Research map / Probabilistic and scientific modeling / Causal discovery
Unbiased Scene Graph Generation via Two-Stage Causal Modeling
Shuzhou Sun, Shuaifeng Zhi, Qing Liao, Janne Heikkilä, Li 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 |
|---|---|---|---|
| apply TsCM (P-Loss plus AL-Adjustment) | MotifsNet · VG150 PredCls | AVG mR@K: 14.8 -> 36.8 | Improved |
| apply TsCM | MotifsNet · VG150 SGCls | AVG mR@K: 8.6 -> 21.6 | Improved |
| apply TsCM | MotifsNet · VG150 SGDet | AVG mR@K: 7.0 -> 16.9 | Improved |
| TsCM versus HML (hybrid debiasing) | MotifsNet · VG150 PredCls | AVG mR@K: 35.0 -> 36.8 | Improved |
| TsCM versus GCL | MotifsNet · VG150 SGDet | AVG mR@K: 16.3 -> 16.9 | Improved |
| add TDE counterfactual adjustment | MotifsNet · VG150 PredCls | AVG mR@K: 14.8 -> 24.4 | Improved |
| add loss reweighting | MotifsNet · VG150 PredCls | AVG mR@K: 14.8 -> 31.6 | Improved |
| add logit reweighting | MotifsNet · VG150 PredCls | AVG mR@K: 14.8 -> 14.8 | No clear effect |
| add logit reweighting | MotifsNet · VG150 SGCls | AVG mR@K: 8.6 -> 7.4 | Hurt |
| add EBM-loss | MotifsNet · VG150 PredCls | AVG mR@K: 14.8 -> 17.2 | Improved |
| apply TsCM-style debiasing: FGPL | VCTree · VG150 PredCls | AVG mR@K: 14.8 -> 36.2 | Improved |
Closest papers by meaning
- A Reverse Causal Framework to Mitigate Spurious Correlations for Debiasing Scene Graph Generation (TPAMI 2025 · similarity 0.95)
- TD²-Net: Toward Denoising and Debiasing for Video Scene Graph Generation (AAAI 2024 · similarity 0.88)
- Hydra-SGG: Hybrid Relation Assignment for One-stage Scene Graph Generation (ICLR 2025 · similarity 0.88)
- Noise-Guided Predicate Representation Extraction and Diffusion-Enhanced Discretization for Scene Graph Generation (ICML 2025 · similarity 0.88)
- Panoptic Scene Graph Generation with Semantics-Prototype Learning (AAAI 2024 · similarity 0.87)
- Towards Unbiased and Robust Spatio-Temporal Scene Graph Generation and Anticipation (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.
