AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

UKPGAN: A General Self-Supervised Keypoint Detector

Yang You, Wenhai Liu, Yanjie Ze, Yong-Lu Li, Weiming Wang, Cewu Lu

CVPR 2022 · Detection and segmentation › Open-set and few-shot detection · approach: Unsupervised object and part discovery · 30 citations (OpenAlex)

Check 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 GAN-based sparsity control (ablation)UKPGAN · KeypointNet airplane, chair, and tableIoU (%): 36.3 -> 68.8; 27.2 -> 36.2; 23.1 -> 34.7Improved
add GAN-based sparsity control (ablation)UKPGAN · KeypointNet airplane, chair, and tableRotation Rep. (%): 99.8 -> 98.3; 95.9 -> 88.3; 99.6 -> 90.6Hurt
add salient information distillation instead of averaging (ablation)UKPGAN · KeypointNet airplane, chair, and tableIoU (%): 51.8 -> 68.8; 33.3 -> 36.2; 19.0 -> 34.7Improved
add salient information distillation instead of averaging (ablation)UKPGAN · KeypointNet airplane, chair, and tableRotation Rep. (%): 94.1 -> 98.3; 87.5 -> 88.3; 99.9 -> 90.6Mixed
use local rotation-invariant features instead of raw XYZ features (ablation)UKPGAN · KeypointNet airplane, chair, and tableIoU (%): 22.4 -> 68.8; 16.0 -> 36.2; 21.2 -> 34.7Improved
use local rotation-invariant features instead of raw XYZ features (ablation)UKPGAN · KeypointNet airplane, chair, and tableRotation Rep. (%): 15.4 -> 98.3; 4.9 -> 88.3; 0.7 -> 90.6Improved
add symmetric regularization (ablation)UKPGAN · KeypointNet airplane, chair, and tableIoU (%): 54.9 -> 68.8; 20.0 -> 36.2; 22.3 -> 34.7Improved
add symmetric regularization (ablation)UKPGAN · KeypointNet airplane, chair, and tableRotation Rep. (%): 85.1 -> 98.3; 77.0 -> 88.3; 73.0 -> 90.6Improved
use UKPGAN keypoints with PerfectMatch rather than random keypoints3DMatch registration · 3DMatch, 100 keypointsRegistration Recall (%): 16.1 -> 37.6Improved
use UKPGAN keypoints with PerfectMatch rather than random keypointsETH registration · ETH, 2500 keypointsRegistration Recall (%): 50.1 -> 58.2Improved
use UKPGAN keypoints with D3Feat descriptors rather than the D3Feat detector3DMatch registration · 3DMatch, 100 keypointsRegistration Recall (%): 67.2 -> 47.4Hurt
use UKPGAN keypoints with D3Feat descriptors rather than the D3Feat detectorETH registration · ETH, 2500 keypointsRegistration Recall (%): 29.2 -> 28.3Hurt

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