Full search page →

Evaluation protocols in detail

Optional · supplementary note 5 of 5 for DLCV 06 · Text in Images: OCR and Document Understanding. Nothing in the main text depends on it; citation numbers refer to the reference list at the bottom of this page.

Navigate: this page and this track: Navigate: this page and this track

On this page

Supplementary notes · 5

  1. CTC prefix beam search, step by step
  2. Shrinking and expanding text polygons
  3. Where the language model sits in a recognizer
  4. Reading vertical and mixed-direction text
  5. Evaluation protocols in detail

Paper guides · 4

  1. Guide to the Modern view surveys
  2. Baek et al., “What Is Wrong With Scene Text Recognition Model Comparisons?” (ICCV 2019)
  3. DBNet: a reading guide
  4. Donut: a reading guide

Deep Learning for Computer Vision

  1. 02 · Deep Image Classification
View the whole course tree →

This note fills in the details of text-detection and recognition metrics that change reported numbers.

“Do not care” regions. Ground truth marks illegible text and other text outside the task as “do not care”. In the protocol described for MLT-2019, a detection tt is discarded if its intersection with a “do not care” region dd covers more than half of dd, that is if ∣t∩d∣/∣d∣>0.5|t \cap d| / |d| \gt 0.5, where ∣⋅∣|\cdot| is area; remaining detections are matched one-to-one to the remaining ground truth with IoU >0.5\gt 0.5 [42]. Total-Text likewise asks that detections of “do not care” areas be filtered out before scoring [10]. A method that detects illegible text is therefore neither rewarded nor punished, which is the intent; but a system that merges a legible word with a neighboring “do not care” region can lose the legible word.

One-to-many matching (ICDAR 2013). The ICDAR 2011/2013 protocol follows the overlap-based method of Wolf and Jolion [51]: a match requires both the fraction of the ground truth covered and the fraction of the detection covered to exceed thresholds, and one-to-many or many-to-one matches (a line detected as several words, or several words as one line) are accepted with a reduced score, usually 0.8, instead of failing [1]. Numbers from this protocol and from the IoU protocol are not comparable.

Single-threshold H-mean. H-mean is computed at the confidence threshold chosen by the method’s authors, usually the one that maximizes it on the test set itself. This is optimistic and hides the precision–recall trade-off; AP, as used by RCTW-17 [39], integrates over all thresholds [1].

End-to-end vocabularies. End-to-end and word-spotting results are reported with a strongly contextualised list (100 words per image, including all words in the image plus distractors), a weakly contextualised list (all words in the test set), or a generic list of about 90,000 words [2]. With the 100-word list, recognition reduces to choosing among a few candidates; numbers with different lists cannot be compared.

Normalized and average edit distance. NED divides each line’s edit distance by the longer of the two strings, so every line contributes at most 1 regardless of length; 1−NED1 - \text{NED} is reported as a score [2]. RCTW-17’s end-to-end task also used the average edit distance (AED): each detection is matched to the ground-truth polygon of maximum IoU (or to nothing if no IoU exceeds 0.5), the edit distances of matched pairs are summed — an unmatched detection or ground truth counts as an edit distance against the empty string — and the sum is divided by the number of test images [39]. AED punishes missed and spurious text by its length, which NED alone does not.

Document metrics. DocVQA’s primary metric, ANLS, is an edit-distance-based similarity between predicted and ground-truth answers, averaged over questions; it was first proposed for scene-text question answering, and the DocVQA authors adopted it because exact-match accuracy gives zero credit to an answer with one OCR slip [44]. Read the metric’s definition on the benchmark’s evaluation page before comparing with other similarity scores. Donut’s TED-based accuracy for structured outputs is max⁡(0, 1−TED(pr,gt)/TED(ϕ,gt))\max\big(0,\ 1 - \text{TED}(\text{pr}, \text{gt}) / \text{TED}(\phi, \text{gt})\big), where TED is the tree edit distance between predicted and ground-truth trees and ϕ\phi is the empty tree; unlike field-level F1, it gives partial credit and measures the nesting of groups [34].

