Research map / Detection and segmentation / CNN features
Ref-NMS: Breaking Proposal Bottlenecks in Two-Stage Referring Expression Grounding
Long Chen, Wenbo Ma, Jun Xiao, Hanwang Zhang, Shih-Fu Chang
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 |
|---|---|---|---|
| Add Ref-NMS with binary XE loss to CM-A-E | CM-A-E · RefCOCO testB | top-1 accuracy (%): 71.32 -> 76.04 | Improved |
| Add Ref-NMS with binary XE loss to CM-A-E | CM-A-E · RefCOCO+ testA | top-1 accuracy (%): 73.04 -> 73.68 | Improved |
| Add Ref-NMS with binary XE loss to CM-A-E | CM-A-E · RefCOCOg test | top-1 accuracy (%): 69.20 -> 70.62 | Improved |
| Add Ref-NMS with binary XE loss to CM-A-E | CM-A-E · RefCOCO testB | overall IoU (%): 53.14 -> 57.41 | Improved |
| Add Ref-NMS with binary XE loss to CM-A-E | CM-A-E · RefCOCO+ testA | overall IoU (%): 53.90 -> 53.84 | Hurt |
| Add Ref-NMS with binary XE loss to proposal selection | Mask R-CNN with GreedyNMS · RefCOCO testB | referent recall (%): 80.77 -> 88.28 | Improved |
| Use ranking loss instead of binary XE loss (ablation) | Ref-NMS · RefCOCO testB | top-1 accuracy (%): 76.04 -> 73.62 | Hurt |
| Remove pseudo ground truths for contextual objects (ablation) | CM-A-E+Ref-NMS · RefCOCO testA | top-1 accuracy (%): 83.63 -> 84.00 | Hurt |
| Remove pseudo ground truths for contextual objects (ablation) | CM-A-E+Ref-NMS · RefCOCO testB | top-1 accuracy (%): 76.09 -> 76.04 | No clear effect |
Closest papers by meaning
- Entity-enhanced Adaptive Reconstruction Network for Weakly Supervised Referring Expression Grounding (TPAMI 2022 · similarity 0.90)
- Rethinking Two-Stage Referring Expression Comprehension: A Novel Grounding and Segmentation Method Modulated by Point (AAAI 2024 · similarity 0.89)
- Hierarchical Alignment-enhanced Adaptive Grounding Network for Generalized Referring Expression Comprehension (AAAI 2025 · similarity 0.88)
- RefDetector: A Simple Yet Effective Matching-based Method for Referring Expression Comprehension (AAAI 2025 · similarity 0.88)
- Unveiling Parts Beyond Objects: Towards Finer-Granularity Referring Expression Segmentation (CVPR 2024 · similarity 0.87)
- Ref-Adv: Exploring MLLM Visual Reasoning in Referring Expression Tasks (ICLR 2026 · similarity 0.86)
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.
