propose SAGE-GRPO (Stable Alignment via Exploration), which applies constraints at both micro and macro levels. At the micro level, we derive a precise manifold-aware SDE with a logarithmic curvature correction and introduce a gradient norm equalizer to stabilize sampling and updates across timesteps. At the macro level, we use a dual trust region with a periodic moving anchor and stepwise constraints so that the trust region tracks checkpoints that are closer to the manifold and limits long-horizon drift. We evaluate SAGE-GRPO on HunyuanVideo1.5 using the original VideoAlign as the reward model and observe consistent gains over previous methods in VQ, MQ, TA, and visual metrics (CLIPScore, PickScore), demonstrating superior performance in both reward maximization and overall video quality. The code and visual gallery are available at https://dungeonmassster.github.io/SAGE-GRPO-Page/.

Juncheng Wang, Chao Xu, Cheng Yu et al.

Video-to-audio generation is essential for synthesizing realistic audio tracks that synchronize effectively with silent videos. Following the perspective of extracting essential signals from videos that can precisely control the mature text-to-audio generative diffusion models, this paper presents how to balance the representation of mel-spectrograms in terms of completeness and complexity through a new approach called Mel Quantization-Continuum Decomposition (Mel-QCD). We decompose the mel-spectrogram into three distinct types of signals, employing quantization or continuity to them, we can effectively predict them from video by a devised video-to-all (V2X) predictor. Then, the predicted signals are recomposed and fed into a ControlNet, along with a textual inversion design, to control the audio generation process. Our proposed Mel-QCD method demonstrates state-of-the-art performance across eight metrics, evaluating dimensions such as quality, synchronization, and semantic consistency. Our codes and demos will be released at \href{Website}{https://wjc2830.github.io/MelQCD/}.

Yisheng He, Xiaodong Gu, Xiaodan Ye et al.

We present LAM, an innovative Large Avatar Model for animatable Gaussian head reconstruction from a single image. Unlike previous methods that require extensive training on captured video sequences or rely on auxiliary neural networks for animation and rendering during inference, our approach generates Gaussian heads that are immediately animatable and renderable. Specifically, LAM creates an animatable Gaussian head in a single forward pass, enabling reenactment and rendering without additional networks or post-processing steps. This capability allows for seamless integration into existing rendering pipelines, ensuring real-time animation and rendering across a wide range of platforms, including mobile phones. The centerpiece of our framework is the canonical Gaussian attributes generator, which utilizes FLAME canonical points as queries. These points interact with multi-scale image features through a Transformer to accurately predict Gaussian attributes in the canonical space. The reconstructed canonical Gaussian avatar can then be animated utilizing standard linear blend skinning (LBS) with corrective blendshapes as the FLAME model did and rendered in real-time on various platforms. Our experimental results demonstrate that LAM outperforms state-of-the-art methods on existing benchmarks. Our code and video are available at https://aigc3d.github.io/projects/LAM/

Linrui Tian, Qi Wang, Bang Zhang et al.

In this work, we tackle the challenge of enhancing the realism and expressiveness in talking head video generation by focusing on the dynamic and nuanced relationship between audio cues and facial movements. We identify the limitations of traditional techniques that often fail to capture the full spectrum of human expressions and the uniqueness of individual facial styles. To address these issues, we propose EMO, a novel framework that utilizes a direct audio-to-video synthesis approach, bypassing the need for intermediate 3D models or facial landmarks. Our method ensures seamless frame transitions and consistent identity preservation throughout the video, resulting in highly expressive and lifelike animations. Experimental results demonsrate that EMO is able to produce not only convincing speaking videos but also singing videos in various styles, significantly outperforming existing state-of-the-art methodologies in terms of expressiveness and realism.

Detao Bai, Zhiheng Ma, Xihan Wei et al.

