Research map / Vision-language and multimodal / Human and embodied modeling
Grounding Referring Expressions in Images by Variational Context
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's title only (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 variational context modeling | referring-expression grounding · RefCOCO Test B | accuracy (Acc%): 82.39 vs. 80.22 | Helped |
| use variational context modeling | referring-expression grounding · RefCOCO+ Test A | accuracy (Acc%): 64.60 vs. 62.56 | Hurt |
| use variational context modeling | referring-expression grounding · RefCOCOg Val | accuracy (Acc%): 73.98 vs. 72.63 | Helped |
| add KL context regularization (ablation) | VC supervised grounding · RefCOCO, RefCOCO+, and RefCOCOg | accuracy (Acc%): VC outperforms VC w/o reg by around 2% | Helped |
| add cue-specific word attention (ablation) | VC supervised grounding · RefCOCO, RefCOCO+, and RefCOCOg | accuracy (Acc%) | Helped |
| remove cue-specific word attention (ablation) | VC unsupervised grounding · RefCOCO, RefCOCO+, and their detected-box settings | accuracy (Acc%): 13.59 -> 33.29; 21.65 -> 30.13; 18.79 -> 34.60; 24.14 -> 31.58 | Helped |
| add cue-specific word attention (ablation) | VC unsupervised grounding · RefCOCOg Val | accuracy (Acc%): 30.26 -> 33.79 | Helped |
| add KL context regularization (ablation) | VC unsupervised grounding · RefCOCOg Val | accuracy (Acc%): 25.14 -> 33.79 | Helped |
| use the jointly trained parser for language features (ablation) | VC unsupervised grounding · RefCOCOg | accuracy (%): 30.73 vs. 33.79 | Hurt |
Closest papers by meaning
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.90)
- RefDetector: A Simple Yet Effective Matching-based Method for Referring Expression Comprehension (AAAI 2025 · similarity 0.89)
- Mask Grounding for Referring Image Segmentation (CVPR 2024 · similarity 0.89)
- Unveiling Parts Beyond Objects: Towards Finer-Granularity Referring Expression Segmentation (CVPR 2024 · similarity 0.89)
- Rex-Thinker: Grounded Object Referring via Chain-of-Thought Reasoning (ICLR 2026 · similarity 0.89)
- Zero-shot Referring Expression Comprehension via Structural Similarity Between Images and Captions (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.