An Empirical Study of Scaling Law for OCR
The laws of model size, data volume, computation and model performance have been extensively studied in the field of Natural Language Processing (NLP). However, the scaling laws in Optical Character Recognition (OCR) have not yet been investigated. To address this, we conducted comprehensive studies that involved examining the correlation between performance and the scale of models, data volume and computation in the field of text recognition.Conclusively, the study demonstrates smooth power laws between performance and model size, as well as training data volume, when other influencing factors are held constant. Additionally, we have constructed a large-scale dataset called REBU-Syn, which comprises 6 million real samples and 18 million synthetic samples. Based on our scaling law and new dataset, we have successfully trained a scene text recognition model, achieving a new state-ofthe-art on 6 common test benchmarks with a top-1 average accuracy of 97.42%. The models and dataset are publicly available at https://github.com/large-ocr-model/large-ocr-model.github.io.
PDF Abstract
Datasets
ICDAR 2013
SVT
Results from the Paper
Ranked #1 on
Scene Text Recognition
on ICDAR2013
(using extra training data)
| Task | Dataset | Model | Metric Name | Metric Value | Global Rank | Uses Extra Training Data |
Benchmark |
|---|---|---|---|---|---|---|---|
| Scene Text Recognition | CUTE80 | CLIP4STR-B* | Accuracy | 99.65 | # 2 | ||
| Scene Text Recognition | ICDAR2013 | CLIP4STR-L* | Accuracy | 99.42 | # 1 | ||
| Scene Text Recognition | ICDAR2015 | CLIP4STR-L* | Accuracy | 92.6 | # 2 | ||
| Scene Text Recognition | SVT | CLIP4STR-B* | Accuracy | 98.76 | # 2 | ||
| Scene Text Recognition | SVTP | CLIP4STR-L* | Accuracy | 98.13 | # 3 |
