Search Results for author:
Found 4 papers, 2 papers with code
Is it an i or an l: Test-time Adaptation of Text Line Recognition Models
While text line recognition models are generally trained on large corpora of real and synthetic data, such models can still make frequent mistakes if the handwriting is inscrutable or the image acquisition process adds corruptions, such as noise, blur, compression, etc.
OCR Language Models with Custom Vocabularies
Language models are useful adjuncts to optical models for producing accurate optical character recognition (OCR) results.
XTREME-UP: A User-Centric Scarce-Data Benchmark for Under-Represented Languages
We evaluate commonly used models on the benchmark.
Rethinking Text Line Recognition Models
Unlike the more common Transformer-based models, this architecture can handle inputs of arbitrary length, a requirement for universal line recognition.
Ranked #2 on
Handwritten Text Recognition
on IAM
(using extra training data)
