AI Paper Atlas

Research map / Detection and segmentation / CNN features

Faceness-Net: Face Detection through Deep Facial Part Responses

Shuo Yang, Ping Luo, Chen Change Loy, Xiaoou Tang

TPAMI 2017 · Detection and segmentation › CNN features · approach: Region-based and query-based detectors · 170 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 Faceness re-scoring to EdgeBox proposalsEdgeBox · FDDB face proposalsproposals needed for 90% detection rate: 600 -> 288Improved
add Faceness re-scoring to MCG proposalsMCG · FDDB face proposalsproposals needed for 90% detection rate: 942 -> 158Improved
add Faceness re-scoring to Selective Search proposalsSelective Search · FDDB face proposalsproposals needed for 90% detection rate: 641 -> 237Improved
add Faceness re-scoring of template proposals (ablation)Template Proposal · FDDB face proposalsproposals needed for 90% detection rate: 847 -> 351Improved
share representation across part networks (Faceness-SR) instead of five separate networksFaceness · FDDB face proposals (MCG)proposals needed for 75% detection rate: 13 -> 69Hurt
share representation across part networks (Faceness-SR) at high recallFaceness · FDDB face proposals (MCG)proposals needed for 90% detection rate: 158 -> 158No clear effect
train on uncropped instead of cropped images (ablation)attribute-aware networks · LFW part detection (350 proposals)part detection rate: Eye 95.87% -> 97.19%; Hair 95.56% -> 94.57%No clear effect
fine-tune with part-level attributes instead of face/non-face or all 25 attributes (ablation)attribute-aware networks · FDDB face proposalsdetection rateImproved
pretrain on facial attributes instead of ImageNet (ablation)multi-task CNN · face detection fine-tuningiterations to converge: 45,000 vs more than 200,000 iterationsImproved
use template proposal (TP) instead of generic proposalsFaceness-Net · FDDBdetection rate at high recall: over 96.5%Improved
use template proposal instead of generic proposals at fewer than 300 proposalsFaceness-Net · FDDBdetection rateHurt
add sharing representation and template proposal (SR-TP)Faceness-Net · FDDB discrete scorerecall: 90.99% -> +1.85%Improved

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