AI Tools.

Search

object detection by microsoft

table-transformer-detection

A DETR-based object detection model from Microsoft Research trained to locate tables in document images. It is the detection stage in a two-step pipeline — a separate structure recognition model then parses the detected table's rows and columns.

Summary text generated by an automated pipeline from the model card · Not individually reviewed or run by us · How this page is made

From the model card

Fields below are copied from the tags and counters on the HuggingFace repository microsoft/table-transformer-detection at our last fetch. They are set by the uploader, not verified by us; rows with no tag are omitted. How this page is made.

Publisher (HF namespace)
microsoft
Pipeline tag
object-detection
Library
Transformers
Framework tags
PyTorch
Weight formats
safetensors
License tag
mit — read the license file in the repo before relying on it
Papers cited
arXiv:2110.00061
Downloads (HF counter at last fetch)
585,407
Likes (HF counter at last fetch)
428
Model card
https://huggingface.co/microsoft/table-transformer-detection

Use cases

  • Locating tables in scanned PDFs before OCR extraction
  • Document intelligence pipelines processing financial reports
  • Pre-processing research papers to extract tabular data
  • Building automated data entry tools from document images

Pros

  • Pretrained on PubTables-1M, a large and diverse table dataset
  • Pairs with table-transformer-structure-recognition for end-to-end parsing
  • MIT licensed
  • PyTorch + SafeTensors weights available

Cons

  • Detects bounding boxes only — does not parse cell content
  • Performance drops on tables with complex spanning cells
  • No native PDF input — requires prior PDF-to-image conversion
  • Two-model pipeline adds latency compared to end-to-end solutions

Tags

transformerspytorchsafetensorstable-transformerobject-detectionarxiv:2110.00061license:mitendpoints_compatibleregion:usdeploy:sagemakerdeploy:azure