Normalization. Many recognition benchmarks evaluate only case-insensitive letters and digits; TrOCR, for example, filters its scene-text outputs to a 36-character set of lowercase letters and digits, while it scores IAM handwriting with case-sensitive CER [19]. Always state case sensitivity, punctuation handling and whitespace rules alongside the number.

References

  1. S. Long, X. He and C. Yao, “Scene text detection and recognition: The deep learning era,” International Journal of Computer Vision, vol. 129, pp. 161–184, 2021. doi
  2. X. Chen, L. Jin, Y. Zhu, C. Luo and T. Wang, “Text recognition in the wild: A survey,” ACM Computing Surveys, vol. 54, no. 2, 2021. doi
  3. L. Cui, Y. Xu, T. Lv and F. Wei, “Document AI: Benchmarks, models and applications,” arXiv:2111.08609, 2021. arXiv
  4. Q. Zhang, B. Wang, V. S.-J. Huang, J. Zhang, Z. Wang, H. Liang, C. He and W. Zhang, “Document parsing unveiled: Techniques, challenges, and prospects for structured information extraction,” arXiv:2410.21169, 2024. arXiv
  5. D. Karatzas, F. Shafait, S. Uchida, M. Iwamura, L. Gomez i Bigorda, S. Robles Mestre, J. Mas et al., “ICDAR 2013 Robust Reading Competition,” in Proc. International Conference on Document Analysis and Recognition (ICDAR), pp. 1484–1493, 2013. doi
  6. D. Karatzas, L. Gomez-Bigorda, A. Nicolaou, S. Ghosh, A. Bagdanov, M. Iwamura, J. Matas et al., “ICDAR 2015 competition on Robust Reading,” in Proc. International Conference on Document Analysis and Recognition (ICDAR), pp. 1156–1160, 2015. doi
  7. J. Baek, G. Kim, J. Lee, S. Park, D. Han, S. Yun, S. J. Oh and H. Lee, “What is wrong with scene text recognition model comparisons? Dataset and model analysis,” in Proc. IEEE/CVF ICCV, pp. 4714–4722, 2019. doi
  8. N. Otsu, “A threshold selection method from gray-level histograms,” IEEE Transactions on Systems, Man, and Cybernetics, vol. 9, no. 1, pp. 62–66, 1979. doi
  9. J. Sauvola and M. Pietikäinen, “Adaptive document image binarization,” Pattern Recognition, vol. 33, no. 2, pp. 225–236, 2000. doi
  10. C. K. Ch’ng and C. S. Chan, “Total-Text: A comprehensive dataset for scene text detection and recognition,” in Proc. ICDAR, pp. 935–942, 2017. doi
  11. Y. Liu, L. Jin, S. Zhang, C. Luo and S. Zhang, “Curved scene text detection via transverse and longitudinal sequence connection,” Pattern Recognition, vol. 90, pp. 337–345, 2019. doi
  12. X. Zhou, C. Yao, H. Wen, Y. Wang, S. Zhou, W. He and J. Liang, “EAST: An efficient and accurate scene text detector,” in Proc. IEEE CVPR, pp. 2642–2651, 2017. doi
  13. W. Wang, E. Xie, X. Li, W. Hou, T. Lu, G. Yu and S. Shao, “Shape robust text detection with progressive scale expansion network,” in Proc. IEEE/CVF CVPR, pp. 9328–9337, 2019. doi
  14. M. Liao, Z. Wan, C. Yao, K. Chen and X. Bai, “Real-time scene text detection with differentiable binarization,” in Proc. AAAI Conference on Artificial Intelligence, vol. 34, pp. 11474–11481, 2020. doi
  15. Y. Baek, B. Lee, D. Han, S. Yun and H. Lee, “Character region awareness for text detection,” in Proc. IEEE/CVF CVPR, pp. 9357–9366, 2019. doi
