Research map / Vision-language and multimodal / CLIP and zero-shot
CRIS: CLIP-Driven Referring Image Segmentation
Check 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 |
|---|---|---|---|
| Add text-to-pixel contrastive learning only (ablation) | CRIS baseline · RefCOCO val | IoU: 62.66 -> 64.64 | Helped |
| Add text-to-pixel contrastive learning only (ablation) | CRIS baseline · RefCOCO+ val | IoU: 50.17 -> 53.15 | Helped |
| Add text-to-pixel contrastive learning only (ablation) | CRIS baseline · G-Ref val (UNC) | IoU: 49.24 -> 52.67 | Helped |
| Add one vision-language decoder layer only (ablation) | CRIS baseline · RefCOCO val | IoU: 62.66 -> 66.31 | Helped |
| Add one vision-language decoder layer only (ablation) | CRIS baseline · RefCOCO+ val | IoU: 50.17 -> 54.73 | Helped |
| Add one vision-language decoder layer only (ablation) | CRIS baseline · G-Ref val (UNC) | IoU: 49.24 -> 51.46 | Helped |
| Combine contrastive learning and decoder, using three decoder layers (ablation) | CRIS baseline · RefCOCO val | IoU: 62.66 -> 69.52 | Helped |
| Combine contrastive learning and decoder, using three decoder layers (ablation) | CRIS baseline · RefCOCO+ val | IoU: 50.17 -> 61.39 | Helped |
| Combine contrastive learning and decoder, using three decoder layers (ablation) | CRIS baseline · G-Ref val (UNC) | IoU: 49.24 -> 59.35 | Helped |
| Increase decoder depth from three to four layers (ablation) | CRIS · RefCOCO val | IoU: 69.52 -> 69.18 | Hurt |
| Increase decoder depth from three to four layers (ablation) | CRIS · RefCOCO+ val | IoU: 61.39 -> 61.15 | Hurt |
| Increase decoder depth from three to four layers (ablation) | CRIS · G-Ref val (UNC) | IoU: 59.35 -> 58.79 | Hurt |
Closest papers by meaning
- Prompt-Driven Referring Image Segmentation with Instance Contrasting (CVPR 2024 · similarity 0.92)
- Boosting Weakly Supervised Referring Image Segmentation via Progressive Comprehension (NeurIPS 2024 · similarity 0.90)
- Curriculum Point Prompting for Weakly-Supervised Referring Image Segmentation (CVPR 2024 · similarity 0.89)
- Mask Grounding for Referring Image Segmentation (CVPR 2024 · similarity 0.88)
- IteRPrimE: Zero-shot Referring Image Segmentation with Iterative Grad-CAM Refinement and Primary Word Emphasis (AAAI 2025 · similarity 0.88)
- LQMFormer: Language-aware Query Mask Transformer for Referring Image Segmentation (CVPR 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.