https://github.com/devanshu-17/logithon-astrea

Team Astrea project for the Logithon 2024 hackathon
https://github.com/devanshu-17/logithon-astrea

Last synced: 7 days ago
JSON representation

Team Astrea project for the Logithon 2024 hackathon

• Host: GitHub
• URL: https://github.com/devanshu-17/logithon-astrea
• Owner: Devanshu-17
• License: mit
• Created: 2024-04-19T09:01:41.000Z (7 months ago)
• Default Branch: main
• Last Pushed: 2024-06-21T23:00:32.000Z (5 months ago)
• Last Synced: 2024-06-22T13:43:28.183Z (5 months ago)
• Language: CSS
• Size: 10.5 MB
• Stars: 0
• Watchers: 1
• Forks: 0
• Open Issues: 0
• Metadata Files:
  • Readme: README.md
  • License: LICENSE

Awesome Lists containing this project

README

        
# Logithon-Astrea

Team Astrea project for the Logithon 2024 hackathon

### Workflow

- [x] Convert the PDF file to image

- [x] Use OCR and a Visual LLM with prompt to extract data from the images

- [x] Define a Pydantic base model with necessary fields for JSON output

- [x] Send OCR text and image description to Mixtral LLM with a prompt to generate JSON output based on the Pydantic base model

- [x] Create a user feedback button to verify data extraction accuracy

- [x] If data extraction is incorrect:

  - [x] Prompt user for reasons for inaccuracy

  - [x] Send reason, prompt, and data to LLM to generate output again

  - [ ] Save input and response in a vector database with a score of 0

- [x] If data extraction is correct:

  - [x] Convert JSON to CSV and display as an editable table

  - [ ] Save input and response in a vector database with a score of 1

- [x] Allow user to make final modifications to the data

- [x] Export data in JSON or CSV format

--------------------------------------------------------------------------------------------------------------

- [x] AFTER WORKFLOW IS IMPLEMENTED, CREATE BACKEND PIPELINE

- [x] CREATE FRONTEND