The inherent synchronization between a speaker's lip movements, voice, and the underlying linguistic content offers a rich source of information for improving speech processing tasks, especially in challenging conditions where traditional audio-only systems falter. We introduce CoGenAV, a powerful and data-efficient model designed to learn versatile audio-visual representations applicable across a wide range of speech and audio-visual tasks. CoGenAV is trained by optimizing a dual objective derived from natural audio-visual synchrony, contrastive feature alignment and generative text prediction, using only 223 hours of labeled data from the LRS2 dataset. This contrastive-generative synchronization strategy effectively captures fundamental cross-modal correlations. We showcase the effectiveness and versatility of the learned CoGenAV representations on multiple benchmarks. When utilized for Audio-Visual Speech Recognition (AVSR) on LRS2, these representations contribute to achieving a state-of-the-art Word Error Rate (WER) of 1.27. They also enable strong performance in Visual Speech Recognition (VSR) with a WER of 20.5 on LRS2, and significantly improve performance in noisy environments by over 70%. Furthermore, CoGenAV representations benefit speech reconstruction tasks, boosting performance in Speech Enhancement and Separation, and achieve competitive results in audio-visual synchronization tasks like Active Speaker Detection (ASD). Our model will be open-sourced to facilitate further development and collaboration within both academia and industry.

Zichen Tang, Yuan Yao, Miaomiao Cui et al.

Text-guided 3D human generation has advanced with the development of efficient 3D representations and 2D-lifting methods like Score Distillation Sampling (SDS). However, current methods suffer from prolonged training times and often produce results that lack fine facial and garment details. In this paper, we propose GaussianIP, an effective two-stage framework for generating identity-preserving realistic 3D humans from text and image prompts. Our core insight is to leverage human-centric knowledge to facilitate the generation process. In stage 1, we propose a novel Adaptive Human Distillation Sampling (AHDS) method to rapidly generate a 3D human that maintains high identity consistency with the image prompt and achieves a realistic appearance. Compared to traditional SDS methods, AHDS better aligns with the human-centric generation process, enhancing visual quality with notably fewer training steps. To further improve the visual quality of the face and clothes regions, we design a View-Consistent Refinement (VCR) strategy in stage 2. Specifically, it produces detail-enhanced results of the multi-view images from stage 1 iteratively, ensuring the 3D texture consistency across views via mutual attention and distance-guided attention fusion. Then a polished version of the 3D human can be achieved by directly perform reconstruction with the refined images. Extensive experiments demonstrate that GaussianIP outperforms existing methods in both visual quality and training efficiency, particularly in generating identity-preserving results. Our code is available at: https://github.com/silence-tang/GaussianIP.

Lianghao Xia, Chao Huang, Yong Xu et al.

Crime prediction is crucial for public safety and resource optimization, yet is very challenging due to two aspects: i) the dynamics of criminal patterns across time and space, crime events are distributed unevenly on both spatial and temporal domains; ii) time-evolving dependencies between different types of crimes (e.g., Theft, Robbery, Assault, Damage) which reveal fine-grained semantics of crimes. To tackle these challenges, we propose Spatial-Temporal Sequential Hypergraph Network (ST-SHN) to collectively encode complex crime spatial-temporal patterns as well as the underlying category-wise crime semantic relationships. In specific, to handle spatial-temporal dynamics under the long-range and global context, we design a graph-structured message passing architecture with the integration of the hypergraph learning paradigm. To capture category-wise crime heterogeneous relations in a dynamic environment, we introduce a multi-channel routing mechanism to learn the time-evolving structural dependency across crime types. We conduct extensive experiments on two real-world datasets, showing that our proposed ST-SHN framework can significantly improve the prediction performance as compared to various state-of-the-art baselines. The source code is available at: https://github.com/akaxlh/ST-SHN.

Lianghao Xia, Chao Huang, Yong Xu et al.

