Research map / Vision-language and multimodal / Human and embodied modeling
Region-aware Anchoring Mechanism for Efficient Referring Visual Grounding
Shuyi Ouyang, Ziwei Niu, Hongyi Wang, Yen-Wei Chen, Lanfen Lin
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 |
|---|---|---|---|
| replace direct interaction (LAVT design) with RaAM (ablation) | RaAM-RVG · gRefCOCO val (GRES) | cIoU: 64.77 -> 67.35 | Improved |
| replace direct interaction (LAVT design) with RaAM (ablation) | RaAM-RVG · gRefCOCO val (GRES) | gIoU: 67.62 -> 70.02 | Improved |
| add L2V gate (ablation) | RaAM-RVG · gRefCOCO val (GRES) | cIoU: 65.74 -> 67.35 | Improved |
| add L2V gate (ablation) | RaAM-RVG · gRefCOCO val (GRES) | Pr@0.9: 25.86 -> 27.43 | Improved |
| add learnable coefficients and linear transform flow control components (ablation) | RaAM-RVG · gRefCOCO val (GRES) | cIoU: 63.87 -> 67.35 | Improved |
| generate anchors in 2 stages instead of 1 (ablation) | RaAM-RVG · gRefCOCO val (GRES) | cIoU: 66.38 -> 67.35 | Improved |
| generate anchors in 3 stages instead of 2 (ablation) | RaAM-RVG · gRefCOCO val (GRES) | cIoU: 67.35 -> 64.58 | Hurt |
| generate anchors in 4 stages instead of 2 (ablation) | RaAM-RVG · gRefCOCO val (GRES) | cIoU: 67.35 -> 60.22 | Hurt |
| add region loss Lr and pixel loss Lseg_p (ablation) | RaAM-RVG · gRefCOCO val (GRES) | cIoU: 65.09 -> 67.35 | Improved |
| integrate RaAM into LAVT | LAVT · gRefCOCO val (GRES) | cIoU: 57.64 -> 61.45 | Improved |
| integrate RaAM into PolyFormer | PolyFormer · gRefCOCO val (GRES) | cIoU: 59.35 -> 63.79 | Improved |
| integrate RaAM into HyperSeg | HyperSeg · gRefCOCO val (GRES) | cIoU: 62.75 -> 66.68 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 3 of 3 on most shared results.
- UniTAB — UniTAB: Unifying Text and Box Outputs for Grounded Vision-Language Modeling (ECCV 2022) · better on 3 of 3 shared results
- HyperSeg — HyperSeg: Hybrid Segmentation Assistant with Fine-grained Visual Perceiver (CVPR 2025) · better on 1 of 1 shared result
- LAVT — LAVT: Language-Aware Vision Transformer for Referring Image Segmentation (CVPR 2022) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Your Large Vision-Language Model Only Needs A Few Attention Heads For Visual Grounding (CVPR 2025 · similarity 0.90)
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.89)
- Multi-task Visual Grounding with Coarse-to-Fine Consistency Constraints (AAAI 2025 · similarity 0.89)
- Connecting the Dots: Training-Free Visual Grounding via Agentic Reasoning (AAAI 2026 · similarity 0.89)
- GroundingME: Exposing the Visual Grounding Gap in MLLMs through Multi-Dimensional Evaluation (CVPR 2026 · similarity 0.88)
- GeM-VG: Towards Generalized Multi-image Visual Grounding with Multimodal Large Language Models (AAAI 2026 · 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.
