Python library to extract tabular data from images and scanned PDFs

Overview

image

image image image

Overview

ExtractTable - API to extract tabular data from images and scanned PDFs

The motivation is to make it easy for developers to extract tabular data from images or scanned PDF files without worrying about the table area, column coordinates, rotation et al.

Prerequisite

API Key: All requests to ExtractTable are authorized by an API Key. FREE credits here. The same API Key can also be used for conversions on the browser at Web Pro.

Installation

pip install -U ExtractTable

Basic Usage

Ok, enough selling. Let the ease in coding do the talk, and the output encourages you to buy credits; put that timer on and count the LOC.

from ExtractTable import ExtractTable
et_sess = ExtractTable(api_key=YOUR_API_KEY)        # Replace your VALID API Key here
print(et_sess.check_usage())        # Checks the API Key validity as well as shows associated plan usage 
table_data = et_sess.process_file(filepath=Location_of_Image_with_Tables, output_format="df")

# To process PDF, make use of pages ("1", "1,3-4", "all") params in the read_pdf function
table_data = et_sess.process_file(filepath=Location_of_PDF_with_Tables, output_format="df", pages="all")

Detailed Library Usage

The tutorial available at Open In Colab takes you through

1. Installation
2. Import and check version
3. Create Session & Validate API Key
    3.1 Create Session with your API Key
    3.2 Validate the Key and check the plan usage
    3.3 Check Usage Details
4. Trigger the extraction process
    4.1 Accepted Input Types
    4.2 Process an IMAGE Input
    4.3 Process a PDF Input
    4.4 Output options
    4.5 Explore session objects
5. Explore the Output
    5.1 Output Structure
    5.2 Output Details
6. Make Corrections
    6.1 Split Merged Rows
    6.2 Split Merged Columns
    6.3 Fix Decimal Format
    6.4 Fix Date Format
7. Helpful Code Snippets
    7.1 Get text data
    7.2 Table output to Excel

Woahh, as simple as that ?!

Certainly. Do you know the current ExtractTable users use it for

  • Bank Statement
  • Medical Records
  • Invoice Details
  • Tax forms
  • Tender Notices

Its up to you now to explore the ways.

Explore

check the complete server response of the latest job with et_sess.ServerResponse.json()

{
    "JobStatus": <string>,                              # Status of the triggered Process  @ JOB-LEVEL
    "Pages": <integer>,                                 # Number of pages processed in this request @ PAGE-LEVEL
    "Tables": [<list of key-value objects of table>     # List of all tables found @ TABLE-LEVEL
        {
            "Page": <integer>,                              ## Page number in which this table is found
            "CharacterConfidence": <float>,                 ## Accuracy of Characters recognized from the input-page
            "LayoutConfidence": <float>,                    ## Accuracy of table layout's design decision
            "TableJson": <dict>,                            ## Table Cell Text in key-value format with index orientation - {row#: {col#: <str>}}
            "TableCoordinates": <dict>,                     ## Top-left & Bottom-right Cell Coordinates - {row#: {col#: <list(x1,y1,x2,y2)>}}
            "TableConfidence": <dict>                       ## Cell level accuracy of detected characters - {row#: {col#: <float>}}
        },
    {...}                                               ## ... more "Tables" objects
    ],
    "Lines": [<list of key-value objects>               # Pagewise Line details @ PAGE-LEVEL
        {
            "Page": <integer>,                          # Page number in which the lines are found
            "CharacterConfidence": <float>,             # Average Accuracy of all Characters recognized from the input-page
            "LinesArray": [
                <list of key-value objects of line>     # Ordered list of lines in this page @ LINE-LEVEL
                {
                    "Line": <str>,                          ## Detected text of the complete line
                    "WordsArray": [
                        <list of key-value objects>         ## Word level datails in this line @ WORD-LEVEL
                        {
                            "Conf": <float>,                    ### Accuracy of recognized characters of the word
                            "Word": <str>,                      ### Detected text of the word
                            "Loc": [x1, y1, x2, y2]             ### Top-left & Bottom-right coordinates, w.r.t the input-page width-height dimensions
                        },
                    {...}                                   ### More "WordsArray" objects
                    ]
                },
            {...}                                       ## More "LinesArray" objects
            ]
        },
    {...}                                               # More Pagewise "Lines" details
    ]
}

Bug Reports

Bug reports/fixes are most welcome and greatly appreciated with API credits. For support reach us at [email protected]

License

This project is licensed under the Apache License 2.0, see the LICENSE file for details.