Many previous studies aim to augment collaborative filtering with deep neural network techniques, so as to achieve better recommendation performance. However, most existing deep learning-based recommender systems are designed for modeling singular type of user-item interaction behavior, which can hardly distill the heterogeneous relations between user and item. In practical recommendation scenarios, there exist multityped user behaviors, such as browse and purchase. Due to the overlook of user's multi-behavioral patterns over different items, existing recommendation methods are insufficient to capture heterogeneous collaborative signals from user multi-behavior data. Inspired by the strength of graph neural networks for structured data modeling, this work proposes a Graph Neural Multi-Behavior Enhanced Recommendation (GNMR) framework which explicitly models the dependencies between different types of user-item interactions under a graph-based message passing architecture. GNMR devises a relation aggregation network to model interaction heterogeneity, and recursively performs embedding propagation between neighboring nodes over the user-item interaction graph. Experiments on real-world recommendation datasets show that our GNMR consistently outperforms state-of-the-art methods. The source code is available at https://github.com/akaxlh/GNMR.

Xiyue Zhang, Chao Huang, Yong Xu et al.

Accurate forecasting of citywide traffic flow has been playing critical role in a variety of spatial-temporal mining applications, such as intelligent traffic control and public risk assessment. While previous work has made significant efforts to learn traffic temporal dynamics and spatial dependencies, two key limitations exist in current models. First, only the neighboring spatial correlations among adjacent regions are considered in most existing methods, and the global inter-region dependency is ignored. Additionally, these methods fail to encode the complex traffic transition regularities exhibited with time-dependent and multi-resolution in nature. To tackle these challenges, we develop a new traffic prediction framework-Spatial-Temporal Graph Diffusion Network (ST-GDN). In particular, ST-GDN is a hierarchically structured graph neural architecture which learns not only the local region-wise geographical dependencies, but also the spatial semantics from a global perspective. Furthermore, a multi-scale attention network is developed to empower ST-GDN with the capability of capturing multi-level temporal dynamics. Experiments on several real-life traffic datasets demonstrate that ST-GDN outperforms different types of state-of-the-art baselines. Source codes of implementations are available at https://github.com/jill001/ST-GDN.

Lianghao Xia, Chao Huang, Yong Xu et al.

Capturing users' precise preferences is of great importance in various recommender systems (eg., e-commerce platforms), which is the basis of how to present personalized interesting product lists to individual users. In spite of significant progress has been made to consider relations between users and items, most of the existing recommendation techniques solely focus on singular type of user-item interactions. However, user-item interactive behavior is often exhibited with multi-type (e.g., page view, add-to-favorite and purchase) and inter-dependent in nature. The overlook of multiplex behavior relations can hardly recognize the multi-modal contextual signals across different types of interactions, which limit the feasibility of current recommendation methods. To tackle the above challenge, this work proposes a Memory-Augmented Transformer Networks (MATN), to enable the recommendation with multiplex behavioral relational information, and joint modeling of type-specific behavioral context and type-wise behavior inter-dependencies, in a fully automatic manner. In our MATN framework, we first develop a transformer-based multi-behavior relation encoder, to make the learned interaction representations be reflective of the cross-type behavior relations. Furthermore, a memory attention network is proposed to supercharge MATN capturing the contextual signals of different types of behavior into the category-specific latent embedding space. Finally, a cross-behavior aggregation component is introduced to promote the comprehensive collaboration across type-aware interaction behavior representations, and discriminate their inherent contributions in assisting recommendations. Extensive experiments on two benchmark datasets and a real-world e-commence user behavior data demonstrate significant improvements obtained by MATN over baselines. Codes are available at: https://github.com/akaxlh/MATN.

Lianghao Xia, Chao Huang, Yong Xu et al.