  16. B. Shi, X. Bai and C. Yao, “An end-to-end trainable neural network for image-based sequence recognition and its application to scene text recognition,” IEEE TPAMI, vol. 39, no. 11, pp. 2298–2304, 2017. doi
  17. A. Graves, S. Fernández, F. Gomez and J. Schmidhuber, “Connectionist temporal classification: Labelling unsegmented sequence data with recurrent neural networks,” in Proc. ICML, pp. 369–376, 2006. doi
  18. B. Shi, M. Yang, X. Wang, P. Lyu, C. Yao and X. Bai, “ASTER: An attentional scene text recognizer with flexible rectification,” IEEE TPAMI, vol. 41, no. 9, pp. 2035–2048, 2019. doi
  19. M. Li, T. Lv, J. Chen, L. Cui, Y. Lu, D. Florencio, C. Zhang, Z. Li and F. Wei, “TrOCR: Transformer-based optical character recognition with pre-trained models,” in Proc. AAAI Conference on Artificial Intelligence, vol. 37, pp. 13094–13102, 2023. doi
  20. D. Bautista and R. Atienza, “Scene text recognition with permuted autoregressive sequence models,” in Proc. ECCV, pp. 178–196, 2022. doi
  21. S. Fang, H. Xie, Y. Wang, Z. Mao and Y. Zhang, “Read like humans: Autonomous, bidirectional and iterative language modeling for scene text recognition,” in Proc. IEEE/CVF CVPR, pp. 7094–7103, 2021. doi
  22. Y. Liu, Z. Li, M. Huang, B. Yang, W. Yu, C. Li, X.-C. Yin, C.-L. Liu, L. Jin and X. Bai, “OCRBench: On the hidden mystery of OCR in large multimodal models,” Science China Information Sciences, vol. 67, 2024. doi
  23. M. Jaderberg, K. Simonyan, A. Vedaldi and A. Zisserman, “Synthetic data and artificial neural networks for natural scene text recognition,” arXiv:1406.2227, 2014. arXiv
  24. A. Gupta, A. Vedaldi and A. Zisserman, “Synthetic data for text localisation in natural images,” in Proc. IEEE CVPR, pp. 2315–2324, 2016. doi
  25. X. Liu, D. Liang, S. Yan, D. Chen, Y. Qiao and J. Yan, “FOTS: Fast oriented text spotting with a unified network,” in Proc. IEEE/CVF CVPR, pp. 5676–5685, 2018. doi
  26. P. Lyu, M. Liao, C. Yao, W. Wu and X. Bai, “Mask TextSpotter: An end-to-end trainable neural network for spotting text with arbitrary shapes,” in Proc. ECCV, pp. 71–88, 2018. doi
  27. Y. Liu, H. Chen, C. Shen, T. He, L. Jin and L. Wang, “ABCNet: Real-time scene text spotting with adaptive Bezier-curve network,” in Proc. IEEE/CVF CVPR, pp. 9806–9815, 2020. doi
  28. X. Zhong, J. Tang and A. Jimeno Yepes, “PubLayNet: Largest dataset ever for document layout analysis,” in Proc. ICDAR, pp. 1015–1022, 2019. doi
  29. B. Pfitzmann, C. Auer, M. Dolfi, A. S. Nassar and P. Staar, “DocLayNet: A large human-annotated dataset for document-layout segmentation,” in Proc. ACM SIGKDD, pp. 3743–3751, 2022. doi
  30. Y. Xu, M. Li, L. Cui, S. Huang, F. Wei and M. Zhou, “LayoutLM: Pre-training of text and layout for document image understanding,” in Proc. ACM SIGKDD, pp. 1192–1200, 2020. doi
  31. R. Smith, “An overview of the Tesseract OCR engine,” in Proc. ICDAR, vol. 2, pp. 629–633, 2007. doi
  32. Y. Huang, T. Lv, L. Cui, Y. Lu and F. Wei, “LayoutLMv3: Pre-training for document AI with unified text and image masking,” in Proc. ACM International Conference on Multimedia, pp. 4083–4091, 2022. doi
  33. G. Jaume, H. K. Ekenel and J.-P. Thiran, “FUNSD: A dataset for form understanding in noisy scanned documents,” in Proc. ICDAR Workshops (OST), 2019. doi
