AI Paper Atlas

Research map / Detection and segmentation / CNN features

A Coarse-Fine Network for Keypoint Localization

Shaoli Huang, Mingming Gong, Dacheng Tao

ICCV 2017 · Detection and segmentation › CNN features · approach: Local features and context in CNNs · 196 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 multi-level supervision instead of strict supervision only (ablation)CFN unified detection · CUB200-2011 (alpha=0.1)PCK (%): 83.7 -> 88.0Improved
add multi-level supervision instead of strict supervision only (ablation)CFN unified detection · CUB200-2011 (alpha=0.05)PCK (%): 69.3 -> 71.0Improved
add multi-level supervision instead of strict supervision only (ablation)CFN unified detection · CUB200-2011 (alpha=0.02)PCK (%): 27.3 -> 27.3No clear effect
add multi-level supervision instead of strict supervision only (ablation)coarse detector 4a · CUB200-2011 (alpha=0.1)PCK (%): 66.1 -> 79.2Improved
add multi-level supervision instead of strict supervision only (ablation)coarse detector 4b · CUB200-2011 (alpha=0.1)PCK (%): 59.6 -> 84.9Improved
add multi-level supervision instead of strict supervision only (ablation)coarse detector 4c · CUB200-2011 (alpha=0.1)PCK (%): 79.9 -> 82.0Improved
add multi-level supervision instead of strict supervision only (ablation)coarse detector 4c · CUB200-2011 (alpha=0.05)PCK (%): 66.6 -> 52.4Hurt
add multi-level supervision instead of strict supervision only (ablation)fine detector · CUB200-2011 (alpha=0.1)PCK (%): 80.8 -> 80.8No clear effect
add deep multi-scale CFN instead of prior best methodCFN · CUB200-2011total PCP (%): 69.1 -> 72.1Improved
use CFN instead of second best method under strict PCKCFN · CUB200-2011 (alpha=0.05)mean PCK (%): 52.4 -> 70.9Improved
use CFN instead of Yu et al. without extra MPII dataCFN · LSP (PCK@0.2)PCK (%): 84.3 -> 84.5No clear effect
use CFN instead of G-RMI (COCO-int)CFN · COCO keypoints test-devAP: 68.5 -> 72.6Improved

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