Cascaded hand pose regression

TLDR

The work builds on prior 2D cascaded pose regression to improve performance on 3D articulated objects. The study extends 2D cascaded pose regression to 3D articulated objects by introducing 3D pose‑indexed features and a hierarchical regression adapted to the object structure. The method employs 3D pose‑indexed features that generalize 2D parameterized features for better invariance, and a principled hierarchical regression aligned with the articulated structure. The proposed approach is more accurate and faster, achieving state‑of‑the‑art accuracy and efficiency on public and new 3D hand pose datasets.

Abstract

We extends the previous 2D cascaded object pose regression work [9] in two aspects so that it works better for 3D articulated objects. Our first contribution is 3D pose-indexed features that generalize the previous 2D parameterized features and achieve better invariance to 3D transformations. Our second contribution is a principled hierarchical regression that is adapted to the articulated object structure. It is therefore more accurate and faster. Comprehensive experiments verify the state-of-the-art accuracy and efficiency of the proposed approach on the challenging 3D hand pose estimation problem, on a public dataset and our new dataset.