Research map / Video / Video segmentation and tracking
SegPVSG: Panoptic Video Scene Graph Generation via Temporal Focusing and Generative Augmentation
YiKai Li, Quhui Ke, Jinglin Liang, Zhi-Yuan Zhang, Zhidi Lin, Shuangping Huang
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 RGVA generative augmentation (ablation) | TFN · OpenPVSG IPS+T | mR@100: 12.87 -> 15.91 | Improved |
| use Gaussian mask temporal focus instead of hard mask (ablation) | TFN · OpenPVSG IPS+T | mR@100: 12.11 -> 12.87 | Improved |
| use Gaussian mask focus instead of Transformer encoder (ablation) | TFN · OpenPVSG IPS+T | mR@100: 12.23 -> 12.87 | Improved |
| add loss reweighting (ablation) | TFN · OpenPVSG IPS+T | R@50 and mR@100: R@50 9.43 -> 9.12; mR@100 8.79 -> 12.87 | Mixed |
| add RGVA to Transformer baseline (ablation) | Transformer · OpenPVSG IPS+T | mR@100: 4.44 -> 7.19 | Improved |
| add RGVA to Convolution baseline (ablation) | Convolution · OpenPVSG IPS+T | mR@100: 5.36 -> 6.01 | Improved |
| combine RGVA with loss reweighting (ablation) | TFN · OpenPVSG IPS+T | mR@100: 11.51 -> 15.91 | Improved |
| add category semantic embedding to visual entity features (ablation) | TFN · OpenPVSG IPS+T | mR@100: 10.98 -> 12.87 | Improved |
| use average instead of max score aggregation (ablation) | TFN · OpenPVSG IPS+T | mR@100: 8.11 -> 12.87 | Improved |
| use ground-truth intervals instead of predicted intervals for classifier supervision (ablation) | TFN · OpenPVSG IPS+T | mR@100: 8.11 -> 12.87 | Improved |
| use K=8 temporal queries (ablation) | TFN · OpenPVSG IPS+T | recall | Improved |
| use ground-truth entity labels at evaluation (ablation) | TFN · OpenPVSG IPS+T | mR@100: 12.87 -> 31.75 | Improved |
Closest papers by meaning
- Panoptic Video Scene Graph Generation (CVPR 2023 · similarity 0.92)
- TextPSG: Panoptic Scene Graph Generation from Textual Descriptions (ICCV 2023 · similarity 0.88)
- A Closer Look at Dynamic Scene Graph Generation in the Era of Multimodal Large Language Models (NeurIPS 2026 · similarity 0.87)
- LASER: A Neuro-Symbolic Framework for Learning Spatio-Temporal Scene Graphs with Weak Supervision (ICLR 2025 · similarity 0.87)
- DiffVsgg: Diffusion-Driven Online Video Scene Graph Generation (CVPR 2025 · similarity 0.87)
- DSGG: Dense Relation Transformer for an End-to-end Scene Graph Generation (CVPR 2024 · 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.
