# microsoft/table-transformer

Table Transformer (TATR) is a deep learning model for extracting tables from unstructured documents (PDFs and images). This is also the official repository for the PubTables-1M dataset and GriTS evaluation metric.

Repository: https://github.com/microsoft/table-transformer
Canonical: https://ross.abutalabs.com/products/table-transformer
Language: Python
License: MIT
License Family: permissive
Topics: table-detection, table-extraction, table-structure-recognition, table-functional-analysis
Last push: 2024-06-24T13:41:20+00:00

## Health v2 (maintenance only)
Score: 23/100 (v2, computed 2026-09-03T02:20:16.233290+00:00)
- activity 0, release rhythm 8, longevity 100
- inputs: {"age_days": 1934, "days_push": 800, "days_rel": null, "gap_med": null, "n_releases_24m": 0}
- flags: none
- formula: round(0.45*activity + 0.35*rhythm + 0.20*longevity); archived -> min(score, 10)

## Adoption (not part of the score)
Stars 2939, forks 315 (observed 2026-08-28T04:07:30.986876+00:00)

## What it is
Table Transformer (TATR) is a deep learning object detection model from Microsoft for detecting, recognizing, and extracting tables from unstructured documents like PDFs and images, outputting HTML or CSV. It is also the official repository for the PubTables-1M dataset and the GriTS evaluation metric.

## Use cases
- extract tables from pdf documents
- detect tables in scanned images
- convert table images to html or csv
- recognize table structure in unstructured documents
- train a custom table extraction model
- evaluate table structure recognition models with grits
- download pubtables-1m dataset for research

## When to choose
- you need state-of-the-art table detection and structure recognition from PDFs or images
- you want pre-trained weights trained on the large PubTables-1M dataset
- you need a research-grade model with training and evaluation code included

## When to avoid
- you need a turnkey end-to-end document parser with OCR included - text extraction must be supplied separately
- you need pre-trained weights for document domains outside those covered by released models
- you want a simple no-GPU inference solution

## Facets
- artifact type: library
- maturity: active
- function: machine-learning, computer-vision, ocr, pdf, image-processing, parser
- domain: computer-vision, pdf, machine-learning, data-science
- platform: python, cross-platform
- tags: table-detection, table-structure-recognition, object-detection, pubtables-1m, grits, document-understanding, pytorch, natural-language-processing

## Member repositories
- microsoft/table-transformer (main) score 23

## Provenance
- Observed fields: from GitHub, fetched 2026-08-28T04:07:30.986876+00:00.
- Health v2: computed from the inputs above; adoption is never an input.
- Inferred fields (summary, facets, guidance): AI-extracted, prompt v1, taxonomy v1, on 2026-08-30T07:32:58.447726+00:00, confidence not recorded.
  - readme: https://github.com/microsoft/table-transformer (fetched 2026-08-28T04:07:30.986876+00:00, sha 0c8326376c78)
- Data as of 2026-08-30T08:39:29.467469+00:00.