AI Paper Atlas

Research map / 3D vision / Human pose estimation

DiffPose: SpatioTemporal Diffusion Model for Video-Based Human Pose Estimation

Runyang Feng, Yixing Gao, Tze Ho Elden Tse, Xueqing Ma, Hyung Jin Chang

ICCV 2023 · 3D vision › Human pose estimation · approach: Temporal video pose estimation · 61 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 LMSFI (lookup-based multi-scale feature interaction) to the direct concatenation baseline (ablation)DiffPose · PoseTrack2017 valmAP: 52.5 -> 82.4Improved
add spatiotemporal representation learner (STRL) as condition (ablation)DiffPose with LMSFI · PoseTrack2017 valmAP: 82.4 -> 86.4Improved
add multi-scale fusion over single low-resolution interaction (ablation)LMSFI · PoseTrack2017 valmAP: 84.4 -> 86.4Improved
use concatenation and aggregation instead of the default multi-scale fusion (ablation)LMSFI · PoseTrack2017 valmAP: 86.4 -> 86.2No clear effect
increase number of initial noises for pose ensemble from 1 to 5 to 10 (ablation)DiffPose · PoseTrack2017 valmAP: 80.4 -> 84.0 -> 86.4Improved
increase ensemble size N from 1 to 10 on challenging joints (ablation)DiffPose · PoseTrack2017 val (wrist, ankle)mAP: wrist 74.6 -> 83.5; ankle 74.1 -> 80.2Improved
increase denoising steps from 1 to 2 to 4 at inference without retraining (ablation)DiffPose · PoseTrack2017 valmAP: 85.9 -> 86.3 -> 86.4Improved
increase temporal interval delta from 0 to 1 to 2 (ablation)STRL · PoseTrack2017 valmAP: 82.4 -> 84.5 -> 86.4Improved
replace deterministic pose estimator FAMI-Pose with DiffPose diffusion frameworkvideo pose estimation · PoseTrack21mAP: 81.2 -> 82.9Improved

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