A transformer predicts per-pixel CLIP embeddings, metric depth, NOCS coordinates, and fixed 3D Gaussians per object, supervised by canonical-space alignment after pose prediction, achieving SOTA on depth, segmentation, and category-level pose tasks.
Omni6dpose: A benchmark and model for universal 6d object pose estima- tion and tracking, 2024
1 Pith paper cite this work. Polarity classification is still indexing.
1
Pith paper citing it
fields
cs.CV 1years
2026 1verdicts
UNVERDICTED 1representative citing papers
citing papers explorer
-
OCH3R: Object-Centric Holistic 3D Reconstruction
A transformer predicts per-pixel CLIP embeddings, metric depth, NOCS coordinates, and fixed 3D Gaussians per object, supervised by canonical-space alignment after pose prediction, achieving SOTA on depth, segmentation, and category-level pose tasks.