Upload Pipeline

This section describes the upload pipeline of the Fluidsdata Digitization and OCR application.

Contents:

• File Handling
  • System Architecture
  • File Upload and Processing Workflow
  • Azure Blob Storage Integration
  • LazyImageManager
  • User Interface Integration
  • Integration with Processing Components
  • File Format Support
  • Integration with Processing Pipeline
• OCR Process
  • Why OCR?
  • OCR Models
  • OCR Model Call
  • Working with OCR Results
• Table Reconstruction
  • Overview
  • Computer Vision Models
  • Table Detection Workflow
  • Word-to-Table Mapping
  • Data Reconstruction Process
  • Error Handling and Edge Cases
  • Performance Optimization
  • Integration with OCR Pipeline
  • Future Enhancements

This pipeline is responsible for handling the upload and processing (OCR and table reconstruction) of various document formats, including PDFs, images, and Excel files.

Currently, processing documents through OCR is done via the file upload page, which allows users to view local and cloud files and select them for processing.