F2G-Pose: real-time category-level pose estimation from a single RGB-D image
sii-research · hf · 2026-10-23
F2G-Pose is a real-time feed-forward framework predicting category-level object pose, metric size, and completed shape from a single RGB-D image. It fuses dense RADIO v2.5 visual features with the point cloud, tokenizes via DGCNN, and runs an 8-layer MoE Transformer with multi-task heads for rotation, translation, and size, iteratively upsampling to a completed 2048×3 point cloud.
More from Research
- 10 of 12 Beta Robots Worked Without Fine-Tuning — Do We Still Need It? — micoolcho · 2026-10-23
- EmbodiedEvo lets robot agents self-train, banking reusable Skills and Lessons across tasks — jiqizhixin · 2026-10-23
- LAW proposes dropping diffusion entirely: language models built on pure self-conditioning — ArnaudDoucet1 · 2026-10-23
- Google paper: AI seller agents inflate tool listings, collapsing buyer agent success from 89% to 30% — rohanpaul_ai · 2026-10-23
- Study: 8 of 13 AI agents charge richer users more, up to $198 extra on flights — 2C_ornot2C · 2026-10-23
- AnyGameRenderer turns AI agent-built graybox game footage into high-fidelity visuals — AlayaLab · 2026-10-23