AI Paper Atlas

Research map / Detection and segmentation / Weakly supervised detection and segmentation

Semi-Weakly Supervised Object Kinematic Motion Prediction

Gengxin Liu, Qian Sun, Haibin Huang, Chongyang Ma, Yulan Guo, Li Yi, Hui Huang, Ruizhen Hu

CVPR 2023 · Detection and segmentation › Weakly supervised detection and segmentation · approach: Weakly supervised object localization · 11 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
augment partial-scan training with pseudo-labeled shapes using hierarchical semantic part segmentationANCSH · Partial scans, meanRotation Error °: 23.66 -> 17.43Improved
use semantic-segmentation-based pseudo labels instead of semi-supervised pseudo labels without that weak supervision (ablation)ANCSH · Partial scans, meanJoint States Error: 22.21 -> 13.35Improved
use fine motion types rather than coarse motion types (ablation)GNN motion-axis prediction · Pre-segmented shapes, 20 categories meanAngle Error°: 12.23 -> 8.93Improved
refine predicted axes using OBB candidate axes and geometric priors (ablation)GNN_fine · Pre-segmented shapes, 20 categories meanDistance Error: 0.205 -> 0.112Improved
train using 80% rather than 100% of labeled mobility data and augment with pseudo-labeled shapes (ablation)ANCSH (Semi-weakly) · Refrigerator partial scansRotation Error °: 5.07 10.05 10.09 -> 7.45 14.10 12.61Mixed
train using 50% rather than 100% of labeled mobility data and augment with pseudo-labeled shapes (ablation)ANCSH (Semi-weakly) · Refrigerator partial scansRotation Error °: 5.07 10.05 10.09 -> 12.86 23.37 18.63Hurt
augment ANCSH training with pseudo-labeled shapes generated by DSG-NetANCSH · StorageFurniture partial scansJoint Angle Error: 9.2 \\ 6.3 -> 1.5 \\ 1.5Improved

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