AI Paper Atlas

Research map / 3D vision / Human pose estimation

DeepIM: Deep Iterative Matching for 6D Pose Estimation

Yi Li, Gu Wang, Xiangyang Ji, Yu Xiang, Dieter Fox

ECCV 2018 · 3D vision › Human pose estimation · approach: Instance-level 6D object pose · 586 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
refine PoseCNN poses with DeepIMPoseCNN · LINEMOD6D Pose accuracy (%): 62.7 -> 88.6Improved
refine PoseCNN poses with DeepIMPoseCNN · Occlusion LINEMOD6D Pose accuracy (%): 26.9 -> 55.5Improved
refine PoseCNN poses with DeepIM (RGB)PoseCNN · YCB-VideoAUC of ADD(-S): 61.31 -> 81.9Improved
use tracking from previous frame instead of PoseCNN initDeepIM · YCB-VideoAUC of ADD(-S): 81.9 -> 85.9Improved
add iterative training (train 2 iterations vs 1), tested with 2 iterations (ablation)DeepIM · LINEMOD6D Pose accuracy (%): 79.0 -> 88.7Improved
test with 4 iterations when trained without iteration (ablation)DeepIM train_iter=1 · LINEMOD6D Pose accuracy (%): 77.9 -> 76.1Hurt
train/test with 4 iterations instead of 2 (ablation)DeepIM · LINEMOD6D Pose accuracy (%): 88.7 -> 88.6No clear effect
add zoom-in on bounding boxes (ablation)DeepIM · LINEMOD6D Pose accuracy (%): 65.2 -> 88.6Improved
use disentangled transformation instead of camera frame (ablation)DeepIM (no zoom) · LINEMOD6D Pose accuracy (%): 44.3 -> 65.2Improved
use disentangled representation instead of model frame (ablation)DeepIM · LINEMOD5cm 5deg accuracy (%): 79.2 -> 86.6Improved
use point matching loss instead of distance loss (ablation)DeepIM · LINEMOD6D Pose accuracy (%): 79.2 -> 88.6Improved
train one shared network instead of separate per-object networks (ablation)DeepIM · LINEMOD6D Pose accuracy (%): 87.6 -> 88.6Improved

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