Research map / Detection and segmentation / Open-set and few-shot detection
ST-SAM: Multimodal Scene Text Segmentation with Dense Visual and Sparse Textual Prompts via SAM
Jin Wei, Yaqiang Wu, Jiayi Yan, Zeng Li, Zhen Xu, Yu Zhou, Lingling Zhang, QianYing Wang
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 adapters to SAM ViT encoder (ablation) | SAM-B · ICDAR13 FST | fgIoU: 65.70 -> 68.18 | Improved |
| add dense visual prompts (ablation) | ST-SAM-B (adapter only) · ICDAR13 FST | fgIoU: 68.18 -> 69.52 | Improved |
| add dense visual prompts (ablation) | ST-SAM-B with text prompts · ICDAR13 FST | fgIoU: 69.48 -> 71.18 | Improved |
| generate text prompts with GPT-4 instead of plain 'Text' (ablation) | ST-SAM-B · ICDAR13 FST | fgIoU: 70.10 -> 71.18 | Improved |
| replace concatenation of text prompts with cross-modal alignment (ablation) | ST-SAM-B · ICDAR13 FST | fgIoU: 68.49 -> 71.18 | Improved |
| add concatenation of text prompts instead of alignment (ablation) | ST-SAM-B visual prompt model · ICDAR13 FST | fgIoU: 69.52 -> 68.49 | Hurt |
| add learnable scenario-specific prompt to LLM prompts (ablation) | ST-SAM-B · ICDAR13 FST | fgIoU: 71.18 -> 72.62 | Improved |
| add LLM-generated prompts to scenario-specific prompt (ablation) | ST-SAM-B · ICDAR13 FST | fgIoU: 70.99 -> 72.62 | Improved |
| use ST-SAM-B instead of Hi-SAM-B | SAM ViT-B · HierText | fgIoU: 73.39 -> 75.69 | Improved |
| ST-SAM-B vs Hi-SAM-B trained on TextSeg, tested on Total-Text | SAM ViT-B · TextSeg to Total-Text | F-score: 82.27 -> 86.56 | Improved |
| use ST-SAM-B instead of UPOCR on difficult cases | SAM ViT-B · Total-Text difficult subset | fgIoU: 76.34 -> 80.67 | Improved |
| ST-SAM-B vs EAFormer on Total-Text | ST-SAM-B · Total-Text | fgIoU: 82.73 -> 82.36 | No clear effect |
Earlier methods this paper compares with
Named baselines from earlier accepted papers in this paper's results tables: better than 2 of 2 on most shared results.
- UPOCR — UPOCR: Towards Unified Pixel-Level OCR Interface (ICML 2024) · better on 3 of 3 shared results
- EAFormer — EAFormer: Scene Text Segmentation with Edge-Aware Transformers (ECCV 2024) · better on 3 of 3 shared results
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
- SAM-CP: Marrying SAM with Composable Prompts for Versatile Segmentation (ICLR 2025 · similarity 0.88)
- Instruction-guided Multi-Granularity Segmentation and Captioning with Large Multimodal Model (AAAI 2025 · similarity 0.87)
- OpenWorldSAM: Extending SAM2 for Universal Image Segmentation with Language Prompts (NeurIPS 2025 · similarity 0.86)
- MSTAR: Box-free Multi-query Scene Text Retrieval with Attention Recycling (NeurIPS 2025 · similarity 0.86)
- ViTEraser: Harnessing the Power of Vision Transformers for Scene Text Removal with SegMIM Pretraining (AAAI 2024 · similarity 0.86)
- Segment and Caption Anything (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.