Social Media

Follow us on Social media for library updates and free credits.

Image      Image

Comments
  • bug: holding when the program running after some samples

    bug: holding when the program running after some samples

    Describe the bug A clear and concise description of what the bug is. keep holding my apI key prefix is o6No6aqYRhrQ2MWxtDDyTeHiiUg**** image

    To Reproduce Steps to reproduce the behavior: or the code you tried

    Expected behavior A clear and concise description of what you expected to happen.

    Additional context Add any other context about the problem here.

    bug 
    opened by franztao 5
  • bug: function

    bug: function "et_sess.save_output(output_folder, output_format="csv")" output file, the file name lack some alpha of the origin full name

    Describe the bug A clear and concise description of what the bug is. my picture name is all suffix png. such as "[email protected]_14-1-4.png"

    image

    To Reproduce Steps to reproduce the behavior: or the code you tried

    Expected behavior A clear and concise description of what you expected to happen.

    Additional context Add any other context about the problem here.

    bug 
    opened by franztao 3
  • found some bugs and list the bugs out

    found some bugs and list the bugs out

    Describe the bug A clear and concise description of what the bug is. 1.不能识别出垮列的文本,识别成表格时,不符合逻辑的分开成两边 image image

    2.不能识别加减号,can not recognize Plus minus sign. 31.2 + 4.98 image 3.不能够识别上下标,can not recognize subscript and supscript. image 4.ocr识别丢失字符 loss some recognized tokens image 5.长的表格,有部分没有识别出来 long size table,can not recognize the bottem part image image 6.cell中有化学式的,识别不出来,when there is chemical formulate in cell, can not recognize the table image

    To Reproduce Steps to reproduce the behavior: or the code you tried

    Expected behavior A clear and concise description of what you expected to happen. I can solve these problems with us.

    Additional context Add any other context about the problem here.

    bug 
    opened by franztao 2
  • question: what meaning is LayoutConfidence?

    question: what meaning is LayoutConfidence?

    "CharacterConfidence": , # Average Accuracy of all Characters recognized from the input-page "LayoutConfidence": , ## Accuracy of table layout's design decision please give out the detaild decription or calculate function code about CharacterConfidence,LayoutConfidence

    good first issue 
    opened by franztao 2
  • Invalid cross-device link

    Invalid cross-device link

    Describe the bug On some OS, we can not save output file to temporary directory (let's say /tmp) and move it to a new place. It throws the following error :

    os.replace(each_tbl_path, os.path.join(output_folder, input_fname+os.path.basename(each_tbl_path)))
    OSError: [Errno 18] Invalid cross-device link: '/tmp/tmp7hqcm0fh/_table_1.csv' -> '/var/www/python/app/tmp/details_table_1.csv'
    

    After checking the source code, it appears ExtractTable use os.replace to move the file. This method does not support moving file from a partition to an other : https://stackoverflow.com/questions/42392600/oserror-errno-18-invalid-cross-device-link

    To Reproduce I use Python 3.6 in a venv. You will need two different system parts, and invoke save_output from ExtractTable-py library, to save file from a filesystem to an other. I have not tried, but I think you can simply reproduce this bug by invoking os.replace without calling ExtractTable-py.

    Expected behavior Move the file from a filesystem to an other. I think using shutil.move would be a preferable way to achieve file moving than os.replace.

    bug 
    opened by Elegye 2
  • MakeCorrections API - How do you chain corrections

    MakeCorrections API - How do you chain corrections

    Hi there, I'm trying to use multiple correction commands but it isn't working as the object becomes a list after the first correction. Is there something I'm missing here? Thanks!

    good first issue 
    opened by kylebutts 1
  • character ocr can support latex format?

    character ocr can support latex format?

