AI Paper Atlas

Research map / 3D vision / Human pose estimation

Collaborative Learning of Gesture Recognition and 3D Hand Pose Estimation with Multi-Order Feature Analysis

Siyuan Yang, Jun Liu, Shijian Lu, Meng Hwa Er, Alex C. Kot

ECCV 2020 · 3D vision › Human pose estimation · approach: Depth-based hand pose regression · 47 citations (OpenAlex)

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
add multi-scale relation module (ablation)Baseline · FPHA gesture recognitionaccuracy (%): 72.17% -> 78.26%Improved
add multi-order multi-stream module (ablation)Baseline + multi-scale · FPHA gesture recognitionaccuracy (%): 78.26% -> 83.83%Improved
add a second collaborative iteration (ablation)Baseline + multi-scale + multi-order · FPHA gesture recognitionaccuracy (%): 83.83% -> 85.22%Improved
add first-order slow-fast features (ablation)Zero-order · FPHA gesture recognitionaccuracy (%): 80.87% -> 82.61%Improved
add second-order slow-fast features (ablation)Zero-order · FPHA gesture recognitionaccuracy (%): 80.87% -> 83.80%Improved
add both first and second order slow-fast features (ablation)Zero-order · FPHA gesture recognitionaccuracy (%): 80.87% -> 85.22%Improved
add slow features only (ablation)Zero-order · FPHA gesture recognitionaccuracy (%): 80.87% -> 82.96%Improved
add fast features only (ablation)Zero-order · FPHA gesture recognitionaccuracy (%): 80.87% -> 82.09%Improved
increase collaborative iterations from 1 to 2 (ablation)zero-order framework · FPHA 3D hand posePCK@30 (%): 87.2% -> 89.3%Improved
increase collaborative iterations from 2 to 5 (ablation)zero-order framework · FPHA 3D hand posePCK@30 (%): 89.3% -> 89.9%No clear effect
increase collaborative iterations from 1 to 5 (ablation)zero-order framework · FPHA gesture recognitionaccuracy (%): 78.3% -> 82.0%Improved
use gesture labels as weak supervision for pose with 15% to 40% pose labels (ablation)Baseline without gesture labels · FPHA 3D hand posePCK@30Improved

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy