Research map / Vision-language and multimodal / Human and embodied modeling
Self-Supervised Relationship Probing
Jiuxiang Gu, Jason Kuen, Shafiq Joty, Jianfei Cai, Vlad I. Morariu, Handong Zhao, Tong Sun
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 data augmentation to pretraining corpus (ablation) | SSRPShare · NLVR2 dev | accuracy: 60.53 -> 61.67 | Improved |
| add data augmentation to pretraining corpus (ablation) | SSRPVisual · NLVR2 dev | accuracy: 69.92 -> 70.75 | Improved |
| add data augmentation to pretraining corpus (ablation) | SSRPCross · NLVR2 dev | accuracy: 74.35 -> 74.48 | Improved |
| use cross-attention (SSRPCross) instead of visual-only attention (ablation) | SSRPVisual · NLVR2 dev | accuracy: 72.03 -> 75.71 | Improved |
| use separate visual attention (SSRPVisual) instead of shared (ablation) | SSRPShare · NLVR2 dev | accuracy: 64.25 -> 72.03 | Improved |
| add language relationships Rw to falign (ablation) | SSRPCross · NLVR2 dev | accuracy: 74.48 -> 74.68 | Improved |
| add visual relationships Rv to falign (ablation) | SSRPCross · NLVR2 dev | accuracy: 74.48 -> 74.25 | No clear effect |
| combine visual and textual relationships Rv+Rw (ablation) | SSRPCross · NLVR2 dev | accuracy: 74.48 -> 75.71 | Improved |
| combine visual and textual relationships Rv+Rw (ablation) | SSRPShare · NLVR2 dev | accuracy: 61.67 -> 64.25 | Improved |
| combine visual and textual relationships Rv+Rw (ablation) | SSRPVisual · NLVR2 dev | accuracy: 70.75 -> 72.03 | Improved |
| compare SSRPCross against prior pretraining models | SSRPCross · NLVR2 dev | accuracy: 75.71 vs 74.9 (LXMERT) | Improved |
| use SSRPVisual refined object features in Up-Down captioner | Up-Down · MSCOCO test split | CIDEr: 120.8 -> 126.7 | Improved |
Closest papers by meaning
- Visual Relationship Detection With Internal and External Linguistic Knowledge Distillation (ICCV 2017 · similarity 0.90)
- Visual Relationship Detection with Language Priors (ECCV 2016 · similarity 0.89)
- Scene-Graph ViT: End-to-End Open-Vocabulary Visual Relationship Detection (ECCV 2024 · similarity 0.89)
- MR-NET: Exploiting Mutual Relation for Visual Relationship Detection (AAAI 2019 · similarity 0.89)
- Remodeling Semantic Relationships in Vision-Language Fine-Tuning (AAAI 2026 · similarity 0.88)
- Groupwise Query Specialization and Quality-Aware Multi-Assignment for Transformer-based Visual Relationship Detection (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.
