Newest papers

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 · 2026 · arXiv

DOI: 10.48550/arXiv.2610.31623 · open-access PDF

Indonesian

Dokumen ini mempelajari perkiraan posisi objek RGB-D tingkat kategori, yang memulihkan rotasi dan terjemahan 3D objek tanpa model CAD khusus atau tampilan referensi. Tugas ini sangat ambigu karena variasi dalam kelas, simetri, dan pengamatan parsial. Solusi saat ini berjuang untuk menyeimbangkan keandalan dan efisiensi: metode yang didasarkan pada struktur menengah cenderung untuk menyebarkan kesalahan, sementara model difusi menderita latensi inferensi tinggi dan ekstraksi kandidat post-hoc yang kompleks.

Machine translation (Meta M2M100); the English original follows.

Abstract

This paper studies category-level RGB-D object pose estimation, which recovers an object's 3D rotation and translation without instance-specific CAD models or reference views. This task is highly ambiguous due to intra-class variations, symmetries, and partial observations. Current solutions struggle to balance reliability and efficiency: methods based on intermediate structures are prone to error propagation, while diffusion models suffer from high inference latency and complex post-hoc candidate extraction. To address this, we present F2G-Pose, a real-time direct, single-pass framework that

Read in: English · Spanish · Chinese · Hindi · Arabic · Portuguese · French · German · Japanese · Russian · Korean · Indonesian · Vietnamese

Sources: OpenAlex (CC0), arXiv, Crossref. Agent Exchange