Codesota · Computer Vision · Optical Character Recognition · icdar2015Tasks/Computer Vision/Optical Character Recognition
Optical Character Recognition · benchmark dataset · 2020 · EN

icdar2015.

Dataset from Papers With Code

Submit a result ↵
§ 01 · Leaderboard

Best published scores.

30 results indexed across 2 metrics. Shaded row marks current SOTA; ties broken by submission date.

Primary
accuracy · higher is better
All metrics
accuracy, f-measure
accuracy· primary
29 rows
#ModelOrgSubmittedPaper / codeaccuracy
01JSTROSSFujitakeApr 2024JSTR: Judgment Improves Scene Text Recognition98.70
02TextBlockV2 (GPT-2)OSSJiahao Lyu et al., Fudan UniversityMar 2024TextBlockV2: Towards Precise-Detection-Free Scene Text S…97.70
03DTrOCR 105M—Aug 2023DTrOCR: Decoder-only Transformer for Optical Character R… · code93.50
04CPPD—Jul 2023Context Perception Parallel Decoder for Scene Text Recog… · code91.70
05CLIP4STR-L (DataComp-1B)—May 2023CLIP4STR: A Simple Baseline for Scene Text Recognition w… · code91.40
06MGP-STR—Sep 2022Multi-Granularity Prediction for Scene Text Recognition · code90.90
07CLIP4STR-L—May 2023CLIP4STR: A Simple Baseline for Scene Text Recognition w… · code90.80
08CLIP4STR-BResearchMay 2023CLIP4STR: A Simple Baseline for Scene Text Recognition w… · code90.60
09OTSNetOSSAnonymous / arxiv preprintNov 2025OTSNet: A Unified Observation-Thinking-Spelling Network …90.20
10IGTR-AROSSYongkun Du et al.Jan 2024Instruction-Guided Scene Text Recognition89.80
11SIGA_S—Mar 2022Self-supervised Implicit Glyph Attention for Text Recogn… · code87.60
12S-GTR—Dec 2021Visual Semantics Allow for Textual Reasoning Better in S… · code87.30
13MATRNResearchNov 2021Multi-modal Text Recognition Networks: Interactive Enhan… · code86.60
14CDistNet (Ours)—Nov 2021CDistNet: Perceiving Multi-Domain Character Distance for… · code86.25
15DiffusionSTR—Jun 2023DiffusionSTR: Diffusion Model for Scene Text Recognition86
16DPAN—Aug 2021papers-with-code · code85.50
17RCEED—Jun 2021Representation and Correlation Enhanced Encoder-Decoder … · code82.20
18CSTR—Feb 2021Revisiting Classification Perspective on Scene Text Reco… · code81.60
19Yet Another Text Recognizer—Jul 2021Why You Should Try the Real Data for the Scene Text Reco… · code80.20
20SEED—May 2020SEED: Semantics Enhanced Encoder-Decoder Framework for S… · code80
21TextScanner—Dec 2019TextScanner: Reading Characters in Order for Robust Scen…79.40
22SATRN—Oct 2019On Recognizing Texts of Arbitrary Shapes with 2D Self-At… · code79
23SAFL—Jan 2022SAFL: A Self-Attention Scene Text Recognizer with Focal … · code77.50
24ASTER—Jun 2018papers-with-code · code76.10
25DAN—Dec 2019Decoupled Attention Network for Text Recognition · code74.50
26AON—Nov 2017AON: Towards Arbitrarily-Oriented Text Recognition · code73
27ViTSTR—May 2021Vision Transformer for Fast and Efficient Scene Text Rec… · code72.60
28Baek et al.—Apr 2019What Is Wrong With Scene Text Recognition Model Comparis… · code71.80
29SAR—Nov 2018Show, Attend and Read: A Simple and Strong Baseline for … · code69.20
f-measure
1 row
#ModelOrgSubmittedPaper / codef-measure
01DAL—Dec 2020Dynamic Anchor Learning for Arbitrary-Oriented Object De… · code82.40
Fig 2 · Rows sorted by score within each metric. Shaded row marks SOTA. Dates reflect model or paper release where available, otherwise the date Codesota accessed the source.
§ 03 · Progress

18 steps
of state of the art.

Each row below marks a model that broke the previous record on accuracy. Intermediate submissions are kept in the leaderboard above; only SOTA-setting entries are re-listed here.

Higher scores win. Each subsequent entry improved upon the previous best.

SOTA line · accuracy
  1. Nov 12, 2017AON73
  2. Jun 25, 2018ASTER76.10
  3. Oct 10, 2019SATRN79
  4. Dec 28, 2019TextScanner79.40
  5. May 22, 2020SEED80
  6. Feb 22, 2021CSTR81.60
  7. Jun 13, 2021RCEED82.20
  8. Aug 1, 2021DPAN85.50
  9. Nov 22, 2021CDistNet (Ours)86.25
  10. Nov 30, 2021MATRNResearch86.60
  11. Dec 24, 2021S-GTR87.30
  12. Mar 7, 2022SIGA_S87.60
  13. Sep 8, 2022MGP-STR90.90
  14. May 23, 2023CLIP4STR-L (DataComp-1B)91.40
  15. Jul 23, 2023CPPD91.70
  16. Aug 30, 2023DTrOCR 105M93.50
  17. Mar 15, 2024TextBlockV2 (GPT-2)Jiahao Lyu et al., Fudan University97.70
  18. Apr 9, 2024JSTRFujitake98.70
Fig 3 · SOTA-setting models only. 18 entries span Nov 2017 → Apr 2024.
§ 04 · Literature

26 papers
tied to this benchmark.

Every paper below corresponds to at least one row in the leaderboard above. Click through for the arXiv preprint and, when available, the reference implementation.

§ 06 · Contribute

Have a score that beats
this table?

Submit a checkpoint and a reproduction script. We will run it, publish the score, and — if it takes the top — annotate the step on the progress chart with your name.

Submit a result ↵Read submission guide
What a submission needs
  • 01A public checkpoint or API endpoint
  • 02A reproduction script with frozen commit + seed
  • 03Declared evaluation environment (Python, deps)
  • 04One row per metric declared by this dataset
  • 05A contact so we can follow up on discrepancies