Research map / Graph learning / Knowledge and temporal graphs
Graph-Based Social Relation Reasoning
Wanhua Li, Yueqi Duan, Jiwen Lu, Jianjiang Feng, Jie Zhou
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 |
|---|---|---|---|
| use graph relational reasoning over whole image instead of pair-wise CNN (224 input) | Pair CNN · PIPA social relation | accuracy (%): 58.0 -> 61.3 | Improved |
| increase input resolution from 224 to 448 (ablation) | GR2N · PIPA social relation | accuracy (%): 61.3 -> 64.3 | Improved |
| increase input resolution from 224 to 448 (ablation) | GR2N · PIPA social domain | accuracy (%): 69.3 -> 72.3 | Improved |
| increase input resolution from 224 to 448 (ablation) | GR2N · PISC coarse relationships | mAP (%): 72.2 -> 83.1 | Improved |
| increase input resolution from 224 to 448 (ablation) | GR2N · PISC fine relationships | mAP (%): 68.6 -> 72.7 | Improved |
| use GR2N over GCN on a fully connected graph (ablation) | GCN · PIPA social relation | accuracy (%): 59.3 -> 61.3 | Improved |
| use GR2N over GGNN on a fully connected graph (ablation) | GGNN · PIPA social relation | accuracy (%): 59.8 -> 61.3 | Improved |
| use any image-based GNN over Pair CNN (ablation) | Pair CNN · PIPA social relation | accuracy (%): 58.0 -> 59.3 | Improved |
| jointly reason over whole image vs per-pair processing | Pair CNN · PISC coarse relationships | mAP (%): 65.1 -> 83.1 | Improved |
| jointly reason over whole image vs per-pair processing | Pair CNN · PISC fine relationships | mAP (%): 48.2 -> 72.7 | Improved |
| process all relations in one pass | Pair CNN · PIPA runtime (batch size 1) | forward time (s/image): 0.077 -> 0.046 | Improved |
| GR2N-448 vs GRM runtime | GRM · PIPA runtime (batch size 1) | forward time (s/image): 0.294 -> 0.046 | Improved |
Closest papers by meaning
- An End-To-End Network for Generating Social Relationship Graphs (CVPR 2019 · similarity 0.91)
- Social Relation Reasoning Based on Triangular Constraints (AAAI 2023 · similarity 0.90)
- Social Relation Recognition From Videos via Multi-Scale Spatial-Temporal Reasoning (CVPR 2019 · similarity 0.88)
- SocialGPT: Prompting LLMs for Social Relation Reasoning via Greedy Segment Optimization (NeurIPS 2024 · similarity 0.87)
- Linking People across Text and Images Based on Social Relation Reasoning (AAAI 2023 · similarity 0.87)
- MINES: Message Intercommunication for Inductive Relation Reasoning over Neighbor-Enhanced Subgraphs (AAAI 2024 · similarity 0.84)
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.
