A Transformer-based object detection model pre-trained for table structure recognition, including row and column detection, using the PubTables1M and FinTabNet datasets.
Table Transformer (TATR) - Table Structure Recognition is a Transformer-based object detection model trained on the PubTables1M and FinTabNet datasets. It is designed to detect and recognize table structures, such as rows, columns, and table boundaries, in unstructured documents. Based on DETR (DEtection TRansformer), the model employs a "normalize before" approach, applying layer normalization before self- and cross-attention. This model is particularly useful for document analysis, table parsing, and structured data extraction from scanned PDFs and images.
MIT
Microsoft
object detection
N.A.
Open
Sector Agnostic
12/03/25 06:34:57
0
MIT
© 2026 - Copyright AIKosh. All rights reserved. This portal is developed by National e-Governance Division for AIKosh mission.