Research map / Graph learning / Knowledge and temporal graphs
Consistency-Aware Graph Network for Human Interaction Understanding
Zhenhua Wang, Jiajun Meng, Dongyan Guo, Jianhua Zhang, Javen Qinfeng Shi, Shengyong Chen
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 third-order graph network (TOGN) on top of Inception V3 features (ablation) | Inception V3 base model · TVHI ablation | F1 (%): 83.00 -> 90.41 | Improved |
| add TOGN on top of Inception V3 features (ablation) | Inception V3 base model · BIT ablation | F1 (%): 87.84 -> 91.26 | Improved |
| add TOGN on top of Inception V3 features (ablation) | Inception V3 base model · UT ablation | F1 (%): 92.20 -> 93.45 | Improved |
| add full CAR module (compatibility and transitivity) to TOGN (ablation) | TOGN · TVHI ablation | F1 (%): 90.41 -> 92.83 | Improved |
| add full CAR module to TOGN (ablation) | TOGN · TVHI ablation | mean IoU (%): 79.40 -> 84.02 | Improved |
| add full CAR module to TOGN (ablation) | TOGN · BIT ablation | F1 (%): 91.26 -> 92.79 | Improved |
| add compatibility oracle only to TOGN (ablation) | TOGN · TVHI ablation | F1 (%): 90.41 -> 91.90 | Improved |
| add transitivity oracle only to TOGN (ablation) | TOGN · TVHI ablation | F1 (%): 90.41 -> 90.35 | No clear effect |
| add transitivity oracle to compatibility-only CAR (ablation) | TOGN+CAR^C · TVHI ablation | F1 (%): 91.90 -> 92.83 | Improved |
| add compatibility oracle to transitivity-only CAR (ablation) | TOGN+CAR^T · TVHI ablation | F1 (%): 90.35 -> 92.83 | Improved |
| add CAR module to base model without TOGN (ablation) | Inception V3 base model · BIT ablation | F1 (%): 87.84 -> 88.72 | Improved |
| add CAR module to base model without TOGN (ablation) | Inception V3 base model · TVHI ablation | F1 (%): 83.00 -> 83.07 | No clear effect |
Closest papers by meaning
- Human Interaction Understanding With Consistency-Aware Learning (TPAMI 2023 · similarity 0.96)
- Cascaded Parsing of Human-Object Interaction Recognition (TPAMI 2021 · similarity 0.87)
- Contextual Heterogeneous Graph Network for Human-Object Interaction Detection (ECCV 2020 · similarity 0.86)
- Cascaded Human-Object Interaction Recognition (CVPR 2020 · similarity 0.86)
- Transferable Interactiveness Knowledge for Human-Object Interaction Detection (CVPR 2019 · similarity 0.86)
- HIG: Hierarchical Interlacement Graph Approach to Scene Graph Generation in Video Understanding (CVPR 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.