Accurate user and item embedding learning is crucial for modern recommender systems. However, most existing recommendation techniques have thus far focused on modeling users' preferences over singular type of user-item interactions. Many practical recommendation scenarios involve multi-typed user interactive behaviors (e.g., page view, add-to-favorite and purchase), which presents unique challenges that cannot be handled by current recommendation solutions. In particular: i) complex inter-dependencies across different types of user behaviors; ii) the incorporation of knowledge-aware item relations into the multi-behavior recommendation framework; iii) dynamic characteristics of multi-typed user-item interactions. To tackle these challenges, this work proposes a Knowledge-Enhanced Hierarchical Graph Transformer Network (KHGT), to investigate multi-typed interactive patterns between users and items in recommender systems. Specifically, KHGT is built upon a graph-structured neural architecture to i) capture type-specific behavior characteristics; ii) explicitly discriminate which types of user-item interactions are more important in assisting the forecasting task on the target behavior. Additionally, we further integrate the graph attention layer with the temporal encoding strategy, to empower the learned embeddings be reflective of both dedicated multiplex user-item and item-item relations, as well as the underlying interaction dynamics. Extensive experiments conducted on three real-world datasets show that KHGT consistently outperforms many state-of-the-art recommendation methods across various evaluation settings. Our implementation code is available at https://github.com/akaxlh/KHGT.

Chao Huang, Huance Xu, Yong Xu et al.

Social recommendation task aims to predict users' preferences over items with the incorporation of social connections among users, so as to alleviate the sparse issue of collaborative filtering. While many recent efforts show the effectiveness of neural network-based social recommender systems, several important challenges have not been well addressed yet: (i) The majority of models only consider users' social connections, while ignoring the inter-dependent knowledge across items; (ii) Most of existing solutions are designed for singular type of user-item interactions, making them infeasible to capture the interaction heterogeneity; (iii) The dynamic nature of user-item interactions has been less explored in many social-aware recommendation techniques. To tackle the above challenges, this work proposes a Knowledge-aware Coupled Graph Neural Network (KCGN) that jointly injects the inter-dependent knowledge across items and users into the recommendation framework. KCGN enables the high-order user- and item-wise relation encoding by exploiting the mutual information for global graph structure awareness. Additionally, we further augment KCGN with the capability of capturing dynamic multi-typed user-item interactive patterns. Experimental studies on real-world datasets show the effectiveness of our method against many strong baselines in a variety of settings. Source codes are available at: https://github.com/xhcdream/KCGN.

Lianghao Xia, Yong Xu, Chao Huang et al.

Modern recommender systems often embed users and items into low-dimensional latent representations, based on their observed interactions. In practical recommendation scenarios, users often exhibit various intents which drive them to interact with items with multiple behavior types (e.g., click, tag-as-favorite, purchase). However, the diversity of user behaviors is ignored in most of the existing approaches, which makes them difficult to capture heterogeneous relational structures across different types of interactive behaviors. Exploring multi-typed behavior patterns is of great importance to recommendation systems, yet is very challenging because of two aspects: i) The complex dependencies across different types of user-item interactions; ii) Diversity of such multi-behavior patterns may vary by users due to their personalized preference. To tackle the above challenges, we propose a Multi-Behavior recommendation framework with Graph Meta Network to incorporate the multi-behavior pattern modeling into a meta-learning paradigm. Our developed MB-GMN empowers the user-item interaction learning with the capability of uncovering type-dependent behavior representations, which automatically distills the behavior heterogeneity and interaction diversity for recommendations. Extensive experiments on three real-world datasets show the effectiveness of MB-GMN by significantly boosting the recommendation performance as compared to various state-of-the-art baselines. The source code is available athttps://github.com/akaxlh/MB-GMN.

Xiaoling Long, Chao Huang, Yong Xu et al.

In recent years, researchers attempt to utilize online social information to alleviate data sparsity for collaborative filtering, based on the rationale that social networks offers the insights to understand the behavioral patterns. However, due to the overlook of inter-dependent knowledge across items (e.g., categories of products), existing social recommender systems are insufficient to distill the heterogeneous collaborative signals from both user and item sides. In this work, we propose a Self-Supervised Metagraph Infor-max Network (SMIN) which investigates the potential of jointly incorporating social- and knowledge-aware relational structures into the user preference representation for recommendation. To model relation heterogeneity, we design a metapath-guided heterogeneous graph neural network to aggregate feature embeddings from different types of meta-relations across users and items, em-powering SMIN to maintain dedicated representations for multi-faceted user- and item-wise dependencies. Additionally, to inject high-order collaborative signals, we generalize the mutual information learning paradigm under the self-supervised graph-based collaborative filtering. This endows the expressive modeling of user-item interactive patterns, by exploring global-level collaborative relations and underlying isomorphic transformation property of graph topology. Experimental results on several real-world datasets demonstrate the effectiveness of our SMIN model over various state-of-the-art recommendation methods. We release our source code at https://github.com/SocialRecsys/SMIN.

