arXiv 2610.31623 · cs.CV · 2026-10-22
F2G-Pose: Geometry-Aware Foundation Feature Lifting for Direct RGB-D Category-Level Object Pose Estimation
Jinyu Zhang, Haitao Lin, Jiashu Hou, Xiangyang Xue, Yanwei Fu
Shanghai Innovation Institute
一句话要点
F2G-Pose提出一种实时直接的单次传递框架,通过几何感知特征提升和混合专家Transformer实现从单帧RGB-D数据中准确估计类别级物体位姿与形状。
F2G-Pose achieves real-time, accurate category-level RGB-D pose estimation via geometry-aware feature lifting and a MoE Transformer.
要点
- 类别级RGB-D位姿估计面临内在变异性、对称性和部分观测导致的高模糊性,现有方法在精度与效率间难以平衡。
- F2G-Pose通过将密集视觉基础特征提升至点云并结合DGCNN分块编码和Mixture-of-Experts Transformer建模局部几何结构,实现单次前向传递预测位姿、尺寸及相机帧完整形状。
- 在SOPE和ROPE数据集上达到当前最优精度,仅需合成的Pose-annotated SOPE数据训练,推理速度达28 FPS,且在HouseCat6D和HANDAL真实物体迁移任务中表现稳健。
- 方法依赖外部预训练的RADIO骨干网络,具备良好的泛化能力,实验验证了其在实际操作场景下的部署潜力。
English overview
F2G-PPose introduces a real-time, direct framework for category-level RGB-D pose estimation by lifting visual features to point clouds and using geometry-aware tokenization with a Mixture-of-Experts Transformer. It achieves state-of-the-art accuracy on synthetic benchmarks while maintaining high inference speed and robustness through dense geometric supervision and metric consistency.
只根据 arXiv 官方摘要,由开源模型整理,经数字/专名自动核对(必须在摘要中出现),以论文原文为准。
阅读 arXiv 原文 PDF Hugging Face 讨论 代码仓库 项目页
主题与学科
同主题:视觉理解与 3D
- Spectral-Spatial Interpretation ▲753
- VisionHOPE: Visual Backbones as Self-Modifying Learning Systems ▲296
- EgoXperience: Learning World-Grounded Egocentric Representations from Human Experience ▲189
- Holi-World: Building Holistic World Intelligence from Human-Ego Experience ▲172
- DONIC: Generalist Dexterous Bimanual Tracking Control from Human Demonstrations ▲161
同主题:高效训练与推理
- MJev-Spatial: System One Spatial Intelligence via Native Attention Matching 2026-10-15 · ▲556
- The Teacher Is a Direction, Not a Destination: Extrapolating RL-Induced Representation Residuals in On-Policy Distillation 2026-09-29 · ▲511
- VideoJev: System One Model for Video Understanding and Interaction 2026-10-13 · ▲375
- Jeva: System One Models from First Principles 2026-10-13 · ▲371
- LEV: Thinking Fast and Deep with Looped Decision Models 2026-10-14 · ▲337
数据说明
元数据来自 arXiv API(已核实);点赞来自 Hugging Face 每日论文;引用来自 Semantic Scholar。最近刷新 2026-10-22。
要点由本站根据摘要自写,不转载摘要原文,请阅读原文。