cs.CV（2025-12-31）

📊 共 19 篇论文 | 🔗 4 篇有代码

🎯 兴趣领域导航

支柱九：具身大模型 (Embodied Foundation Models) (7) 支柱三：空间感知与语义 (Perception & Semantics) (5 🔗1) 支柱二：RL算法与架构 (RL & Architecture) (4 🔗1) 支柱一：机器人控制 (Robot Control) (1 🔗1) 支柱四：生成式动作 (Generative Motion) (1) 支柱七：动作重定向 (Motion Retargeting) (1 🔗1)

🔬 支柱九：具身大模型 (Embodied Foundation Models) (7 篇)

#题目一句话要点标签🔗⭐
1 VLN-MME: Diagnosing MLLMs as Language-guided Visual Navigation agents VLN-MME：诊断多模态大语言模型在语言引导视觉导航任务中的能力 VLN large language model multimodal
2 FinMMDocR: Benchmarking Financial Multimodal Reasoning with Scenario Awareness, Document Understanding, and Multi-Step Computation FinMMDocR：提出金融多模态推理基准，关注场景感知、文档理解和多步计算。 large language model multimodal
3 A Spatially Masked Adaptive Gated Network for multimodal post-flood water extent mapping using SAR and incomplete multispectral data 提出SMAGNet，利用SAR和不完整MSI数据进行洪水范围精确高效的多模态映射。 multimodal
4 MoniRefer: A Real-world Large-scale Multi-modal Dataset based on Roadside Infrastructure for 3D Visual Grounding 提出MoniRefer数据集和Moni3DVG方法，用于路侧基础设施的3D视觉定位。 visual grounding
5 RGBT-Ground Benchmark: Visual Grounding Beyond RGB in Complex Real-World Scenarios 提出RGBT-Ground基准，用于评估复杂场景下RGB-T图像的视觉定位能力 visual grounding
6 UR-Bench: A Benchmark for Multi-Hop Reasoning over Ultra-High-Resolution Images 提出UR-Bench，用于评估多模态大模型在超高分辨率图像上的多跳推理能力。 large language model multimodal
7 EchoFoley: Event-Centric Hierarchical Control for Video Grounded Creative Sound Generation 提出EchoFoley任务与EchoVidia框架，用于视频事件驱动的精细化创意声音生成。 multimodal

🔬 支柱三：空间感知与语义 (Perception & Semantics) (5 篇)

#题目一句话要点标签🔗⭐
8 Splatwizard: A Benchmark Toolkit for 3D Gaussian Splatting Compression Splatwizard：用于3D高斯溅射压缩的综合基准测试工具包 3D gaussian splatting 3DGS gaussian splatting ✅
9 FoundationSLAM: Unleashing the Power of Depth Foundation Models for End-to-End Dense Visual SLAM FoundationSLAM：利用深度基础模型实现端到端稠密视觉SLAM visual SLAM geometric consistency foundation model
10 Spatial4D-Bench: A Versatile 4D Spatial Intelligence Benchmark 提出Spatial4D-Bench，用于全面评估多模态大语言模型在4D空间智能方面的能力。 scene understanding spatial relationship spatiotemporal
11 Projection-based Adversarial Attack using Physics-in-the-Loop Optimization for Monocular Depth Estimation 提出基于物理环路优化的投影对抗攻击，用于单目深度估计 depth estimation monocular depth
12 HaineiFRDM: Explore Diffusion to Restore Defects in Fast-Movement Films 提出HaineiFRDM，利用扩散模型修复快速移动影片中的缺陷。 optical flow

🔬 支柱二：RL算法与架构 (RL & Architecture) (4 篇)

#题目一句话要点标签🔗⭐
13 UniC-Lift: Unified 3D Instance Segmentation via Contrastive Learning UniC-Lift：通过对比学习实现统一的3D实例分割 contrastive learning 3D gaussian splatting 3DGS
14 TeleWorld: Towards Dynamic Multimodal Synthesis with a 4D World Model TeleWorld：基于4D世界模型的动态多模态实时合成框架 world model distillation scene reconstruction
15 VideoCuRL: Video Curriculum Reinforcement Learning with Orthogonal Difficulty Decomposition VideoCuRL：提出正交难度分解的视频课程强化学习，提升视频理解能力。 reinforcement learning optical flow spatiotemporal
16 PhyGDPO: Physics-Aware Groupwise Direct Preference Optimization for Physically Consistent Text-to-Video Generation 提出PhyGDPO框架，通过物理感知的群体偏好优化实现物理一致的文本生成视频。 direct preference optimization chain-of-thought ✅

🔬 支柱一：机器人控制 (Robot Control) (1 篇)

#题目一句话要点标签🔗⭐
17 ShowUI-$π$: Flow-based Generative Models as GUI Dexterous Hands ShowUI-$π$：提出基于Flow的生成模型，实现GUI界面的灵巧操作。 manipulation dexterous hand dexterous manipulation ✅

🔬 支柱四：生成式动作 (Generative Motion) (1 篇)

#题目一句话要点标签🔗⭐
18 Hierarchical Vector-Quantized Latents for Perceptual Low-Resolution Video Compression 提出一种分层矢量量化隐变量的感知低分辨率视频压缩方法，适用于带宽受限场景。 VQ-VAE spatiotemporal

🔬 支柱七：动作重定向 (Motion Retargeting) (1 篇)

#题目一句话要点标签🔗⭐
19 GaMO: Geometry-aware Multi-view Diffusion Outpainting for Sparse-View 3D Reconstruction GaMO：基于几何感知的多视角扩散外绘用于稀疏视角3D重建 geometric consistency ✅

⬅️ 返回 cs.CV 首页 · 🏠 返回主页