README.md

metadata

title: Singtel Bill Scanner
emoji: 📱
colorFrom: red
colorTo: red
sdk: streamlit
sdk_version: 1.28.0
app_file: app.py
pinned: false
tags:
  - computer-vision
  - ocr
  - trocr
  - bill-processing
  - singtel
  - document-ai

Singtel Bill Scanner 📱💡

An AI-powered optical character recognition (OCR) system specifically designed for processing Singtel telecommunications bills.

🚀 Try the Live Demo

Upload your Singtel bill image and get instant results!

Features

• 🔍 Text Extraction: Uses Microsoft TrOCR for accurate OCR
• 📊 Bill Parsing: Extracts amounts, dates, account numbers
• ⚡ Fast Processing: Real-time results
• 🎯 Singtel Optimized: Tailored for Singtel bill formats
• 🔒 Privacy First: Images processed locally, not stored

How to Use

1. 📸 Take a clear photo of your Singtel bill
2. 📤 Upload the image using the interface above
3. 🔍 Click "Extract Information"
4. 📊 View the extracted data

Technical Details

• Model: Microsoft TrOCR (microsoft/trocr-base-handwritten)
• Framework: Streamlit + Hugging Face Transformers
• Processing: ~3-5 seconds per image
• Accuracy: High for clear, well-lit images

Local Installation

git clone https://huggingface.co/spaces/Cosmo125/Singtel_Bill_Scanner
cd Singtel_Bill_Scanner
pip install -r requirements.txt
streamlit run app.py

API Usage

from transformers import pipeline
from PIL import Image

# Load model
pipe = pipeline("image-to-text", model="microsoft/trocr-base-handwritten")

# Process image
image = Image.open("bill.jpg")
result = pipe(image)
text = result[0]['generated_text']

Team members Ninette Zi Xuan Sai Kamal

Created for the Singtel community with ❤️