Research map / Detection and segmentation / CNN features
DPText-DETR: Towards Better Scene Text Detection with Dynamic Points in Transformer
Read the paperCheck 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 |
|---|---|---|---|
| Replace box-based positional queries with explicit sampled points and update points across decoder layers (ablation) | DPText-DETR · Total-Text | F-measure: 83.90 -> 85.15 | Helped |
| Add positional label form (ablation) | DPText-DETR · Rot.Total-Text | F-measure: 71.37 -> 74.11 | Helped |
| Add positional label form (ablation) | DPText-DETR · Inverse-Text | F-measure: 78.71 -> 81.20 | Helped |
| Add positional label form (ablation) | DPText-DETR · Total-Text | F-measure: 84.71 -> 84.77 | Helped |
| Add EPQM to positional-label model (ablation) | DPText-DETR · Total-Text | F-measure: 84.77 -> 85.80 | Helped |
| Add EPQM to positional-label model (ablation) | DPText-DETR · Rot.Total-Text | F-measure: 74.11 -> 73.82 | Hurt |
| Add EPQM to positional-label model (ablation) | DPText-DETR · Inverse-Text | F-measure: 81.20 -> 80.86 | Hurt |
| Add EFSA (ablation) | DPText-DETR · Total-Text | F-measure: 85.28 -> 86.17 | Helped |
| Add EFSA (ablation) | DPText-DETR · Total-Text with 25% training data | F-measure: 58.54 -> 70.49 | Helped |
| Add EFSA (ablation) | DPText-DETR · Inverse-Text with 25% training data | F-measure: 52.32 -> 60.15 | Helped |
| Use rotation augmentation (ablation) | DPText-DETR · Rot.Total-Text | F-measure: 70.02 -> 83.99 | Helped |
| Use EFSA with circular convolution for Bezier control points (ablation) | Bezier variant detector · Total-Text | F-measure: 85.80 -> 85.21 | Hurt |
Closest papers by meaning
- DeepSolo: Let Transformer Decoder With Explicit Points Solo for Text Spotting (CVPR 2023 · similarity 0.87)
- Few Could Be Better Than All: Feature Sampling and Grouping for Scene Text Detection (CVPR 2022 · similarity 0.87)
- SAP-DETR: Bridging the Gap Between Salient Points and Queries-Based Transformer Detector for Fast Model Convergency (CVPR 2023 · similarity 0.86)
- LRANet: Towards Accurate and Efficient Scene Text Detection with Low-Rank Approximation Network (AAAI 2024 · similarity 0.86)
- Arbitrary Reading Order Scene Text Spotter with Local Semantics Guidance (AAAI 2025 · similarity 0.85)
- ViTEraser: Harnessing the Power of Vision Transformers for Scene Text Removal with SegMIM Pretraining (AAAI 2024 · similarity 0.84)
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.