Research map / Vision-language and multimodal / Human and embodied modeling
Context Disentangling and Prototype Inheriting for Robust Visual Grounding
Wei Tang, Liang Li, Xuejing Liu, Lu Jin, Jinhui Tang, Zechao Li
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 Hadamard fusion instead of Vision-Language Transformer fusion (ablation) | TransCP · ReferIt test (trained on RefCOCO) | top-1 accuracy: 24.24 -> 27.71 | Improved |
| use Hadamard fusion instead of Vision-Language Transformer fusion (ablation) | TransCP · RefCOCO val | top-1 accuracy: 83.02 -> 84.25 | Improved |
| add linguistic disentangling to baseline (ablation) | TransVG · RefCOCO val | top-1 accuracy: 80.48 -> 82.46 | Improved |
| add visual disentangling to baseline (ablation) | TransVG · RefCOCO val | top-1 accuracy: 80.48 -> 84.47 | Improved |
| add prototype module alone to baseline (ablation) | TransVG · RefCOCO val | top-1 accuracy: 80.48 -> 81.48 | No clear effect |
| add Hadamard fusion to baseline (ablation) | TransVG · RefCOCO val | top-1 accuracy: 80.48 -> 82.90 | Improved |
| add linguistic disentangling to Hadamard fusion (ablation) | TransCP · ReferIt test | top-1 accuracy: 24.19 -> 26.59 | Improved |
| add visual disentangling on top of H.D. and L.D. (ablation) | TransCP · ReferIt val | top-1 accuracy: 27.27 -> 28.53 | Mixed |
| add prototype discovering and inheriting to H.D.+L.D.+V.D. (ablation) | TransCP · ReferIt test | top-1 accuracy: 27.11 -> 27.71 | Improved |
| add prototype discovering and inheriting to H.D.+L.D.+V.D. (ablation) | TransCP · RefCOCO val | top-1 accuracy: 83.90 -> 84.25 | Improved |
| vary prototype bank size (too small or too large) (ablation) | TransCP · ReferIt open-vocabulary | top-1 accuracy | Mixed |
| number of regression Transformer encoder layers (ablation) | TransCP · RefCOCO and ReferIt | top-1 accuracy | Mixed |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- TransVG — TransVG: End-to-End Visual Grounding With Transformers (ICCV 2021) · better on 6 of 6 shared results
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
- GroundVLP: Harnessing Zero-Shot Visual Grounding from Vision-Language Pre-training and Open-Vocabulary Object Detection (AAAI 2024 · similarity 0.91)
- SimVG: A Simple Framework for Visual Grounding with Decoupled Multi-modal Fusion (NeurIPS 2024 · similarity 0.90)
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.89)
- GroundingME: Exposing the Visual Grounding Gap in MLLMs through Multi-Dimensional Evaluation (CVPR 2026 · similarity 0.89)
- Your Large Vision-Language Model Only Needs A Few Attention Heads For Visual Grounding (CVPR 2025 · similarity 0.89)
- AugRefer: Advancing 3D Visual Grounding via Cross-Modal Augmentation and Spatial Relation-based Referring (AAAI 2025 · 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.
