Research map / Video / Video segmentation and tracking
Target Adaptive Context Aggregation for Video Scene Graph Generation
Yao Teng, Limin Wang, Zhifeng Li, Gangshan 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 HRTree context aggregation (scheme 1) (ablation) | TRACE · ImageNet-VidVRD (GT trajectories) | mAP: 28.62 -> 29.32 | Improved |
| use HRTree instead of fully connected graph (FC-G) (ablation) | TRACE · ImageNet-VidVRD (GT trajectories) | mAP: 28.11 -> 29.32 | Improved |
| use tree scheme 1 instead of scheme 2 (ablation) | TRACE · ImageNet-VidVRD (GT trajectories) | Recall@50: 17.70 -> 18.45 | Improved |
| add temporal fusion scheme 1 (ablation) | TRACE · ImageNet-VidVRD (GT trajectories) | mAP: 29.94 -> 29.32 | Hurt |
| add temporal fusion scheme 1 (ablation) | TRACE · ImageNet-VidVRD (GT trajectories) | Recall@100: 23.56 -> 23.85 | Improved |
| add temporal fusion scheme 2 (ablation) | TRACE · ImageNet-VidVRD (GT trajectories) | mAP: 29.94 -> 29.80 | No clear effect |
| add temporal fusion scheme 1 (ablation) | TRACE · Action Genome Recall@20 (lying on) | Recall@20: 19.55 -> 26.64 | Improved |
| add temporal fusion scheme 1 (ablation) | TRACE · Action Genome Recall@20 (beneath) | Recall@20: 47.65 -> 46.70 | Hurt |
| add temporal fusion scheme 2 (ablation) | TRACE · Action Genome Recall@20 (standing on) | Recall@20: 45.40 -> 46.08 | No clear effect |
| use 2 groups instead of 4 (ablation) | TRACE · ImageNet-VidVRD (GT trajectories) | mAP: 29.32 -> 29.81 | No clear effect |
| use HRTree with full parameters instead of reduced FC-G (ablation) | TRACE · Action Genome SGDet | Recall@20: 32.32 -> 33.41 | Improved |
| use HRTree instead of fully connected graph at reduced parameters (ablation) | TRACE · Action Genome SGDet | Recall@20: 32.32 -> 32.24 | No clear effect |
Closest papers by meaning
- OED: Towards One-stage End-to-End Dynamic Scene Graph Generation (CVPR 2024 · similarity 0.88)
- A Closer Look at Dynamic Scene Graph Generation in the Era of Multimodal Large Language Models (NeurIPS 2026 · similarity 0.88)
- Weakly Supervised Video Scene Graph Generation via Natural Language Supervision (ICLR 2025 · similarity 0.88)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · similarity 0.86)
- DiffVsgg: Diffusion-Driven Online Video Scene Graph Generation (CVPR 2025 · similarity 0.86)
- HIG: Hierarchical Interlacement Graph Approach to Scene Graph Generation in Video Understanding (CVPR 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.