    Is your feature request related to a problem? Please describe. A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

    Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered.

    Describe the solution you'd like [optional, but helpful] A clear and concise description of what you want to happen.

    Additional context Add any other context or screenshots about the feature request here.

    opened by franztao 1
  • please, do you have tools of transform ExtracTable output file type to CoCo file type(other open source Detection file type)?

    please, do you have tools of transform ExtracTable output file type to CoCo file type(other open source Detection file type)?

    Is your feature request related to a problem? Please describe. A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

    Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered.

    Describe the solution you'd like [optional, but helpful] A clear and concise description of what you want to happen.

    Additional context Add any other context or screenshots about the feature request here.

    opened by franztao 1
  • Custom output path when the output_format is csv

    Custom output path when the output_format is csv

    Is your feature request related to a problem? Please describe. When the output_format is set to csv the csv file is written to some random path in /tmp location.

    Describe the solution you'd like [optional, but helpful] Define a parameter in the process_file like output_file which takes the absolute path where the file needs to be written along with the file name

    opened by padmano 1
  • Is it possible to  get the data in excel by maintaining table structure?

    Is it possible to get the data in excel by maintaining table structure?

    Is your feature request related to a problem? Please describe. A clear and concise description of what the problem is. Ex. I'm always frustrated when [...]

    Describe alternatives you've considered A clear and concise description of any alternative solutions or features you've considered.

    Describe the solution you'd like [optional, but helpful] A clear and concise description of what you want to happen.

    Additional context Add any other context or screenshots about the feature request here.

    opened by jcthink 1
  • Character and Layout Confidence

    Character and Layout Confidence

    Hi, need some definition material for Character and Layout Confidence like how it is calculated mathematically using below code. Thanks.

    for idx, each_table in enumerate(et_sess.ServerResponse.json()['Tables']):
        print("CharacterConfidence = ", each_table['CharacterConfidence'])
        print("LayoutConfidence = ", each_table['LayoutConfidence'])
    
    good first issue 
    opened by muhdzubair 1
  • Consider user hints on the table structure information

    Consider user hints on the table structure information

    Is your feature request related to a problem? Please describe. "while you do whatever you want, why not consider the our hints" is the developers feedback on many instances

    Describe alternatives you've considered Developers are tackling with their custom post processing.

    Describe the solution you'd like [optional, but helpful] Pros: May be it is a worth taking a look as most of the post processing involves in similar approaches that resolves majority issues. Cons: computing cost

    feature/idea 
    opened by akshowhini 0
  • Capture Vertically center aligned columns

    Capture Vertically center aligned columns

    Refer: https://stackoverflow.com/questions/58238981/extracting-table-from-a-pdf-table-without-vertical-lines

    Do not miss: Joelgeraci's comment to the question

    feature/idea 
    opened by akshowhini 0
Releases(v2.4.0)
Owner
Org. Account
You, I and they have the same problem to solve !?!?
Org. Account
A semi-automatic open-source tool for Layout Analysis and Region EXtraction on early printed books.

LAREX LAREX is a semi-automatic open-source tool for layout analysis on early printed books. It uses a rule based connected components approach which

162 Jan 05, 2023
A simple python program to record security cam footage by detecting a face and body of a person in the frame.

SecurityCam A simple python program to record security cam footage by detecting a face and body of a person in the frame. This code was created by me,

1 Nov 08, 2021
End-to-end pipeline for real-time scene text detection and recognition.

Real-time-Scene-Text-Detection-and-Recognition-System End-to-end pipeline for real-time scene text detection and recognition. The detection model use

Fangneng Zhan 89 Aug 04, 2022
Detect handwritten words in a text-line (classic image processing method).

