Research map / Detection and segmentation / Tracking
PropVG: End-to-End Proposal-Driven Visual Grounding with Multi-Granularity Discrimination
Ming Dai, Wenxuan Cheng, Jiedong Zhuang, Jiang-jiang Liu, Hongshen Zhao, Zhenhua Feng, Wankou Yang
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 CRS module (ablation) | PropVG baseline · gRefCOCO val (ablation, 224px) | gIoU: 69.85 -> 71.30 | Improved |
| add MTD module (ablation) | PropVG baseline · gRefCOCO val (ablation, 224px) | gIoU: 69.85 -> 71.26 | Improved |
| add both CRS and MTD (ablation) | PropVG baseline · gRefCOCO val (ablation, 224px) | gIoU: 69.85 -> 71.99 | Improved |
| add foreground supervision in proposal stage (ablation) | PropVG baseline · gRefCOCO val (ablation, 224px) | gIoU: 66.37 -> 69.85 | Improved |
| add foreground supervision in proposal stage (ablation) | PropVG baseline · gRefCOCO val (ablation, 224px) | N-acc.: 61.06 -> 70.39 | Improved |
| add sentence-level contrastive scoring instead of linear layer (ablation) | CRS · gRefCOCO val (ablation, 224px) | N-acc.: 70.39 -> 73.39 | Improved |
| add word-level contrastive scoring instead of linear layer (ablation) | CRS · gRefCOCO val (ablation, 224px) | N-acc.: 70.39 -> 73.88 | Improved |
| use adaptive learned weighting instead of plain averaging of sentence/word similarities (ablation) | CRS · gRefCOCO val (ablation, 224px) | N-acc.: 74.11 -> 74.78 | Improved |
| add mask-level SPCA (ablation) | MTD · gRefCOCO val (ablation, 224px) | gIoU: 69.85 -> 70.87 | Improved |
| add query-level SPCA (ablation) | MTD · gRefCOCO val (ablation, 224px) | cIoU: 66.24 -> 66.13 | No clear effect |
| add SimFPN, UNet decoder, deformable decoder, channel splitting, query projection (ablation) | PropVG basic setting · gRefCOCO val (ablation, 224px) | gIoU: 65.98 -> 69.85 | Improved |
| add existence loss weight 0.2 (vs 0.0) (ablation) | PropVG · gRefCOCO val (loss weight ablation) | N-acc: 67.8 -> 71.0 | Improved |
Closest papers by meaning
- GroundVLP: Harnessing Zero-Shot Visual Grounding from Vision-Language Pre-training and Open-Vocabulary Object Detection (AAAI 2024 · similarity 0.88)
- RefDetector: A Simple Yet Effective Matching-based Method for Referring Expression Comprehension (AAAI 2025 · similarity 0.88)
- Interpreting Object-level Foundation Models via Visual Precision Search (CVPR 2025 · similarity 0.88)
- Referencing Where to Focus: Improving Visual Grounding with Referential Query (NeurIPS 2024 · similarity 0.88)
- SimVG: A Simple Framework for Visual Grounding with Decoupled Multi-modal Fusion (NeurIPS 2024 · similarity 0.88)
- Rex-Thinker: Grounded Object Referring via Chain-of-Thought Reasoning (ICLR 2026 · 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.
