AI Paper Atlas

Research map / 3D vision / Human pose estimation

Joint Learning of Pose Regression and Denoising Diffusion with Score Scaling Sampling for Category-level 6D Pose Estimation

Seunghyun Lee, Tae-Kyun Kim

ICCV 2025 · 3D vision › Human pose estimation · approach: Instance-level 6D object pose · 5 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
pre-train encoder with direct 6D pose regression, then jointly train regression and diffusion heads (ablation)GenPose-style diffusion pose estimator · REAL275 training timetotal training time: 11 days -> 7.5 daysImproved
joint learning reaches baseline best accuracy (ablation)joint learning · REAL275 10deg5cmtraining time: 11 days -> 0.5 days to reach baseline bestImproved
add joint learning (J) vs diffusion-only training, K=50 without ranker (ablation)GenPose · REAL275mAP 5deg2cm: 49.4 -> 50.8Improved
add time-dependent score scaling guidance for single-pose sampling (ablation)GenPose · REAL275 K=1mAP 5deg2cm: 13.5 -> 52.2Improved
combine joint learning and score scaling guidance (ablation)single-pose sampling · REAL275 K=1mAP 5deg2cm: 52.2 -> 52.6Improved
remove energy-based ranker, halving parametersGenPose pipeline · REAL275parameters (M): 4.4 -> 2.2Improved
add score scaling guidance with DDIM sampler at 500 steps (ablation)DDIM sampler · REAL275 K=1mAP 5deg2cm: 15.9 -> 52.7Improved
reduce DDIM sampling steps from 500 to 10 with score scaling (ablation)DDIM + guidance · REAL275 K=1mAP 5deg2cm: 52.7 -> 50.8, 2.70 -> 42.9 FPSMixed
pre-train encoder with classification or reconstruction instead of pose regression (ablation)GenPose encoder · NOCS convergenceconvergence speedNo clear effect
geodesic loss instead of L2 loss for rotation regression pre-training (ablation)regression head · REAL275pre-training performanceImproved
constant score scaling schedule instead of exponential (ablation)score scaling guidance · REAL275 symmetric objectspose distribution diversity: 5deg2cm 52.7 vs 52.6 (no accuracy gain)Hurt
increase w_max from 1 to 4, then beyond 4 (ablation)exponential score scaling · REAL275pose accuracyMixed

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