Word segmentation Implementation of scale space technique for word segmentation as proposed by R. Manmatha and N. Srimal. Even though the paper is fro

Harald Scheidl 190 Jan 03, 2023
Library used to deskew a scanned document

Deskew //Note: Skew is measured in degrees. Deskewing is a process whereby skew is removed by rotating an image by the same amount as its skew but in

Stéphane Brunner 273 Jan 06, 2023
Code for the paper STN-OCR: A single Neural Network for Text Detection and Text Recognition

STN-OCR: A single Neural Network for Text Detection and Text Recognition This repository contains the code for the paper: STN-OCR: A single Neural Net

Christian Bartz 496 Jan 05, 2023
Text modding tools for FF7R (Final Fantasy VII Remake)

FF7R_text_mod_tools Subtitle modding tools for FF7R (Final Fantasy VII Remake) There are 3 tools I made. make_dualsub_mod.exe: Merges (or swaps) subti

10 Dec 19, 2022
A curated list of papers, code and resources pertaining to image composition

A curated list of resources including papers, datasets, and relevant links pertaining to image composition.

BCMI 391 Dec 30, 2022
code for our ICCV 2021 paper "DeepCAD: A Deep Generative Network for Computer-Aided Design Models"

DeepCAD This repository provides source code for our paper: DeepCAD: A Deep Generative Network for Computer-Aided Design Models Rundi Wu, Chang Xiao,

Rundi Wu 85 Dec 31, 2022
A curated list of promising OCR resources

Call for contributor(paper summary,dataset generation,algorithm implementation and any other useful resources) awesome-ocr A curated list of promising

wanghaisheng 1.6k Jan 04, 2023
Code for the paper: Fusformer: A Transformer-based Fusion Approach for Hyperspectral Image Super-resolution

Fusformer Code for the paper: "Fusformer: A Transformer-based Fusion Approach for Hyperspectral Image Super-resolution" Plateform Python 3.8.5 + Pytor

Jin-Fan Hu (胡锦帆) 11 Dec 12, 2022
第一届西安交通大学人工智能实践大赛(2018AI实践大赛--图片文字识别)第一名;仅采用densenet识别图中文字

OCR 第一届西安交通大学人工智能实践大赛(2018AI实践大赛--图片文字识别)冠军 模型结果 该比赛计算每一个条目的f1score,取所有条目的平均,具体计算方式在这里。这里的计算方式不对一句话里的相同文字重复计算,故f1score比提交的最终结果低: - train val f1score 0

尹畅 441 Dec 22, 2022
document image degradation

ocrodeg The ocrodeg package is a small Python library implementing document image degradation for data augmentation for handwriting recognition and OC

NVIDIA Research Projects 134 Nov 18, 2022
A python programusing Tkinter graphics library to randomize questions and answers contained in text files

RaffleOfQuestions Um programa simples em python, utilizando a biblioteca gráfica Tkinter para randomizar perguntas e respostas contidas em arquivos de

Gabriel Ferreira Rodrigues 1 Dec 16, 2021
Image augmentation library in Python for machine learning.

Augmentor is an image augmentation library in Python for machine learning. It aims to be a standalone library that is platform and framework independe

Marcus D. Bloice 4.8k Jan 04, 2023
Reference Code for AAAI-20 paper "Multi-Stage Self-Supervised Learning for Graph Convolutional Networks on Graphs with Few Labels"

Reference Code for AAAI-20 paper "Multi-Stage Self-Supervised Learning for Graph Convolutional Networks on Graphs with Few Labels" Please refer to htt

Ke Sun 1 Feb 14, 2022
Toolbox for OCR post-correction

Ochre Ochre is a toolbox for OCR post-correction. Please note that this software is experimental and very much a work in progress! Overview of OCR pos

National Library of the Netherlands / Research 117 Nov 10, 2022
Zoom , GoogleMeets에서 Vtuber 데뷔하기

EasyVtuber Facial landmark와 GAN을 이용한 Character Face Generation Google Meets, Zoom 등에서 자신만의 웹툰, 만화 캐릭터로 대화해보세요! 악세사리는 어느정도 추가해도 잘 작동해요! 안타깝게도 RTX 2070

Gunwoo Han 140 Dec 23, 2022
Detect the mathematical formula from the given picture and the same formula is extracted and converted into the latex code

Mathematical formulae extractor The goal of this project is to create a learning based system that takes an image of a math formula and returns corres

6 May 22, 2022