| What Is Wrong with Synthetic Data for Scene Text Recognition? A Strong Synthetic Engine with Diverse Simulations and Self-Evolution | ✓ Link | 95.42 | | | 86.67 | 91.09 | 87.74 | | SVTRv2-AR + UnionST-SP + R | 2026-02-06 |
| Advancing WordArt-Oriented Scene Text Recognition: Datasets and Methods | ✓ Link | 94.27 | 86.91 | 89.32 | 84.11 | 89.89 | 85.6 | 95.54 | WATERec-RS (extra data) | 2026-06-23 |
| Embedding Rotation Invariance for Provable Multi-Oriented Scene Text Recognition | | 94.1 | 80.6 | 87.4 | 79.4 | 90.3 | 83.7 | 96.6 | RISTER-L (extra data) | 2026-08-11 |
| SVTRv2: CTC Beats Encoder-Decoder Models in Scene Text Recognition | ✓ Link | 90.6 | 86.1 | 86.7 | 79.3 | 86.2 | 85.1 | 89.0 | SVTRv2-B | 2024-11-24 |
| Linguistics-aware Masked Image Modeling for Self-supervised Scene Text Recognition | | 90.6 | | | | | 85.7 | 86.9 | LMIM (20-epoch, ARD fine-tuning) (extra data) | 2025-03-24 |
| Revisiting Scene Text Recognition: A Data Perspective | ✓ Link | 88.8 | 85.5 | 87.5 | 80.0 | 84.9 | 85.8 | 83.9 | MAERec-B (Union14M-U PT + Union14M-L FT) | 2023-07-17 |
| Focus on the Whole Character: Discriminative Character Modeling for Scene Text Recognition | ✓ Link | 86.8 | 85.5 | 85.9 | 77.5 | 83.5 | 84.4 | 80.4 | CFE-Base (Union14M-L training) (extra data) | 2024-07-08 |
| OTE: Exploring Accurate Scene Text Recognition Using One Token | ✓ Link | 83.1 | 73.7 | 70.3 | 73.5 | 79.7 | 82.2 | 82.8 | OTE A/SVTR | 2024-06-17 |
| Context Perception Parallel Decoder for Scene Text Recognition | ✓ Link | | | 65.5 | | | | | NRTR-CPPD | 2023-07-23 |
| Instruction-Guided Scene Text Recognition | ✓ Link | | 83.6 | | | | | | IGTR-PR-PT (extra data) | 2024-01-31 |
| Linguistics-aware Masked Image Modeling for Self-supervised Scene Text Recognition | | | 85.5 | 87.9 | 80.7 | 88.2 | | | LMIM (20-epoch, Union14M-L fine-tuning) (extra data) | 2025-03-24 |
| What Is Wrong with Synthetic Data for Scene Text Recognition? A Strong Synthetic Engine with Diverse Simulations and Self-Evolution | ✓ Link | | | | | | | 97.3 | SVTRv2-AR + UnionST-PR + R | 2026-02-06 |
| What Is Wrong with Synthetic Data for Scene Text Recognition? A Strong Synthetic Engine with Diverse Simulations and Self-Evolution | ✓ Link | | 91.14 | | | | | | SVTRv2-AR + UnionST-SP + ISR (D_P2 + D_L-hard) | 2026-02-06 |
| What Is Wrong with Synthetic Data for Scene Text Recognition? A Strong Synthetic Engine with Diverse Simulations and Self-Evolution | ✓ Link | | | 92.48 | | | | | SVTRv2-AR + UnionST-SR + R | 2026-02-06 |