table-transformer 1.0.5

Table Transformer

Table Transformer Library

Original repository: https://github.com/microsoft/table-transformer

Introduction

This is the Table Transformer Model developed by Brandon Smock et al. of Microsoft AI. This repository consists of Table Structure Recognition (TATR) for detecting and extracting table infomation into popular formats such as CSV or HTML table, plus text recognition using EasyOCR.

Installation

pip install table-transformer

Usage

The full model usage can be found here:

from table_transformer import TableExtractionPipeline, get_cell_coordinates_by_row, apply_ocr, get_detection_class_thresholds, get_structure_class_thresholds

import numpy as np
import csv
import easyocr
from tqdm.auto import tqdm

det_config = "table_transformer\src\detection_config.json"
str_config = "table_transformer\src\structure_config.json"


pipe = TableExtractionPipeline(det_device="cpu", str_device="cpu",
                 det_model_path=".\pubtables1m_detection_detr_r18.pth",
                 str_model_path=".\TATR-v1.1-Pub-msft.pth",
                 det_config_path=det_config, str_config_path=str_config,)

from PIL import Image

img = Image.open("table.jpg")

reader = easyocr.Reader(['en','vi']) # this needs to run only once to load the model into memory

pipe(reader, img)
     

Evaluation

With structure recognition, the original author has evaluated the v1.0 model on PubTables-1M with great results. With other datasets such as PubTabNet, the score is quite good.

You can check out the score and run the evaluation with your own dataset in this link.

Version history

• v1.0.5: Added Table Detection, ending up with a full Table Extraction Pipeline.
• v1.0.3: Removed unnecessary code and added new functionalities.
• v1.0.2: Initial version.