Research map / Detection and segmentation / Open-set and few-shot detection
Key-Word-Aware Network for Referring Expression Image Segmentation
Hengcan Shi, Hongliang Li, Fanman Meng, Qingbo Wu
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 query attention to extract key words per image region (ablation) | Baseline [9] (VGG16) · ReferItGame test (ablation) | IoU: 48.03% -> 50.28% | Improved |
| add vision-only spatial pyramid visual context (ablation) | Key-word-model · ReferItGame test (ablation) | IoU: 50.28% -> 51.01% | Improved |
| make visual context key-word-aware (ablation) | Context-model · ReferItGame test (ablation) | IoU: 51.01% -> 52.19% | Improved |
| replace unit step with soft attention to aggregate context (ablation) | Full-model · ReferItGame test (ablation) | IoU: 52.19% -> 51.93% | No clear effect |
| use KWAN instead of [18] with VGG16 | [18] · ReferItGame test | IoU: 48.84% -> 52.19% | Improved |
| use KWAN instead of [18] with Deeplab101 | [18] · ReferItGame test | IoU: 57.34% -> 59.09% | Improved |
| use KWAN instead of [18] with VGG16 | [18] · Google-Ref val | IoU: 28.60% -> 31.36% | Improved |
| use KWAN instead of [18] with Deeplab101 | [18] · Google-Ref val | IoU: 34.40% -> 36.92% | Improved |
| use KWAN instead of [18] at high precision threshold | [18] Deeplab101 · ReferItGame test | Pr@0.9: 6.43% -> 11.79% | Improved |
| use KWAN instead of [18] on long queries | [18] VGG16 · ReferItGame test query length 7-20 | IoU: 22.25% -> 26.61% | Improved |
| use KWAN instead of [18] on single-word queries | [18] VGG16 · ReferItGame test query length 1 | IoU: 63.19% -> 65.59% | Improved |
| use KWAN instead of [18] with VGG16 | [18] · ReferItGame test | Pr@0.5: 35.79% -> 35.61% | No clear effect |
Closest papers by meaning
- Locate Then Segment: A Strong Pipeline for Referring Image Segmentation (CVPR 2021 · similarity 0.90)
- Vision-Language Transformer and Query Generation for Referring Segmentation (ICCV 2021 · similarity 0.89)
- Beyond One-to-One: Rethinking the Referring Image Segmentation (ICCV 2023 · similarity 0.89)
- LQMFormer: Language-aware Query Mask Transformer for Referring Image Segmentation (CVPR 2024 · similarity 0.87)
- Mask Grounding for Referring Image Segmentation (CVPR 2024 · similarity 0.87)
- Unveiling Parts Beyond Objects: Towards Finer-Granularity Referring Expression Segmentation (CVPR 2024 · 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.
