Open Models Almanac
Methodsdata curationsynthetic data

single mention · data curation

Visual augmentation for robust document OCR

Applies visual distortions such as blur, noise, and geometric transformations to OCR training data.

The sources on this page only mention it in passing. The line above is the extraction's one-line gloss. See synthetic data for the methods filed beside it.

Filed under data curation › synthetic data