AI Paper Atlas

Research map / Detection and segmentation / CNN features

Supervised Transformer Network for Efficient Face Detection

Dong Chen, Gang Hua, Fang Wen, Jian Sun

ECCV 2016 · Detection and segmentation › CNN features · approach: Region-based and query-based detectors · 188 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 the multi-task RPN with facial landmarks (instead of direct rectangle regression) (ablation)Supervised Transformer Network · FDDBrecall rate at 10 false alarms: 87.1 -> 89.6Improved
add the Supervised Transformer to RPN + feature combination (ablation)Supervised Transformer Network · FDDBrecall rate at 10 false alarms: 88.8 -> 89.6Improved
add the Supervised Transformer to the multi-task RPN (ablation)Supervised Transformer Network · FDDBrecall rate at 10 false alarms: 87.1 -> 88.3Improved
add multi-granularity feature combination (ablation)Supervised Transformer Network · FDDBrecall rate at 10 false alarms: 88.3 -> 89.6Improved
add feature combination to the multi-task RPN (ablation)Supervised Transformer Network · FDDBrecall rate at 10 false alarms: 87.1 -> 88.8Improved
add the multi-task RPN to feature combination alone (ablation)Supervised Transformer Network · FDDBrecall rate at 10 false alarms: 88.0 -> 88.8Improved
replace NMS with non-top K suppression (K=3) (ablation)Supervised Transformer Network · FDDBrecall rate: 88.7 -> 89.3Improved
increase K in non-top K suppression from 3 to 5 (ablation)Supervised Transformer Network · FDDBrecall rate: 89.3 -> 89.4Improved
add ROI convolution with pre-filter threshold 0Supervised Transformer Network · FDDB (640x480)total time and recall: 107.2 -> 55.3 ms; 89.3% -> 89.2% recallImproved
add ROI convolution with pre-filter threshold 3Supervised Transformer Network · FDDB (640x480)total time and recall: 107.2 -> 29.3 ms; 89.3% -> 88.1% recallMixed
raise pre-filter threshold from 2 to 3Supervised Transformer Network · FDDB (640x480)total time and recall: 32.1 -> 29.3 ms; 88.7% -> 88.1% recallMixed
initialize canonical positions too large, too small or offsetSupervised Transformer Network · learned canonical positionscanonical point convergenceNo clear effect

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