AI Paper Atlas

Research map / 3D vision / Human pose estimation

Normalized Human Pose Features for Human Action Video Alignment

Jingyuan Liu, Mingyi Shi, Qifeng Chen, Hongbo Fu, Chiew-Lan Tai

ICCV 2021 · 3D vision › Human pose estimation · approach: Temporal video pose estimation · 17 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 pose normalization (FK onto a condition skeleton) before embedding (ablation)Ours (Pose Features) · Dense correspondencehit ratio (%): 53.69 -> 74.95Improved
add pose normalization before embedding (ablation)Ours (Pose Features) · Dense correspondenceMSE: 17.6033 -> 12.7106Improved
use adaptive triplet mining instead of fixed temporal windows (ablation)Ours (Pose Features) · Dense correspondencehit ratio (%): 41.92 -> 74.95Improved
use adaptive triplet mining instead of fixed temporal windows (ablation)Ours (Pose Features) · Dense correspondenceMSE: 28.7598 -> 12.7106Improved
use fully connected embedding network instead of SemGCN (ablation)Ours (Pose Features) · Dense correspondencehit ratio (%): 61.05 -> 74.95Improved
use fully connected embedding network instead of PointNet (ablation)Ours (Pose Features) · Dense correspondencehit ratio (%): 68.96 -> 74.95Improved
train normalization and embedding networks separately rather than jointly (ablation)Ours (Pose Features) · Dense correspondencehit ratio (%): 25.76 -> 74.95Improved
do not add a temporal convolution window to the pose feature (ablation)Ours (Pose Features) · Dense correspondencehit ratio (%): 74.95 -> 44.81Hurt
use normalized poses with L2 distance instead of learned pose featuresOurs · Dense correspondencehit ratio (%): 74.95 -> 94.25Improved
replace direct-supervised 3D pose estimators with FK-based normalization for anthropometry robustnessOurs · Human3.6M (bone-length augmented)Delta MPJPE (mm): 58.27 -> 53.35Mixed
use 3D pose-based Euler angles instead of 2D pose-based Euler anglesEuler angle feature · Dense correspondencehit ratio (%): 33.20 -> 66.92Improved

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