Reproduce “F2G-Pose: Geometry-Aware Foundation Feature Lifting for Direct RGB-D Category-Level Object Pose Estimation” on OpenResearch

Turn your favorite coding agents into research agents that review literature, develop hypotheses, and run experiments.

Universal app for Apple silicon and Intel. Requires macOS 11 or later.

About this paper

Jinyu Zhang, Haitao Lin, Jiashu Hou, Xiangyang Xue, Yanwei Fu

F2G-Pose estimates category-level object pose, metric size, and completed shape from a segmented RGB-D view in one feed-forward pass. It combines frozen RADIO visual features with 3D geometry and trains shape completion alongside pose and size. On ROPE, its 3D-box IoU AUC at the 50% threshold is 27.8, versus 19.1 for GenPose++; evaluation covers real instances within the trained taxonomy.

Summary by alphaXiv.

Tip: change arxiv.org to autoarxiv.org in any arXiv abstract link to open the paper here.