AI Paper Atlas

Research map / Architectures and efficiency / CNN design

SSH: Single Stage Headless Face Detector

Mahyar Najibi, Pouya Samangouei, Rama Chellappa, Larry S. Davis

ICCV 2017 · Architectures and efficiency › CNN design · approach: Biologically inspired CNN explanation · 516 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 image pyramid at inference to SSHSSH (VGG-16) · WIDER val (hard)AP: 81.4 -> 84.5Improved
add multiple detection modules M1 and M3 on different strides (vs only M2 from conv5-3) (ablation)SSH (VGG-16) · WIDER val (hard)AP: AP drops by ~12.8%Improved
add the context module (5x5 and 7x7 receptive field via sequential 3x3 convs) (ablation)SSH (VGG-16) · WIDER valmAP: more than half a percentImproved
add OHEM for anchor selection in each detection module (ablation)SSH (VGG-16) · WIDER valAPImproved
add conv4-3/conv5-3 feature fusion for module M1 (ablation)SSH (VGG-16) · WIDER valAPImproved
use finer anchor scale sets (5 scales per module) instead of 2 (ablation)SSH (VGG-16) · WIDER valAPHurt
use SSH without pyramid instead of HR (VGG-16) with pyramidSSH (VGG-16) · WIDER val (hard)AP: outperforms HR VGG-16 by 5.7%, 6.3%, and 6.5%Improved
use multiple aspect ratios for anchors (not used)SSH (VGG-16) · WIDERAPNo 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