Research map / Vision-language and multimodal / Human and embodied modeling
Referring Image Segmentation via Cross-Modal Progressive Comprehension
Shaofei Huang, Tianrui Hui, Si Liu, Guanbin Li, Yunchao Wei, Jizhong Han, Luoqi Liu, Bo Li
Check 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 entity perception (EP) to the baseline (ablation) | CMPC-Refseg · UNC val | Overall IoU: 47.36 -> 49.06 | Improved |
| Add relation-aware reasoning (RAR) to the baseline (ablation) | CMPC-Refseg · UNC val | Overall IoU: 47.36 -> 53.40 | Improved |
| Add EP and RAR together (ablation) | CMPC-Refseg · UNC val | Overall IoU: 47.36 -> 55.38 | Improved |
| Initialize word embeddings with GloVe (ablation) | EP + RAR · UNC val | Overall IoU: 55.38 -> 56.00 | Improved |
| Add TGFE to a multi-level baseline (ablation) | CMPC-Refseg · UNC val | Overall IoU: 56.38 -> 58.81 | Improved |
| Add EP and RAR to the multi-level TGFE model (ablation) | CMPC-Refseg · UNC val | Overall IoU: 58.81 -> 60.72 | Improved |
| Increase TGFE exchange rounds from one to three (ablation) | CMPC-Refseg · UNC val | Overall IoU: 60.72 -> 61.25 | Improved |
| Use three graph convolution layers instead of one (ablation) | CMPC-Refseg · UNC val | Overall IoU: 55.38 -> 50.70 | Hurt |
| Use two graph convolution layers instead of one (ablation) | CMPC-Refseg · G-Ref val | Overall IoU: 38.19 -> 40.12 | Improved |
| Use three graph convolution layers instead of two (ablation) | CMPC-Refseg · G-Ref val | Overall IoU: 40.12 -> 38.96 | Hurt |
| Use CMPC-Refseg instead of STEP | Referring segmentation benchmarks · UNC val | Overall IoU: 60.04 -> 61.36 | Improved |
| Use CMPC-Refseg instead of STEP | Referring segmentation benchmarks · UNC testA | Overall IoU: 63.46 -> 64.53 | Improved |
Closest papers by meaning
- Cross-Modal Progressive Comprehension for Referring Segmentation (TPAMI 2021 · similarity 0.97)
- Beyond One-to-One: Rethinking the Referring Image Segmentation (ICCV 2023 · similarity 0.91)
- LQMFormer: Language-aware Query Mask Transformer for Referring Image Segmentation (CVPR 2024 · similarity 0.88)
- Mask Grounding for Referring Image Segmentation (CVPR 2024 · similarity 0.88)
- Boosting Weakly Supervised Referring Image Segmentation via Progressive Comprehension (NeurIPS 2024 · similarity 0.88)
- X-RefSeg3D: Enhancing Referring 3D Instance Segmentation via Structured Cross-Modal Graph Neural Networks (AAAI 2024 · 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.
