cs.CV（2025-11-10）

📊 共 38 篇论文 | 🔗 7 篇有代码

🎯 兴趣领域导航

支柱三：空间感知 (Perception & SLAM) (22 🔗4) 支柱二：RL算法与架构 (RL & Architecture) (6 🔗1) 支柱一：机器人控制 (Robot Control) (4 🔗1) 支柱七：动作重定向 (Motion Retargeting) (3) 支柱四：生成式动作 (Generative Motion) (2 🔗1) 支柱八：物理动画 (Physics-based Animation) (1)

🔬 支柱三：空间感知 (Perception & SLAM) (22 篇)

#题目一句话要点标签🔗⭐
1 Robust and High-Fidelity 3D Gaussian Splatting: Fusing Pose Priors and Geometry Constraints for Texture-Deficient Outdoor Scenes 针对纹理缺失的室外场景，提出融合位姿先验和几何约束的鲁棒高保真3D高斯溅射方法 3D gaussian splatting 3DGS gaussian splatting ✅
2 YoNoSplat: You Only Need One Model for Feedforward 3D Gaussian Splatting YoNoSplat：仅需单模型的前馈3D高斯溅射重建，适用于各种相机内外参场景 3D gaussian splatting gaussian splatting scene reconstruction ✅
3 Sparse4DGS: 4D Gaussian Splatting for Sparse-Frame Dynamic Scene Reconstruction Sparse4DGS：提出纹理感知正则化与优化，解决稀疏帧动态场景的4D高斯重建问题。 gaussian splatting NeRF scene reconstruction
4 GFix: Perceptually Enhanced Gaussian Splatting Video Compression GFix：提出感知增强的高斯溅射视频压缩方法，提升视觉质量和压缩率。 3D gaussian splatting 3DGS gaussian splatting
5 Rethinking Rainy 3D Scene Reconstruction via Perspective Transforming and Brightness Tuning 提出REVR-GSNet以解决雨天3D场景重建问题 3D gaussian splatting gaussian splatting scene reconstruction
6 MUGSQA: Novel Multi-Uncertainty-Based Gaussian Splatting Quality Assessment Method, Dataset, and Benchmarks 提出MUGSQA数据集与评测方法，用于评估高斯溅射重建三维物体的感知质量。 gaussian splatting point cloud
7 ConeGS: Error-Guided Densification Using Pixel Cones for Improved Reconstruction with Fewer Primitives ConeGS：利用像素锥误差引导稠密化，以更少图元实现更优重建 3D gaussian splatting 3DGS gaussian splatting
8 DIAL-GS: Dynamic Instance Aware Reconstruction for Label-free Street Scenes with 4D Gaussian Splatting DIAL-GS：用于无标签街景的动态实例感知4D高斯溅射重建 gaussian splatting scene reconstruction
9 FlowFeat: Pixel-Dense Embedding of Motion Profiles 提出FlowFeat，通过运动轮廓嵌入实现像素级密集图像表征，提升多种视觉任务性能。 depth estimation monocular depth optical flow
10 LiveNeRF: Efficient Face Replacement Through Neural Radiance Fields Integration LiveNeRF：通过神经辐射场集成实现高效人脸替换 neural radiance
11 RaLD: Generating High-Resolution 3D Radar Point Clouds with Latent Diffusion 提出RaLD，利用潜在扩散模型从雷达频谱生成高分辨率3D点云。 point cloud
12 3D-ANC: Adaptive Neural Collapse for Robust 3D Point Cloud Recognition 提出3D-ANC，利用神经崩溃机制提升3D点云识别的鲁棒性，对抗恶意攻击。 point cloud
13 Certified L2-Norm Robustness of 3D Point Cloud Recognition in the Frequency Domain FreqCert：提出频域认证框架，提升3D点云识别对L2范数扰动的鲁棒性 point cloud
14 PanoNav: Mapless Zero-Shot Object Navigation with Panoramic Scene Parsing and Dynamic Memory PanoNav：基于全景场景解析与动态记忆的无地图零样本物体导航 navigation
15 PointCubeNet: 3D Part-level Reasoning with 3x3x3 Point Cloud Blocks PointCubeNet：提出一种基于3x3x3点云块的无监督3D部件级推理框架 point cloud
16 Omni-View: Unlocking How Generation Facilitates Understanding in Unified 3D Model based on Multiview images Omni-View：提出基于多视角图像的统一3D模型，探索生成促进理解的原理。 novel view synthesis scene understanding
17 LeCoT: revisiting network architecture for two-view correspondence pruning LeCoT：通过空间-通道融合Transformer改进双视图对应关系剪枝 pose estimation localization ✅
18 4DSTR: Advancing Generative 4D Gaussians with Spatial-Temporal Rectification for High-Quality and Consistent 4D Generation 提出4DSTR网络，通过时空校正生成高质量、时序一致的4D高斯模型。 gaussian splatting
19 Geometric implicit neural representations for signed distance functions 提出几何隐式神经表示，用于有向距离函数的表面重建 point cloud
20 Mono3DVG-EnSD: Enhanced Spatial-aware and Dimension-decoupled Text Encoding for Monocular 3D Visual Grounding 提出Mono3DVG-EnSD框架，增强单目3D视觉定位中空间感知和维度解耦的文本编码。 localization
21 Gaussian-Augmented Physics Simulation and System Identification with Complex Colliders 提出AS-DiffMPM，解决复杂碰撞体下基于视频的物理属性辨识难题 novel view synthesis
22 UniADC: A Unified Framework for Anomaly Detection and Classification 提出UniADC，统一异常检测与分类框架，解决信息孤岛问题。 localization ✅

