📄 DocuFlow

DocuFlow is an AI-based document processing and Excel automation system that converts user-uploaded documents into clean, structured Excel records through a deterministic and explainable backend pipeline.

The system is designed to be mentor-safe, academically defensible, and practically useful, focusing on real-world document workflows rather than black-box automation.

---

✨ Key Features

• 📂 Single & multi-file document upload
• 🧭 User-selected document type (Receipts, Invoices, ID Cards)
• 🖼️ Image preprocessing using OpenCV
• 🔤 OCR with spatial text extraction (Tesseract)
• 📐 Layout-aware text grouping & analysis
• 🧠 Rule-based field extraction (deterministic)
• ✅ Validation & confidence scoring
• 📦 Batch processing with per-file error isolation
• 📊 Automatic Excel (.xlsx) generation
• ⬇️ Secure Excel download endpoint
• 🖥️ Clean frontend with progressive disclosure UI

---

🧠 System Philosophy

DocuFlow is intentionally designed to be:

• Deterministic – no black-box ML decisions
• Explainable – every processing step is inspectable
• Modular – each pipeline stage is independent
• Realistic – mirrors real-world document processing systems

This makes the system ideal for academic evaluation, project demos, and interview discussions.

---

🏗️ High-Level Architecture

Upload → Preprocessing → OCR → Layout Analysis
     → Field Extraction → Validation → Excel Generation
     → API Response → Frontend Download

---

📁 Project Structure

docuflow-backend/
├── app/
│   ├── main.py
│   ├── api/
│   │   ├── routes.py
│   │   ├── upload.py
│   │   └── download.py
│   └── core/
│       ├── storage.py
│       ├── preprocessing.py
│       ├── ocr.py
│       ├── layout.py
│       ├── extraction.py
│       ├── validation.py
│       └── excel.py
├── uploads/
├── processed/
├── exports/
├── index.html   # Frontend
├── requirements.txt
└── README.md

---

⚙️ Tech Stack

Backend

• FastAPI – API framework
• OpenCV – image preprocessing
• Tesseract OCR – text extraction
• Pillow – image handling
• openpyxl – Excel generation

Frontend

• HTML, CSS, Vanilla JavaScript
• Single-page application (no framework)

---

🚀 Getting Started

1️⃣ Clone the Repository

git clone https://github.com/your-username/docuflow.git
cd docuflow-backend

2️⃣ Create Virtual Environment

python3 -m venv venv
source venv/bin/activate

3️⃣ Install Dependencies

python3 -m pip install -r requirements.txt

4️⃣ Install Tesseract OCR

macOS

brew install tesseract

Verify:

tesseract --version

---

▶️ Run the Backend

python3 -m uvicorn app.main:app --reload

Access:

• API Docs: http://127.0.0.1:8000/docs

---

🖥️ Run the Frontend

• Open index.html directly in your browser
• Ensure backend is running at http://127.0.0.1:8000

---

📤 Usage Flow

1. Select document type (Receipt / Invoice / ID Card)
2. Upload one or more documents
3. Click Process Documents
4. View processing summary and warnings
5. Download generated Excel file

Each document corresponds to one row in the Excel output.

---

📊 Output Example (Excel)

Date	Merchant	Amount	Confidence	Warnings
05-01-2025	Swiggy	326	0.85	—

---

🧪 Batch Processing Behavior

• Each file is processed independently
• Failure in one document does not stop others
• Partial extraction is allowed with warnings
• Excel is generated from successfully processed files only

---

📌 One-Line Summary

DocuFlow processes user-selected document types through a structured pipeline involving image preprocessing, OCR, layout analysis, validation, and automatic Excel generation.

---

👤 Author

Developed as an academic and practical project for demonstrating document intelligence pipelines.

---

If you’re reviewing this project:

• Check /docs for API clarity
• Upload sample documents
• Inspect generated Excel outputs

Thank you for exploring DocuFlow 🚀