Research map / Video / Video segmentation and tracking
TRKT: Weakly Supervised Dynamic Scene Graph Generation with Temporal-enhanced Relation-aware Knowledge Transferring
Zhu Xu, Ting Lei, Zhimin Li, Guan Wang, Qingchao Chen, Yuxin Peng, Yang Liu
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 Inter-frame Attention Augmentation (optical-flow temporal cues) on top of CBM+LRM (ablation) | TRKT · Action Genome object detection | AP (maxDets=1): 14.1 -> 23.0 | Improved |
| add IAA to attention-map-only detections (ablation) | Internal detections · Action Genome object detection | AP (maxDets=1): 2.8 -> 5.3 | Improved |
| combine attention-map detections with external detections (ablation) | PLA · Action Genome WS-DSGG (With Constraint) | R@10: 14.32 -> 16.52 | Improved |
| use attention-map detections alone instead of external detector (ablation) | PLA · Action Genome object detection | AP (maxDets=1): 11.4 -> 2.8 | Hurt |
| add Confidence Boosting Module (ablation) | PLA · Action Genome object detection | AP (maxDets=1): 11.4 -> 12.6 | Improved |
| add Localization Refinement Module (ablation) | PLA · Action Genome object detection | AP (maxDets=1): 11.4 -> 13.4 | Improved |
| combine CBM and LRM (ablation) | PLA · Action Genome object detection | AP (maxDets=1): 11.4 -> 14.1 | Improved |
| add object tokens to class decoder (ablation) | PLA · Action Genome object detection | AP (maxDets=1): 11.4 -> 21.6 | Improved |
| add relation tokens to object tokens (ablation) | TRKT · Action Genome object detection | AP (maxDets=1): 21.6 -> 23.0 | Improved |
| add relation tokens to object tokens (ablation) | TRKT · Action Genome object detection | AR (maxDets=1): 31.8 -> 28.8 | Hurt |
| replace PLA with TRKT detections for pseudo labels | WS-DSGG · Action Genome (No Constraints) | R@10: 15.87 -> 18.76 | Improved |
| apply TRKT on UCML for tracklet-level WS-DSGG | UCML · VidVRD | mAP: 17.17% -> 17.93% | Improved |
Later papers that compared with this method
1 later accepted paper reports TRKT as a baseline in their results tables; 1 of them report a better number on most of the results they share with it.
- Revisiting Weakly-Supervised Video Scene Graph Generation via Pair Affinity Learning (ECCV 2026 · better on 2 of 2 shared results)
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- OED: Towards One-stage End-to-End Dynamic Scene Graph Generation (CVPR 2024 · similarity 0.91)
- A Closer Look at Dynamic Scene Graph Generation in the Era of Multimodal Large Language Models (NeurIPS 2026 · similarity 0.91)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · similarity 0.90)
- TD²-Net: Toward Denoising and Debiasing for Video Scene Graph Generation (AAAI 2024 · similarity 0.88)
- Weakly Supervised Video Scene Graph Generation via Natural Language Supervision (ICLR 2025 · similarity 0.88)
- DiffVsgg: Diffusion-Driven Online Video Scene Graph Generation (CVPR 2025 · similarity 0.87)
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.
