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
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 changed | Where | Measured | Effect |
|---|---|---|---|
| add Patch-Level graph module (ablation) | GKGNet · MS-COCO 448x448 | mAP: 79.9 -> 82.5 | Improved |
| add Cross-Level graph (label nodes connected to patches) (ablation) | GKGNet · MS-COCO 448x448 | mAP: 82.5 -> 85.5 | Improved |
| replace KNN with Group KNN graph construction (ablation) | GKGNet · MS-COCO 448x448 | mAP: 85.5 -> 86.7 | Improved |
| apply Group KNN to Pyramid ViG-Ti for general classification (ablation) | Pyramid ViG-Ti · CIFAR-100 | top-1 accuracy: 74.4 -> 76.5 | Improved |
| apply Group KNN to Pyramid ViG-Ti for general classification (ablation) | Pyramid ViG-Ti · CIFAR-10 | top-1 accuracy: 94.6 -> 94.9 | Improved |
| use G=2 groups instead of G=1 (traditional KNN) (ablation) | GKGNet · MS-COCO | mAP | Improved |
| increase G beyond 2 (ablation) | GKGNet · MS-COCO | mAP | No clear effect |
| use K=3 neighbors instead of larger K (ablation) | GKGNet · MS-COCO | mAP | Hurt |
| use GKGNet instead of Q2L for small objects | multi-label recognition · MS-COCO small objects | mAP: 30.7 -> 35.6 | Improved |
| use GKGNet instead of Q2L under same ViG backbone | multi-label recognition · MS-COCO 448x448 | mAP: 85.9 -> 86.7 | Improved |
| use GKGNet instead of Q2L-R101 | multi-label recognition · MS-COCO 576x576 | mAP and FLOPs: 86.5 -> 87.7 mAP, 80.8 -> 40.1 GFLOPs | Improved |
Closest papers by meaning
- Multi-Label Image Recognition With Graph Convolutional Networks (CVPR 2019 · similarity 0.93)
- Learning Graph Convolutional Networks for Multi-Label Recognition and Applications (TPAMI 2021 · similarity 0.92)
- Attention-Driven Dynamic Graph Convolutional Network for Multi-Label Image Recognition (ECCV 2020 · similarity 0.91)
- GLDL: Graph Label Distribution Learning (AAAI 2024 · similarity 0.88)
- Enhancing the Influence of Labels on Unlabeled Nodes in Graph Convolutional Networks (ICML 2025 · similarity 0.87)
- HGDL: Heterogeneous Graph Label Distribution Learning (NeurIPS 2024 · similarity 0.86)
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.
