AI Paper Atlas

Research map / Label-efficient and robust learning / Recognition and long-tail

GKGNet: Group K-Nearest Neighbor based Graph Convolutional Network for Multi-Label Image Recognition

Ruijie Yao, Sheng Jin, Lumin Xu, Wang Zeng, Wentao Liu, Chen Qian, Ping Luo, Ji Wu

ECCV 2024 · Label-efficient and robust learning › Recognition and long-tail · approach: Partial and multi-label learning · 13 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 Patch-Level graph module (ablation)GKGNet · MS-COCO 448x448mAP: 79.9 -> 82.5Improved
add Cross-Level graph (label nodes connected to patches) (ablation)GKGNet · MS-COCO 448x448mAP: 82.5 -> 85.5Improved
replace KNN with Group KNN graph construction (ablation)GKGNet · MS-COCO 448x448mAP: 85.5 -> 86.7Improved
apply Group KNN to Pyramid ViG-Ti for general classification (ablation)Pyramid ViG-Ti · CIFAR-100top-1 accuracy: 74.4 -> 76.5Improved
apply Group KNN to Pyramid ViG-Ti for general classification (ablation)Pyramid ViG-Ti · CIFAR-10top-1 accuracy: 94.6 -> 94.9Improved
use G=2 groups instead of G=1 (traditional KNN) (ablation)GKGNet · MS-COCOmAPImproved
increase G beyond 2 (ablation)GKGNet · MS-COCOmAPNo clear effect
use K=3 neighbors instead of larger K (ablation)GKGNet · MS-COCOmAPHurt
use GKGNet instead of Q2L for small objectsmulti-label recognition · MS-COCO small objectsmAP: 30.7 -> 35.6Improved
use GKGNet instead of Q2L under same ViG backbonemulti-label recognition · MS-COCO 448x448mAP: 85.9 -> 86.7Improved
use GKGNet instead of Q2L-R101multi-label recognition · MS-COCO 576x576mAP and FLOPs: 86.5 -> 87.7 mAP, 80.8 -> 40.1 GFLOPsImproved

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