Research map / Vision-language and multimodal / Human and embodied modeling
Improving Referring Expression Grounding With Cross-Modal Attention-Guided Erasing
Check 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 |
|---|---|---|---|
| add cross-modal attention-guided erasing during training | CM-Att · RefCOCOg | accuracy: 78.10 -> 80.23; 78.12 -> 80.37 | Helped |
| replace attention-guided erasing with random erasing (ablation) | CM-Att-Erase · RefCOCOg | accuracy: 80.23 -> 79.08; 80.37 -> 79.05 | Hurt |
| use a separate adversarial network to choose erasures (ablation) | CM-Att-Erase · RefCOCOg | accuracy: 80.23 -> 79.31; 80.37 -> 79.23 | Hurt |
| use self-attention rather than cross-modal attention to guide erasing (ablation) | CM-Att-Erase · RefCOCOg | accuracy: 80.23 -> 79.27; 80.37 -> 79.22 | Hurt |
| remove visual erasing, retaining textual erasing (ablation) | CM-Att-Erase · RefCOCOg | accuracy: 80.23 -> 79.21; 80.37 -> 79.55 | Hurt |
| remove textual erasing, retaining visual erasing (ablation) | CM-Att-Erase · RefCOCOg | accuracy: 80.23 -> 79.05; 80.37 -> 79.37 | Hurt |
| apply erasing iteratively (ablation) | CM-Att-Erase · RefCOCOg | accuracy: 80.23 -> 80.13; 80.37 -> 79.97 | Hurt |
| also erase during inference and ensemble scores (ablation) | CM-Att-Erase · RefCOCOg | accuracy: 80.23 -> 79.25; 80.37 -> 79.56 | Hurt |
| replace erasing with multiple steps of attention (ablation) | CM-Att-Erase · RefCOCOg | accuracy: 80.23 -> 79.31; 80.37 -> 78.49 | Hurt |
Closest papers by meaning
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.89)
- RefDetector: A Simple Yet Effective Matching-based Method for Referring Expression Comprehension (AAAI 2025 · similarity 0.88)
- Task-aware Cross-modal Feature Refinement Transformer with Large Language Models for Visual Grounding (CVPR 2025 · similarity 0.88)
- Hierarchical Alignment-enhanced Adaptive Grounding Network for Generalized Referring Expression Comprehension (AAAI 2025 · similarity 0.88)
- 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 0.88)
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.