Bo Liu, Chaowei Tan, Jiazhou Wang et al.

In this paper, we present Fedlearn-Algo, an open-source privacy preserving machine learning platform. We use this platform to demonstrate our research and development results on privacy preserving machine learning algorithms. As the first batch of novel FL algorithm examples, we release vertical federated kernel binary classification model and vertical federated random forest model. They have been tested to be more efficient than existing vertical federated learning models in our practice. Besides the novel FL algorithm examples, we also release a machine communication module. The uniform data transfer interface supports transferring widely used data formats between machines. We will maintain this platform by adding more functional modules and algorithm examples. The code is available at https://github.com/fedlearnAI/fedlearn-algo.

Guanghan Ning, Guang Chen, Chaowei Tan et al.

It is challenging to train a robust object detector under the supervised learning setting when the annotated data are scarce. Thus, previous approaches tackling this problem are in two categories: semi-supervised learning models that interpolate labeled data from unlabeled data, and self-supervised learning approaches that exploit signals within unlabeled data via pretext tasks. To seamlessly integrate and enhance existing supervised object detection methods, in this work, we focus on addressing the data scarcity problem from a fundamental viewpoint without changing the supervised learning paradigm. We propose a new offline data augmentation method for object detection, which semantically interpolates the training data with novel views. Specifically, our new system generates controllable views of training images based on differentiable neural rendering, together with corresponding bounding box annotations which involve no human intervention. Firstly, we extract and project pixel-aligned image features into point clouds while estimating depth maps. We then re-project them with a target camera pose and render a novel-view 2d image. Objects in the form of keypoints are marked in point clouds to recover annotations in new views. Our new method is fully compatible with online data augmentation methods, such as affine transform, image mixup, etc. Extensive experiments show that our method, as a cost-free tool to enrich images and labels, can significantly boost the performance of object detection systems with scarce training data. Code is available at \url{https://github.com/Guanghan/DANR}.

Xinyao Wang, Liefeng Bo, Li Fuxin

Heatmap regression with a deep network has become one of the mainstream approaches to localize facial landmarks. However, the loss function for heatmap regression is rarely studied. In this paper, we analyze the ideal loss function properties for heatmap regression in face alignment problems. Then we propose a novel loss function, named Adaptive Wing loss, that is able to adapt its shape to different types of ground truth heatmap pixels. This adaptability penalizes loss more on foreground pixels while less on background pixels. To address the imbalance between foreground and background pixels, we also propose Weighted Loss Map, which assigns high weights on foreground and difficult background pixels to help training process focus more on pixels that are crucial to landmark localization. To further improve face alignment accuracy, we introduce boundary prediction and CoordConv with boundary coordinates. Extensive experiments on different benchmarks, including COFW, 300W and WFLW, show our approach outperforms the state-of-the-art by a significant margin on various evaluation metrics. Besides, the Adaptive Wing loss also helps other heatmap regression tasks. Code will be made publicly available at https://github.com/protossw512/AdaptiveWingLoss.

Kai Xu, Longyin Wen, Guorong Li et al.

In this paper, we present a unified, end-to-end trainable spatiotemporal CNN model for VOS, which consists of two branches, i.e., the temporal coherence branch and the spatial segmentation branch. Specifically, the temporal coherence branch pretrained in an adversarial fashion from unlabeled video data, is designed to capture the dynamic appearance and motion cues of video sequences to guide object segmentation. The spatial segmentation branch focuses on segmenting objects accurately based on the learned appearance and motion cues. To obtain accurate segmentation results, we design a coarse-to-fine process to sequentially apply a designed attention module on multi-scale feature maps, and concatenate them to produce the final prediction. In this way, the spatial segmentation branch is enforced to gradually concentrate on object regions. These two branches are jointly fine-tuned on video segmentation sequences in an end-to-end manner. Several experiments are carried out on three challenging datasets (i.e., DAVIS-2016, DAVIS-2017 and Youtube-Object) to show that our method achieves favorable performance against the state-of-the-arts. Code is available at https://github.com/longyin880815/STCNN.

