EXTRACTIFY ML9Forces

Problem Statement-2

Jan 8, 2022

1. GitHub-Repo:

1. Documentation : ( Setup > Readme ) https://github.com/ML9-FORCES/extractify-doc

2. Model-API : https://github.com/ML9-FORCES/extractify-api

3. Frontend : https://github.com/ML9-FORCES/extractify-frontend

2. Model File Link:

1. https://drive.google.com/drive/folders/1iWytxFuWd8J9n7Sgk-yX_GDN7ZxPBkk8?usp=sharing

3. Public Test Data Submission File Link:

1. https://drive.google.com/drive/folders/1KjIRC28jPodX3t7s6ClHLPFGCr8e-4j4?usp=sharing

4. Demo Video Link:

1. https://drive.google.com/file/d/1giaZY8oIBYVjnQTgk-jCa1DDu769GxTS/view?usp=sharing

5. Supported Formats for document upload on the UI:

1. PNG, JPG, PDF, JPEG.

ML Model Performance Matric

• Classification of Keys and Values:

• Model Training/Metrics/Usage Notebook: https://github.com/ML9-FORCES/extractify-doc/blob/main/Classifier_Model.ipynb

• Mapping of Keys with their respective Values:

• Model Training/Metrics/Usage Notebook: https://github.com/ML9-FORCES/extractify-doc/blob/main/Linking_Model.ipynb

• Wrapper For Integrating Both Models:

• Wrapper Implementation: https://github.com/ML9-FORCES/extractify-doc/blob/main/Model_Wrapper.ipynb

 wrapper.generate(Input_json)=> Output_json -

• Input_json ⇒ OCR Input-JSON
• Output_json ⇒ Output-JSON with Classification & linking

—-----------------------------------------------------------

• API:
• Wrapper Is Integrated into a REST API
• Endpoint

• /api

• Input - Input-OCR-JSON (as raw-json in       request body)
• Output - Output-JSON with Classification & Linking
• Method - POST/GET

ML-Presentation : https://github.com/ML9-FORCES/extractify-doc/blob/main/ML9FORCES.pdf