AI Paper Atlas

Research map / Detection and segmentation / Open-set and few-shot detection

Pairwise Body-Part Attention for Recognizing Human-Object Interactions

Hao-Shu Fang, Jinkun Cao, Yu-Wing Tai, Cewu Lu

ECCV 2018 · Detection and segmentation › Open-set and few-shot detection · approach: Human-object interaction detection · 145 citations (OpenAlex)

Read the paper

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 object features (union box) (ablation)baseline model · HICO testmAP: 35.6 -> 37.0Improved
use union box instead of tight box for object features (ablation)baseline model · HICO testmAP: 36.3 -> 37.0Improved
add body-part features without attention (ablation)baseline + object · HICO testmAP: 37.0 -> 38.0Improved
use body-part pairs instead of single body parts (no attention) (ablation)body-part model · HICO testmAP: 38.0 -> 38.9Improved
add pairwise body-part attention with feature selection (ablation)body-part pairs model · HICO testmAP: 38.9 -> 39.9Improved
attend over both single parts and pairs (ablation)pairwise attention model · HICO testmAP: 39.9 -> 39.1Hurt
use feature selection (k=20) instead of keeping all 45 pairs (ablation)pairwise attention model · HICO valmAP: 1.2 mAP higher than k=45Improved
add weighted sigmoid cross entropy lossOurs · HICO testmAP: 37.5 -> 39.9Improved
replace VGG16 with ResNet101 base networkOurs · MPII testmAP: 36.8 -> 37.5Improved
use pairwise body-part attention vs Mallya & Lazebnik (weighted loss)Mallya & Lazebnik · HICO testmAP: 36.1 -> 39.9Improved
use pairwise body-part attention on apple buying classMallya & Lazebnik · HICO test, class: apple buyingAP: 19.3 -> 59.0Improved
use pairwise body-part attention on oven cleaning classMallya & Lazebnik · HICO test, class: oven cleaningAP: 22.1 -> 13.1Hurt

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