Rui Zhu, Shifeng Zhang, Xiaobo Wang et al.

Current state-of-the-art object objectors are fine-tuned from the off-the-shelf networks pretrained on large-scale classification dataset ImageNet, which incurs some additional problems: 1) The classification and detection have different degrees of sensitivity to translation, resulting in the learning objective bias; 2) The architecture is limited by the classification network, leading to the inconvenience of modification. To cope with these problems, training detectors from scratch is a feasible solution. However, the detectors trained from scratch generally perform worse than the pretrained ones, even suffer from the convergence issue in training. In this paper, we explore to train object detectors from scratch robustly. By analysing the previous work on optimization landscape, we find that one of the overlooked points in current trained-from-scratch detector is the BatchNorm. Resorting to the stable and predictable gradient brought by BatchNorm, detectors can be trained from scratch stably while keeping the favourable performance independent to the network architecture. Taking this advantage, we are able to explore various types of networks for object detection, without suffering from the poor convergence. By extensive experiments and analyses on downsampling factor, we propose the Root-ResNet backbone network, which makes full use of the information from original images. Our ScratchDet achieves the state-of-the-art accuracy on PASCAL VOC 2007, 2012 and MS COCO among all the train-from-scratch detectors and even performs better than several one-stage pretrained methods. Codes will be made publicly available at https://github.com/KimSoybean/ScratchDet.

Jiaxing Zhao, Xihan Wei, Liefeng Bo

In this work, we present the first application of Reinforcement Learning with Verifiable Reward (RLVR) to an Omni-multimodal large language model in the context of emotion recognition, a task where both visual and audio modalities play crucial roles. We leverage RLVR to optimize the Omni model, significantly enhancing its performance in three key aspects: reasoning capability, emotion recognition accuracy, and generalization ability. The introduction of RLVR not only improves the model's overall performance on in-distribution data but also demonstrates superior robustness when evaluated on out-of-distribution datasets. More importantly, the improved reasoning capability enables clear analysis of the contributions of different modalities, particularly visual and audio information, in the emotion recognition process. This provides valuable insights into the optimization of multimodal large language models.

Qi Zuo, Xiaodong Gu, Lingteng Qiu et al.

Generating multi-view images based on text or single-image prompts is a critical capability for the creation of 3D content. Two fundamental questions on this topic are what data we use for training and how to ensure multi-view consistency. This paper introduces a novel framework that makes fundamental contributions to both questions. Unlike leveraging images from 2D diffusion models for training, we propose a dense consistent multi-view generation model that is fine-tuned from off-the-shelf video generative models. Images from video generative models are more suitable for multi-view generation because the underlying network architecture that generates them employs a temporal module to enforce frame consistency. Moreover, the video data sets used to train these models are abundant and diverse, leading to a reduced train-finetuning domain gap. To enhance multi-view consistency, we introduce a 3D-Aware Denoising Sampling, which first employs a feed-forward reconstruction module to get an explicit global 3D model, and then adopts a sampling strategy that effectively involves images rendered from the global 3D model into the denoising sampling loop to improve the multi-view consistency of the final images. As a by-product, this module also provides a fast way to create 3D assets represented by 3D Gaussians within a few seconds. Our approach can generate 24 dense views and converges much faster in training than state-of-the-art approaches (4 GPU hours versus many thousand GPU hours) with comparable visual quality and consistency. By further fine-tuning, our approach outperforms existing state-of-the-art methods in both quantitative metrics and visual effects. Our project page is aigc3d.github.io/VideoMV.

