cs.CV（2026-02-10）

📊 共 32 篇论文 | 🔗 6 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (12 🔗4) 支柱三：空间感知与语义 (Perception & Semantics) (9) 支柱二：RL算法与架构 (RL & Architecture) (7 🔗2) 支柱一：机器人控制 (Robot Control) (3) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (12 篇)

#题目一句话要点标签🔗⭐
1 Towards Training-free Multimodal Hate Localisation with Large Language Models 提出LELA：一种基于大语言模型的无训练多模态仇恨内容定位框架 large language model multimodal
2 SchröMind: Mitigating Hallucinations in Multimodal Large Language Models via Solving the Schrödinger Bridge Problem SchröMind：通过求解薛定谔桥问题缓解多模态大语言模型中的幻觉 large language model multimodal
3 Tele-Omni: a Unified Multimodal Framework for Video Generation and Editing Tele-Omni：统一多模态视频生成与编辑框架，支持文本、图像和视频指令。 large language model multimodal
4 Scalpel: Fine-Grained Alignment of Attention Activation Manifolds via Mixture Gaussian Bridges to Mitigate Multimodal Hallucination Scalpel：通过混合高斯桥精细对齐注意力激活流形，缓解多模态幻觉问题 large language model multimodal
5 SAGE: Scalable Agentic 3D Scene Generation for Embodied AI SAGE：用于具身AI的可扩展Agentic 3D场景生成 embodied AI ✅
6 ARK: A Dual-Axis Multimodal Retrieval Benchmark along Reasoning and Knowledge 提出ARK基准，用于评估多模态检索在推理和知识方面的能力。 multimodal
7 Reason-IAD: Knowledge-Guided Dynamic Latent Reasoning for Explainable Industrial Anomaly Detection 提出Reason-IAD，利用知识引导的动态隐空间推理进行可解释的工业异常检测 large language model multimodal ✅
8 Kelix Technique Report Kelix：一种全离散自回归统一模型，缩小了离散和连续视觉表示之间的理解差距。 large language model multimodal
9 Understanding and Enhancing Encoder-based Adversarial Transferability against Large Vision-Language Models 提出语义引导的多模态攻击SGMA，提升针对大型视觉语言模型的编码器对抗迁移性 multimodal visual grounding
10 AdaTSQ: Pushing the Pareto Frontier of Diffusion Transformers via Temporal-Sensitivity Quantization AdaTSQ：通过时序敏感量化提升扩散Transformer的Pareto前沿 large language model ✅
11 AGMark: Attention-Guided Dynamic Watermarking for Large Vision-Language Models AGMark：注意力引导的动态水印技术，用于保护大型视觉-语言模型的内容版权。 visual grounding
12 AUHead: Realistic Emotional Talking Head Generation via Action Units Control AUHead：通过动作单元控制实现逼真情感的说话头生成 chain-of-thought ✅

🔬 支柱三：空间感知与语义 (Perception & Semantics) (9 篇)

#题目一句话要点标签🔗⭐
13 CompSplat: Compression-aware 3D Gaussian Splatting for Real-world Video 提出CompSplat，解决真实世界视频压缩感知的3D高斯溅射问题 3D gaussian splatting gaussian splatting splatting
14 Faster-GS: Analyzing and Improving Gaussian Splatting Optimization Faster-GS：通过分析与改进高斯溅射优化，实现更快的训练速度和高质量重建 3D gaussian splatting 3DGS gaussian splatting
15 Single-Slice-to-3D Reconstruction in Medical Imaging and Natural Objects: A Comparative Benchmark with SAM 3D 对比评估单切片医学影像三维重建模型，揭示几何先验迁移的局限性 sam 3D SAM 3D foundation model
16 RAD: Retrieval-Augmented Monocular Metric Depth Estimation for Underrepresented Classes RAD：检索增强的单目深度估计，提升欠表示类别的深度预测精度 depth estimation metric depth
17 Stability and Concentration in Nonlinear Inverse Problems with Block-Structured Parameters: Lipschitz Geometry, Identifiability, and an Application to Gaussian Splatting 针对块结构参数非线性反问题，提出稳定性与集中度分析框架，并应用于高斯溅射。 gaussian splatting splatting
18 VersaViT: Enhancing MLLM Vision Backbones via Task-Guided Optimization VersaViT：通过任务引导优化增强MLLM视觉骨干网络 depth estimation large language model multimodal
19 Toward Fine-Grained Facial Control in 3D Talking Head Generation 提出FG-3DGS框架，解决3D说话头生成中精细面部控制难题 3D gaussian splatting 3DGS gaussian splatting
20 Learning to Detect Baked Goods with Limited Supervision 提出基于有限监督的目标检测方法，用于识别烘焙产品以优化生产。 open-vocabulary open vocabulary
21 4RC: 4D Reconstruction via Conditional Querying Anytime and Anywhere 提出4RC，通过条件查询实现任意时间、任意位置的单目视频4D重建 scene flow

🔬 支柱二：RL算法与架构 (RL & Architecture) (7 篇)

#题目一句话要点标签🔗⭐
22 Beyond Next-Token Alignment: Distilling Multimodal Large Language Models via Token Interactions 提出Align-TI，通过token交互蒸馏多模态大语言模型，显著提升模型性能。 distillation large language model multimodal ✅
23 Hand2World: Autoregressive Egocentric Interaction Generation via Free-Space Hand Gestures 提出Hand2World以解决自由空间手势交互生成问题 world model egocentric geometric consistency
24 BabyMamba-HAR: Lightweight Selective State Space Models for Efficient Human Activity Recognition on Resource Constrained Devices 提出BabyMamba-HAR以解决资源受限设备上的人类活动识别问题 Mamba SSM state space model
25 Code2World: A GUI World Model via Renderable Code Generation 提出Code2World，通过可渲染代码生成实现GUI世界模型，提升GUI Agent的交互能力。 reinforcement learning world model ✅
26 Weakly Supervised Contrastive Learning for Histopathology Patch Embeddings 提出弱监督对比学习WeakSupCon，提升病理切片嵌入表示质量，改善MIL性能 representation learning contrastive learning
27 Olaf-World: Orienting Latent Actions for Video World Modeling Olaf-World：通过对齐潜在动作来构建可控的视频世界模型 world model
28 Fake-HR1: Rethinking reasoning of vision language model for synthetic image detection 提出Fake-HR1，自适应推理检测合成图像，提升效率与性能。 reinforcement learning chain-of-thought

🔬 支柱一：机器人控制 (Robot Control) (3 篇)

#题目一句话要点标签🔗⭐
29 VideoAfford: Grounding 3D Affordance from Human-Object-Interaction Videos via Multimodal Large Language Model VideoAfford：利用多模态大语言模型从人-物交互视频中学习3D可供性 manipulation affordance human-object interaction
30 MVISTA-4D: View-Consistent 4D World Model with Test-Time Action Inference for Robotic Manipulation MVISTA-4D：用于机器人操作的视角一致性4D世界模型与测试时动作推断 manipulation world model
31 VideoWorld 2: Learning Transferable Knowledge from Real-world Videos VideoWorld 2：提出动态增强潜在动力学模型，从真实视频中学习可迁移知识 manipulation latent dynamics

🔬 支柱八：物理动画 (Physics-based Animation) (1 篇)

#题目一句话要点标签🔗⭐
32 Time2General: Learning Spatiotemporal Invariant Representations for Domain-Generalization Video Semantic Segmentation 提出Time2General框架，学习时空不变表征，解决域泛化视频语义分割问题。 spatiotemporal

⬅️ 返回 cs.CV 首页 · 🏠 返回主页