Research map / Vision-language and multimodal / Human and embodied modeling
Deconfounded Visual Grounding
Jianqiang Huang, Yu Qin, Jiaxin Qi, Qianru Sun, Hanwang Zhang
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 RED to LBYL | LBYL · RefCOCO testB | Acc@0.5%: 74.15 -> 77.66 | Improved |
| Add RED to LBYL | LBYL · RefCOCO+ testB | Acc@0.5%: 59.49 -> 62.20 | Improved |
| Add RED to LBYL | LBYL · RefCOCOg | Acc@0.5%: 69.06 -> 71.11; 68.34 -> 70.67 | Improved |
| Add RED to LBYL | LBYL · ReferItGame | Acc@0.5%: 66.51 -> 67.27 | Improved |
| Add RED to Yang's-V1† | Yang's-V1† (YOLO-V4) · Flickr30K Entities | Acc@0.5%: 70.25 -> 70.50 | Improved |
| Use an autoencoder on image and language features rather than language features alone (ablation) | Yang's-V1† + RED · RefCOCO+ and RefCOCOg | Acc@0.5%: 66.74 vs 66.87; 57.23 vs 57.51; 69.01 vs 69.46; 68.55 vs 69.51 | Hurt |
| Add attention to image features alongside language features (ablation) | Yang's-V1† + RED · RefCOCO+ and RefCOCOg | Acc@0.5%: 65.57 vs 66.87; 56.37 vs 57.51; 67.83 vs 69.46; 68.11 vs 69.51 | Hurt |
| Add RED to Yang's-V1† | Yang's-V1† · Visual grounding inference | latency per sample: 24.56ms -> 26.53ms | Hurt |
Closest papers by meaning
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.89)
- Look Around Before Locating: Considering Content and Structure Information for Visual Grounding (AAAI 2025 · similarity 0.88)
- Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning (AAAI 2026 · similarity 0.88)
- GroundingME: Exposing the Visual Grounding Gap in MLLMs through Multi-Dimensional Evaluation (CVPR 2026 · similarity 0.88)
- Multi-task Visual Grounding with Coarse-to-Fine Consistency Constraints (AAAI 2025 · similarity 0.87)
- GeM-VG: Towards Generalized Multi-image Visual Grounding with Multimodal Large Language Models (AAAI 2026 · 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.