Xusen Sun, Longhao Zhang, Hao Zhu et al.

Audio-driven talking head generation has drawn much attention in recent years, and many efforts have been made in lip-sync, expressive facial expressions, natural head pose generation, and high video quality. However, no model has yet led or tied on all these metrics due to the one-to-many mapping between audio and motion. In this paper, we propose VividTalk, a two-stage generic framework that supports generating high-visual quality talking head videos with all the above properties. Specifically, in the first stage, we map the audio to mesh by learning two motions, including non-rigid expression motion and rigid head motion. For expression motion, both blendshape and vertex are adopted as the intermediate representation to maximize the representation ability of the model. For natural head motion, a novel learnable head pose codebook with a two-phase training mechanism is proposed. In the second stage, we proposed a dual branch motion-vae and a generator to transform the meshes into dense motion and synthesize high-quality video frame-by-frame. Extensive experiments show that the proposed VividTalk can generate high-visual quality talking head videos with lip-sync and realistic enhanced by a large margin, and outperforms previous state-of-the-art works in objective and subjective comparisons.

Li Hu, Guangyuan Wang, Zhen Shen et al.

Recent character image animation methods based on diffusion models, such as Animate Anyone, have made significant progress in generating consistent and generalizable character animations. However, these approaches fail to produce reasonable associations between characters and their environments. To address this limitation, we introduce Animate Anyone 2, aiming to animate characters with environment affordance. Beyond extracting motion signals from source video, we additionally capture environmental representations as conditional inputs. The environment is formulated as the region with the exclusion of characters and our model generates characters to populate these regions while maintaining coherence with the environmental context. We propose a shape-agnostic mask strategy that more effectively characterizes the relationship between character and environment. Furthermore, to enhance the fidelity of object interactions, we leverage an object guider to extract features of interacting objects and employ spatial blending for feature injection. We also introduce a pose modulation strategy that enables the model to handle more diverse motion patterns. Experimental results demonstrate the superior performance of the proposed method.

Linrui Tian, Siqi Hu, Qi Wang et al.

In this paper, we propose a novel audio-driven talking head method capable of simultaneously generating highly expressive facial expressions and hand gestures. Unlike existing methods that focus on generating full-body or half-body poses, we investigate the challenges of co-speech gesture generation and identify the weak correspondence between audio features and full-body gestures as a key limitation. To address this, we redefine the task as a two-stage process. In the first stage, we generate hand poses directly from audio input, leveraging the strong correlation between audio signals and hand movements. In the second stage, we employ a diffusion model to synthesize video frames, incorporating the hand poses generated in the first stage to produce realistic facial expressions and body movements. Our experimental results demonstrate that the proposed method outperforms state-of-the-art approaches, such as CyberHost and Vlogger, in terms of both visual quality and synchronization accuracy. This work provides a new perspective on audio-driven gesture generation and a robust framework for creating expressive and natural talking head animations.

Lingteng Qiu, Xiaodong Gu, Peihao Li et al.

Animatable 3D human reconstruction from a single image is a challenging problem due to the ambiguity in decoupling geometry, appearance, and deformation. Recent advances in 3D human reconstruction mainly focus on static human modeling, and the reliance of using synthetic 3D scans for training limits their generalization ability. Conversely, optimization-based video methods achieve higher fidelity but demand controlled capture conditions and computationally intensive refinement processes. Motivated by the emergence of large reconstruction models for efficient static reconstruction, we propose LHM (Large Animatable Human Reconstruction Model) to infer high-fidelity avatars represented as 3D Gaussian splatting in a feed-forward pass. Our model leverages a multimodal transformer architecture to effectively encode the human body positional features and image features with attention mechanism, enabling detailed preservation of clothing geometry and texture. To further boost the face identity preservation and fine detail recovery, we propose a head feature pyramid encoding scheme to aggregate multi-scale features of the head regions. Extensive experiments demonstrate that our LHM generates plausible animatable human in seconds without post-processing for face and hands, outperforming existing methods in both reconstruction accuracy and generalization ability.