🔬 支柱二：RL算法与架构 (RL & Architecture) (6 篇)

#题目一句话要点标签🔗⭐
23 TiS-TSL: Image-Label Supervised Surgical Video Stereo Matching via Time-Switchable Teacher-Student Learning 提出TiS-TSL，通过时序可切换的师生学习解决手术视频立体匹配中的时序一致性问题 teacher-student stereo matching navigation
24 Learning from the Right Patches: A Two-Stage Wavelet-Driven Masked Autoencoder for Histopathology Representation Learning WISE-MAE：一种基于小波变换的双阶段掩码自编码器，用于病理图像表征学习 representation learning masked autoencoder MAE
25 MirrorMamba: Towards Scalable and Robust Mirror Detection in Videos MirrorMamba：提出一种可扩展且鲁棒的视频镜像检测方法 Mamba
26 Spatial-Frequency Enhanced Mamba for Multi-Modal Image Fusion 提出空间-频率增强Mamba融合网络，提升多模态图像融合性能 Mamba ✅
27 ConsistTalk: Intensity Controllable Temporally Consistent Talking Head Generation with Diffusion Noise Search ConsistTalk：提出基于扩散噪声搜索的、强度可控且时序一致的说话人头部生成框架 teacher-student optical flow
28 MRT: Learning Compact Representations with Mixed RWKV-Transformer for Extreme Image Compression 提出混合RWKV-Transformer的MRT模型，用于极低码率图像压缩，显著提升压缩性能。 linear attention representation learning

🔬 支柱一：机器人控制 (Robot Control) (4 篇)

#题目一句话要点标签🔗⭐
29 TrueCity: Real and Simulated Urban Data for Cross-Domain 3D Scene Understanding TrueCity：提出城市三维场景理解的真实与模拟跨域数据集 sim-to-real scene understanding point cloud ✅
30 Improving Deepfake Detection with Reinforcement Learning-Based Adaptive Data Augmentation 提出基于强化学习的自适应数据增强方法CRDA，提升Deepfake检测器的泛化能力。 manipulation reinforcement learning
31 FoCLIP: A Feature-Space Misalignment Framework for CLIP-Based Image Manipulation and Detection 提出FoCLIP框架，通过特征空间错位攻击和防御CLIP模型，提升图像篡改检测能力。 manipulation
32 Breaking the Stealth-Potency Trade-off in Clean-Image Backdoors with Generative Trigger Optimization 提出GCB框架，通过生成式触发器优化解决clean-image后门攻击的隐蔽性与效力权衡问题 manipulation

🔬 支柱七：动作重定向 (Motion Retargeting) (3 篇)

#题目一句话要点标签🔗⭐
33 AvatarTex: High-Fidelity Facial Texture Reconstruction from Single-Image Stylized Avatars AvatarTex：单图像生成高保真风格化头像纹理，解决几何一致性难题 latent optimization geometric consistency
34 SPAN: Spatial-Projection Alignment for Monocular 3D Object Detection 提出SPAN，通过空间投影对齐解决单目3D目标检测中的几何不一致问题 geometric consistency
35 On Accurate and Robust Estimation of 3D and 2D Circular Center: Method and Application to Camera-Lidar Calibration 提出基于共形几何代数的鲁棒圆形标靶中心估计方法，用于相机-激光雷达标定 geometric consistency

🔬 支柱四：生成式动作 (Generative Motion) (2 篇)

#题目一句话要点标签🔗⭐
36 DIMO: Diverse 3D Motion Generation for Arbitrary Objects 提出DIMO以生成任意物体的多样化3D运动 motion generation ✅
37 Slow - Motion Video Synthesis for Basketball Using Frame Interpolation 通过微调RIFE网络，实现高质量篮球赛事慢动作视频合成 motion synthesis motion generation

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

#题目一句话要点标签🔗⭐
38 Integrating Reweighted Least Squares with Plug-and-Play Diffusion Priors for Noisy Image Restoration 提出基于重加权最小二乘与即插即用扩散先验的图像恢复框架，用于去除噪声。 PULSE

⬅️ 返回 cs.CV 首页 · 🏠 返回主页