Research map / Detection and segmentation / Weakly supervised detection and segmentation
Weakly-Supervised Visual Grounding of Phrases With Linguistic Structures
Fanyi Xiao, Leonid Sigal, Yong Jae Lee
Check 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 parent-child and sibling structural constraints to discriminative training | phrase-grounding model · Visual Genome | pointing game accuracy: 0.230 -> 0.244 | Improved |
| add parent-child constraint (ablation) | phrase-grounding model · Visual Genome | pointing game accuracy: 0.230 -> 0.236 | Improved |
| add sibling constraint alone (ablation) | phrase-grounding model · Visual Genome | pointing game accuracy: 0.230 -> 0.231 | No clear effect |
| train using phrases rather than noun tokens | phrase-grounding model · Visual Genome | pointing game accuracy: 0.222 -> 0.230 | Improved |
| add parent-child and sibling structural constraints to discriminative training | phrase-grounding model · MS COCO val2014 | Avg mAP: 0.203 -> 0.251 | Improved |
| add parent-child constraint (ablation) | phrase-grounding model · MS COCO val2014 | Avg mAP: 0.203 -> 0.219 | Improved |
| add sibling constraint (ablation) | phrase-grounding model · MS COCO val2014 | Avg mAP: 0.203 -> 0.211 | Improved |
| train using noun tokens rather than phrases | phrase-grounding model · MS COCO val2014 | Avg mAP: 0.203 -> 0.238 | Improved |
| combine parent-child and sibling constraints rather than use either constraint alone (ablation) | phrase-grounding model · Visual Genome | pointing game accuracy: PC 0.236; SIB 0.231; Ours 0.244 | Improved |
| combine parent-child and sibling constraints rather than use either constraint alone (ablation) | phrase-grounding model · MS COCO val2014 | Avg mAP: PC 0.219; SIB 0.211; Ours 0.251 | Improved |
Closest papers by meaning
- Align2Ground: Weakly Supervised Phrase Grounding Guided by Image-Caption Alignment (ICCV 2019 · similarity 0.92)
- Improving Weakly Supervised Visual Grounding by Contrastive Knowledge Distillation (CVPR 2021 · similarity 0.92)
- Contrastive Learning for Weakly Supervised Phrase Grounding (ECCV 2020 · similarity 0.91)
- Weakly-Supervised Generation and Grounding of Visual Descriptions With Conditional Generative Models (CVPR 2022 · similarity 0.91)
- Grounding of Textual Phrases in Images by Reconstruction (ECCV 2016 · similarity 0.90)
- Improved Visual Grounding through Self-Consistent Explanations (CVPR 2024 · similarity 0.88)
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.
