Research map / Vision-language and multimodal / Human and embodied modeling
Cross-Modal Progressive Comprehension for Referring Segmentation
Si Liu, Tianrui Hui, Shaofei Huang, Yunchao Wei, Bo Li, Guanbin Li
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 entity perception stage (ablation) | baseline · UNC val | overall IoU: 47.36 -> 49.06 | Improved |
| add relation-aware reasoning stage (ablation) | baseline · UNC val | overall IoU: 47.36 -> 53.40 | Improved |
| add EP and RAR together (ablation) | baseline · UNC val | overall IoU: 47.36 -> 55.38 | Improved |
| add GloVe initialization (ablation) | CMPC-I (EP+RAR) · UNC val | overall IoU: 55.38 -> 56.00 | Improved |
| add CMF (concatenate multimodal feature) (ablation) | CMPC-I with GloVe · UNC val | overall IoU: 56.00 -> 57.12 | Improved |
| add TGFE to multi-level baseline (ablation) | multi-level baseline · UNC val | overall IoU: 56.38 -> 58.81 | Improved |
| increase TGFE exchange rounds from 1 to 3 (ablation) | CMPC-I + TGFE · UNC val | overall IoU: 60.72 -> 61.25 | Improved |
| add necessary words extraction (ablation) | TGFE full model · UNC val | overall IoU: 61.23 -> 61.80 | Improved |
| add action-aware reasoning (AAR) (ablation) | video model with TGFE, EP, RAR · A2D Sentences | Overall IoU: 63.2 -> 64.9 | Improved |
| use direct relevance instead of adaptive relevance in AAR (ablation) | CMPC-V · A2D Sentences | Overall IoU: 64.5 -> 64.9 | Improved |
| use 1 instead of 0 graph convolution layers (ablation) | CMPC-I · UNC val | overall IoU: 49.06 -> 55.38 | Improved |
| use 2 instead of 1 graph convolution layers (ablation) | CMPC-I · G-Ref val | overall IoU: 38.19 -> 40.12 | Improved |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 1 of 1 on most shared results.
- MAttNet — MAttNet: Modular Attention Network for Referring Expression Comprehension (CVPR 2018) · better on 1 of 1 shared result
Matched by method name in results tables read by AI. Authors' own tables tend to favour the newer method, and a comparison can be missed when a method is renamed.
Closest papers by meaning
- Referring Image Segmentation via Cross-Modal Progressive Comprehension (CVPR 2020 · similarity 0.97)
- Referring Segmentation in Images and Videos with Cross-Modal Self-Attention Network (TPAMI 2021 · similarity 0.92)
- Language-Aware Spatial-Temporal Collaboration for Referring Video Segmentation (TPAMI 2023 · similarity 0.91)
- Semantic-Assisted Object Clustering for Multi-Modal Referring Video Segmentation (TPAMI 2025 · similarity 0.90)
- Referring Segmentation via Encoder-Fused Cross-Modal Attention Network (TPAMI 2022 · similarity 0.90)
- Language-Aware Vision Transformer for Referring Segmentation (TPAMI 2024 · similarity 0.90)
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.
