Mustard Dataset (Table Structure Recognition)

About Dataset

MUSTARD (Multilingual Scanned and Scene Table Structure Recognition Dataset) is a diverse dataset curated for table structure recognition across multiple languages. The dataset consists of tables extracted from magazines, including printed, scanned, and scene-text tables, labeled with Optimized Table Structure Language (OTSL) sequences. It is designed to facilitate research in multilingual table structure recognition, particularly for non-English documents.