We gratefully acknowledge support from
the Simons Foundation and member institutions.
Full-text links:


Current browse context:


Change to browse by:

References & Citations

DBLP - CS Bibliography


(what is this?)
CiteULike logo BibSonomy logo Mendeley logo del.icio.us logo Digg logo Reddit logo ScienceWISE logo

Computer Science > Machine Learning

Title: PubTables-1M: Towards a universal dataset and metrics for training and evaluating table extraction models

Abstract: Recently, interest has grown in applying machine learning to the problem of table structure inference and extraction from unstructured documents. However, progress in this area has been challenging both to make and to measure, due to several issues that arise in training and evaluating models from labeled data. This includes challenges as fundamental as the lack of a single definitive ground truth output for each input sample and the lack of an ideal metric for measuring partial correctness for this task. To address these issues we propose a new dataset, PubMed Tables One Million (PubTables-1M), and a new class of metric, grid table similarity (GriTS). PubTables-1M is nearly twice as large as the previous largest comparable dataset, contains highly-detailed structure annotations, and can be used for models across multiple architectures and modalities. Further, it addresses issues such as ambiguity and lack of consistency in the annotations via a novel canonicalization and quality control procedure. We apply DETR to table extraction for the first time and show that object detection models trained on PubTables-1M produce excellent results out-of-the-box for all three tasks of detection, structure recognition, and functional analysis. It is our hope that PubTables-1M and GriTS can further progress in this area by creating data and metrics suitable for training and evaluating a wide variety of models for table extraction. Data and code will be released at this https URL
Subjects: Machine Learning (cs.LG); Computer Vision and Pattern Recognition (cs.CV)
Cite as: arXiv:2110.00061 [cs.LG]
  (or arXiv:2110.00061v2 [cs.LG] for this version)

Submission history

From: Brandon Smock [view email]
[v1] Thu, 30 Sep 2021 19:42:07 GMT (5395kb,D)
[v2] Tue, 12 Oct 2021 17:43:57 GMT (5377kb,D)
[v3] Thu, 18 Nov 2021 20:28:13 GMT (1905kb,D)

Link back to: arXiv, form interface, contact.