AI Paper Atlas

Research map / Probabilistic and scientific modeling / Uncertainty and Gaussian processes

Face Alignment With Kernel Density Deep Neural Network

Lisha Chen, Hui Su, Qiang Ji

ICCV 2019 · Probabilistic and scientific modeling › Uncertainty and Gaussian processes · approach: Priors and functional variational uncertainty · 36 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 Gaussian kernel density loss instead of softlabel lossHourglass cascade · 300W-testNME / NLL: NME 2.32 -> 2.21; NLL 4.67 -> 2.93Improved
use Gaussian kernel density loss instead of softlabel lossHourglass cascade · COFW-68 testNME / NLL: NME 2.92 -> 2.73; NLL 5.27 -> 3.21Improved
use Gaussian kernel density loss instead of softlabel lossHourglass cascade · Menpo-frontalNME / AUC: NME 2.27 -> 2.01; AUC 67.4 -> 71.1Improved
use Gaussian kernel instead of uniform kernel (ablation)KDN · 300W-testNME: 2.38 -> 2.21Improved
use Gaussian kernel instead of uniform kernel (ablation)KDN · COFW-68 testNME: 2.92 -> 2.73Improved
use KDN loss instead of softlabel in a single stage (ablation)single-stage Hourglass · 300W-testNME / NLL: NME 2.58 -> 2.52; NLL 4.79 -> 3.01Improved
use Gaussian KDN instead of softlabelHourglass cascade · COFW-29NME / NLL: NME 2.51 -> 2.28; NLL 6.13 -> 3.19Improved
use Gaussian KDN instead of softlabelHourglass cascade · AFLW-fullNME / NLL: NME 2.87 -> 2.80; NLL 5.75 -> 3.56Improved
use KDN uncertainty instead of softlabel pseudo-variance for occlusion detectionKDN-Gaussian · COFW-68occlusion precision/recall: precision 56/40 -> 70/40Improved
add multi-stage cascade with probability propagation (ablation)KDN · 300W-testNME and uncertaintyImproved
use KDN-Gaussian instead of Gaussian for AU intensityfacial AU intensity · BP4DMAE: 0.748 -> 0.757Hurt
use KDN-Gaussian instead of KDN-Uniform for AU intensity (ablation)facial AU intensity · BP4DICC: 0.559 -> 0.588Improved

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