Research map / Vision-language and multimodal / Human and embodied modeling
MAttNet: Modular Attention Network for Referring Expression Comprehension
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 modular subject and location matching instead of concatenated visual and language features (ablation) | MAttNet · RefCOCO val | accuracy: 79.14 -> 79.68 | Helped |
| Add relative-location features (ablation) | MAttNet with subject and location modules · RefCOCO val | accuracy: 79.68 -> 82.06 | Helped |
| Add the relationship module (ablation) | MAttNet with relative-location features · RefCOCO val | accuracy: 82.06 -> 82.54 | Helped |
| Add the attribute prediction branch (ablation) | MAttNet with relationship and location modules · RefCOCO val | accuracy: 82.54 -> 83.54 | Helped |
| Add phrase-guided in-box subject attention (ablation) | MAttNet with attribute prediction · RefCOCO testA | accuracy: 82.66 -> 85.26 | Helped |
| Replace learned soft phrase attention with a template parser (ablation) | MAttNet · RefCOCO val | accuracy: 85.65 -> 80.20 | Hurt |
| Use ResNet-101 Faster R-CNN features instead of VGG-16 features (ablation) | MAttNet without phrase-guided subject attention · RefCOCO val | accuracy: 80.94 -> 83.54 | Helped |
| Use Mask R-CNN features instead of Faster R-CNN features (ablation) | Fully automatic MAttNet · RefCOCO val | accuracy: 76.40 -> 76.65 | Helped |
| Use MAttNet for expression-guided segmentation instead of D+RMI+DCRF | RefCOCO segmentation · RefCOCO val | IoU: 45.18 -> 56.51 | Helped |
| Use MAttNet for expression-guided segmentation instead of D+RMI+DCRF | RefCOCO+ segmentation · RefCOCO+ val | IoU: 29.86 -> 46.67 | Helped |
| Separate box localization from segmentation and use GrabCut for the mask | MAttNet referential segmentation · RefCOCO+ val | IoU: 29.86 -> 37.18 | Helped |
| Separate box localization from segmentation and use GrabCut for the mask | MAttNet referential segmentation · RefCOCO val | IoU: 45.18 -> 42.86 | Hurt |
Closest papers by meaning
- Rethinking Two-Stage Referring Expression Comprehension: A Novel Grounding and Segmentation Method Modulated by Point (AAAI 2024 · similarity 0.89)
- ScanFormer: Referring Expression Comprehension by Iteratively Scanning (CVPR 2024 · similarity 0.88)
- Ref-Adv: Exploring MLLM Visual Reasoning in Referring Expression Tasks (ICLR 2026 · similarity 0.88)
- DViN: Dynamic Visual Routing Network for Weakly Supervised Referring Expression Comprehension (CVPR 2025 · similarity 0.87)
- Boosting Weakly Supervised Referring Image Segmentation via Progressive Comprehension (NeurIPS 2024 · similarity 0.87)
- RefDetector: A Simple Yet Effective Matching-based Method for Referring Expression Comprehension (AAAI 2025 · 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.