Optical Character Recognition2020en

bbcsport

Dataset from Papers With Code

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

MPAD-path

Unknown

99.59

accuracy

accuracy Progress Over Time

Showing 2 breakthroughs from Apr 2019 to Aug 2019

95.396.597.798.8100.0Apr 2019Aug 2019accuracyDate

Key Milestones

Apr 2019
ApproxRepSet

From paper: Rep the Set: Neural Networks for Learning Set Representations

95.7
Aug 2019
MPAD-pathCurrent SOTA

From paper: Message Passing Attention Networks for Document Understanding

99.6
+4.0%
Total Improvement
4.0%
Time Span
4m
Breakthroughs
2
Current SOTA
99.6

Top Models Performance Comparison

Top 4 models ranked by accuracy

accuracy1MPAD-path99.6100.0%2Orthogonalized Soft VSM97.798.1%3ApproxRepSet95.796.1%4REL-RWMD k-NN95.295.6%0%25%50%75%100%% of best
Best Score
99.6
Top Model
MPAD-path
Models Compared
4
Score Range
4.4

accuracyPrimary

Related Papers4

Other Optical Character Recognition Datasets

bbcsport Benchmark - Optical Character Recognition | CodeSOTA