AI Paper Atlas

Research map / Detection and segmentation / CNN features

Body-Face Joint Detection via Embedding and Head Hook

Junfeng Wan, Jiangfan Deng, Xiaosong Qiu, Feng Zhou

ICCV 2021 · Detection and segmentation › CNN features · approach: Pedestrian and crowd detection · 16 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 embedding guidance to position-mode association (ablation)FPN + BFJ · CrowdHuman valmMR-2: 66.4 -> 55.7Improved
add head hook guidance to position-mode association (ablation)FPN + BFJ · CrowdHuman valmMR-2: 66.4 -> 54.2Improved
fuse embedding and head hook similarities (add embedding to head hook) (ablation)FPN + BFJ · CrowdHuman valmMR-2: 54.2 -> 52.5Improved
add embedding guidance to position-mode association (ablation)RetinaNet + BFJ · CrowdHuman valmMR-2: 73.8 -> 68.9Improved
add head hook guidance to position-mode association (ablation)RetinaNet + BFJ · CrowdHuman valmMR-2: 73.8 -> 64.5Improved
fuse both guidance cues (ablation)RetinaNet + BFJ · CrowdHuman valmMR-2: 64.5 -> 63.7Improved
add Angular loss to Smooth L1 head hook loss (ablation)FPN + BFJ · CrowdHuman valmMR-2: 53.7 -> 52.5Improved
add Angular loss to Smooth L1 head hook loss (ablation)RetinaNet + BFJ · CrowdHuman valmMR-2: 64.8 -> 63.7Improved
use BFJ association instead of cascade, position and position-with-head baselinesFPN · CrowdHuman valmMR-2: 67.2 -> 52.5Improved
use BFJ association on CrowdDet instead of baselinesCrowdDet · CrowdHuman valmMR-2: 66.1 -> 52.3Improved
use head as bridge between body and face in detection (POSH vs POS)FPN + POS · CrowdHuman valface MR-2: 54.3 -> 57.1Hurt
use cascade face detector on body crops (CAS vs POS)FPN · CrowdHuman valface AP@0.5: 70.3 -> 59.3Hurt

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