Research map / Detection and segmentation / Open-set and few-shot detection
Using Syntax to Ground Referring Expressions in Natural Images
Volkan Cirik, Taylor Berg-Kirkpatrick, Louis-Philippe Morency
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 |
|---|---|---|---|
| Use syntax-guided module text spans rather than letting all modules attend to the full expression (ablation) | GroundNet · GoogleRef supporting-object annotations | supporting object accuracy: 10.6 -> 60.6 | Improved |
| Use syntax-guided module text spans rather than letting all modules attend to the full expression (ablation) | GroundNet · GoogleRef | accuracy: 68.9 -> 65.7 | Hurt |
| Restrict Locate nodes to their text spans while allowing Relate nodes to attend to the full expression (ablation) | GroundNet · GoogleRef supporting-object annotations | supporting object accuracy: 10.6 -> 60.0 | Improved |
| Restrict Locate nodes to their text spans while allowing Relate nodes to attend to the full expression (ablation) | GroundNet · GoogleRef | accuracy: 68.9 -> 66.7 | Hurt |
| Use dynamic, modular computation guided by syntax rather than a recursive sentence representation or fixed syntax-guided tuple | GroundNet · GoogleRef | accuracy: 51.5 -> 65.7; 53.5 -> 65.7 | Improved |
| Model supporting-object predictions with syntax-guided compositional modules | GroundNet · GoogleRef supporting-object annotations | supporting object accuracy: 11.1 -> 60.6; 15.0 -> 60.6 | Improved |
| Add visual difference features for object pairs | LSTM+CNN - MMI · GoogleRef | accuracy: 60.7 -> 64.0 | Improved |
Closest papers by meaning
- Adaptive Reconstruction Network for Weakly Supervised Referring Expression Grounding (ICCV 2019 · similarity 0.91)
- Grounding of Textual Phrases in Images by Reconstruction (ECCV 2016 · similarity 0.91)
- Grounding Referring Expressions in Images by Variational Context (CVPR 2018 · similarity 0.91)
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.90)
- Improved Visual Grounding through Self-Consistent Explanations (CVPR 2024 · similarity 0.89)
- Boosting Weakly Supervised Referring Image Segmentation via Progressive Comprehension (NeurIPS 2024 · similarity 0.89)
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.