  34. G. Kim, T. Hong, M. Yim, J. Nam, J. Park, J. Yim, W. Hwang, S. Yun, D. Han and S. Park, “OCR-free document understanding transformer,” in Proc. ECCV, pp. 498–517, 2022. doi
  35. K. Lee, M. Joshi, I. R. Turc, H. Hu, F. Liu, J. M. Eisenschlos, U. Khandelwal, P. Shaw, M.-W. Chang and K. Toutanova, “Pix2Struct: Screenshot parsing as pretraining for visual language understanding,” in Proc. ICML, PMLR vol. 202, pp. 18893–18912, 2023. PMLR
  36. L. Blecher, G. Cucurull, T. Scialom and R. Stojnic, “Nougat: Neural optical understanding for academic documents,” in Proc. ICLR, 2024. OpenReview · arXiv
  37. U.-V. Marti and H. Bunke, “The IAM-database: An English sentence database for offline handwriting recognition,” International Journal on Document Analysis and Recognition, vol. 5, pp. 39–46, 2002. doi
  38. C.-L. Liu, F. Yin, D.-H. Wang and Q.-F. Wang, “CASIA online and offline Chinese handwriting databases,” in Proc. ICDAR, pp. 37–41, 2011. doi
  39. B. Shi, C. Yao, M. Liao, M. Yang, P. Xu, L. Cui, S. Belongie, S. Lu and X. Bai, “ICDAR2017 competition on reading Chinese text in the wild (RCTW-17),” in Proc. ICDAR, pp. 1429–1434, 2017. doi · arXiv
  40. T.-L. Yuan, Z. Zhu, K. Xu, C.-J. Li, T.-J. Mu and S.-M. Hu, “A large Chinese text dataset in the wild,” Journal of Computer Science and Technology, vol. 34, pp. 509–521, 2019. doi · arXiv
  41. F. Xue and R. Ishida (eds.), “Requirements for Chinese Text Layout,” W3C Group Note (draft of 7 October 2026). W3C
  42. N. Nayef, Y. Patel, M. Busta, P. N. Chowdhury, D. Karatzas, W. Khlif, J. Matas et al., “ICDAR2019 Robust Reading Challenge on multi-lingual scene text detection and recognition — RRC-MLT-2019,” in Proc. ICDAR, pp. 1582–1587, 2019. doi · arXiv
  43. R. A. Wagner and M. J. Fischer, “The string-to-string correction problem,” Journal of the ACM, vol. 21, no. 1, pp. 168–173, 1974. doi
  44. M. Mathew, D. Karatzas and C. V. Jawahar, “DocVQA: A dataset for VQA on document images,” in Proc. IEEE/CVF WACV, pp. 2199–2208, 2021. doi
  45. A. Mishra, K. Alahari and C. V. Jawahar, “Scene text recognition using higher order language priors,” in Proc. BMVC, 2012. doi
  46. K. Wang, B. Babenko and S. Belongie, “End-to-end scene text recognition,” in Proc. IEEE ICCV, pp. 1457–1464, 2011. doi
  47. L. Ouyang, Y. Qu, H. Zhou, J. Zhu, R. Zhang, Q. Lin, B. Wang, Z. Zhao et al., “OmniDocBench: Benchmarking diverse PDF document parsing with comprehensive annotations,” in Proc. IEEE/CVF CVPR, pp. 24838–24848, 2025. CVF
  48. H. Wei, C. Liu, J. Chen, J. Wang, L. Kong, Y. Xu, Z. Ge, L. Zhao et al., “General OCR theory: Towards OCR-2.0 via a unified end-to-end model,” arXiv:2409.01704, 2024. arXiv
  49. J. Poznanski, A. Rangapur, J. Borchardt, J. Dunkelberger, R. Huff, D. Lin, C. Wilhelm et al., “olmOCR: Unlocking trillions of tokens in PDFs with vision language models,” arXiv:2502.18443, 2025. arXiv
  50. H. Wei, Y. Sun and Y. Li, “DeepSeek-OCR: Contexts optical compression,” arXiv:2510.18234, 2025. arXiv
  51. C. Wolf and J.-M. Jolion, “Object count/area graphs for the evaluation of object detection and segmentation algorithms,” International Journal of Document Analysis and Recognition, vol. 8, pp. 280–296, 2006. doi