Optical Character Recognition2020en

mldoc-zero-shot-english-to-japanese

Dataset from Papers With Code

Metrics:accuracy, cer, wer, f1
Current State of the Art

MultiFiT, pseudo

Unknown

69.57

accuracy

accuracy Progress Over Time

Showing 2 breakthroughs from May 2018 to Sep 2019

67.468.068.669.269.8May 2018Sep 2019accuracyDate

Key Milestones

May 2018
MultiCCA + CNN

From paper: A Corpus for Multilingual Document Classification in Eight Languages

67.6
Sep 2019
MultiFiT, pseudoCurrent SOTA

From paper: MultiFiT: Efficient Multi-lingual Language Model Fine-tuning

69.6
+2.9%
Total Improvement
2.9%
Time Span
1y 4m
Breakthroughs
2
Current SOTA
69.6

Top Models Performance Comparison

Top 3 models ranked by accuracy

accuracy1MultiFiT, pseudo69.6100.0%2MultiCCA + CNN67.697.2%3Massively Multilingual Se...60.386.7%0%25%50%75%100%% of best
Best Score
69.6
Top Model
MultiFiT, pseudo
Models Compared
3
Score Range
9.3

accuracyPrimary

Related Papers3

Other Optical Character Recognition Datasets