Research map / Detection and segmentation / Open-set and few-shot detection
Referring Expression Generation and Comprehension via Attributes
Jingyu Liu, Liang Wang, Ming-Hsuan Yang
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 learned attributes to the comprehension embedding | MMI · RefCOCO TestA (ground-truth boxes) | accuracy: 71.72% -> 78.12% | Improved |
| add learned attributes to the comprehension embedding | MMI · RefCOCOg Val (ground-truth boxes) | accuracy: 62.14% -> 67.43% | Improved |
| add learned attributes to the comprehension embedding | MMI · RefCOCO TestB (ground-truth boxes) | accuracy: 71.09% -> 75.89% | Improved |
| add learned attributes to the comprehension embedding | MMI · RefCOCO+ TestB (detected boxes) | accuracy: 42.81% -> 43.14% | Improved |
| add learned attributes to the comprehension embedding | MMI · RefCOCO TestA (detected boxes) | accuracy: 64.90% -> 70.55% | Improved |
| add learned attributes to the comprehension embedding | MMI · RefCOCOg Val (detected boxes) | accuracy: 45.85% -> 50.02% | Improved |
| add visdif to attr+MMI (ablation) | attr+MMI · RefCOCO TestB (ground-truth boxes) | accuracy: 75.89% -> 78.07% | Improved |
| add visdif to attr+MMI (ablation) | attr+MMI · RefCOCOg Val (ground-truth boxes) | accuracy: 67.43% -> 69.83% | Improved |
| add attributes to the generation LSTM | baseline · RefCOCO TestA | ROUGE: 0.413 -> 0.472 | Improved |
| add attributes to the generation LSTM | baseline · RefCOCO+ TestA | ROUGE: 0.356 -> 0.362 | Improved |
| add attributes to the generation LSTM | baseline · RefCOCOg Val | ROUGE: 0.363 -> 0.389 | Improved |
| add visdif to attr generation model (ablation) | attr · RefCOCOg Val | ROUGE: 0.389 -> 0.378 | Hurt |
Closest papers by meaning
- Modeling Context in Referring Expressions (ECCV 2016 · similarity 0.92)
- Revisiting Counterfactual Problems in Referring Expression Comprehension (CVPR 2024 · similarity 0.88)
- Ref-Adv: Exploring MLLM Visual Reasoning in Referring Expression Tasks (ICLR 2026 · similarity 0.86)
- RefDetector: A Simple Yet Effective Matching-based Method for Referring Expression Comprehension (AAAI 2025 · similarity 0.86)
- Rethinking Two-Stage Referring Expression Comprehension: A Novel Grounding and Segmentation Method Modulated by Point (AAAI 2024 · similarity 0.86)
- ScanFormer: Referring Expression Comprehension by Iteratively Scanning (CVPR 2024 · similarity 0.85)
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.
