Fudan's F2G-Pose estimates 6D pose, size and shape from one RGB-D image at 28 FPS
joemeno · x · 2026-10-24
Researchers from Fudan University and Shanghai Innovation Institute introduce F2G-Pose, which estimates an object's 6D pose, real-world size, and completed shape from a single RGB-D image.
- Takes a pre-segmented image and predicts pose, metric size, and shape in one shot
- Fuses foundation-model features with 3D geometric information for single-pass inference
- Reports 28 FPS for pose estimation
Related event: Fudan's F2G-Pose Estimates 6D Pose and Shape from a Single RGB-D Image(2 posts)→
More from Embodied
- SenseTime open-sources 27B robot model matching GPT-6-Astra at ~1/50 the cost — rohanpaul_ai · 2026-10-24
- Samsung TVs found running facial recognition code that returns face bounding boxes — basedjensen · 2026-10-24
- Open-source humanoid Asimov 1 tests new grippers on day 363 of development — freelerobot · 2026-10-24
- Chinese factories reportedly pay workers extra to wear robotics data-collection gear daily — jwt0625 · 2026-10-24
- Ropedia Super Week: 6 embodied AI releases from $30M startup led by NTU's Ziwei Liu — liuziwei7 · 2026-10-24
- aernata shows autonomous aircraft routinely flying Alaska's backcountry skies — edgarpavlovsky · 2026-10-24