AI Paper Atlas

Research map / Architectures and efficiency / CNN design

CRF-CNN: Modeling Structured Information in Human Pose Estimation

Xiao Chu, Wanli Ouyang, hongsheng Li, Xiaogang Wang

NeurIPS 2016 · Architectures and efficiency › CNN design · approach: Filter design and trained-filter analysis · 37 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
use CRF-CNN instead of Chu et al.VGG pose estimator · LSPmean PCP: 81.1 -> 83.1Improved
use CRF-CNN instead of Yang et al.VGG pose estimator · LSPmean PCP: 81.1 -> 83.1Improved
use CRF-CNN instead of Chen&YuilleVGG pose estimator · LSPmean PCP: 75.0 -> 83.1Improved
use CRF-CNN instead of Chen and Yuillepose estimator · FLICPCK@0.2 elbow: 95.3 -> 98.0Improved
use CRF-CNN instead of Chen and Yuillepose estimator · FLICPCK@0.2 wrist: 92.4 -> 94.1Improved
pass messages twice instead of once with flooding (ablation)CRF-CNN tree · LSPmean PCP: 76.6 -> 77.1Improved
add extra connections between joints (loopy graph) to 2-iter flooding (ablation)CRF-CNN · LSPmean PCP: 77.1 -> 78.4Improved
replace flooding with serial message passing (ReLU) (ablation)CRF-CNN tree · LSPmean PCP: 77.1 -> 80.1Improved
replace ReLU with scaled softmax message nonlinearity (ablation)CRF-CNN serial tree · LSPmean PCP: 80.1 -> 83.1Improved
scale softmax with alpha=0.5 and beta=N_c (ablation)CRF-CNN · training convergenceconvergence speedImproved
serial instead of flooding message passing (ablation)CRF-CNN · LSPmean PCP: 78.4 -> 83.1Improved

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