Update README.md

README.md

@@ -0,0 +1,54 @@
+---
+title: Invoice Information Extractor
+emoji: 📄
+colorFrom: blue
+colorTo: green
+sdk: docker
+pinned: false
+license: mit
+app_port: 7860
+---
+
+# Invoice Information Extractor
+
+AI-powered REST API for extracting structured information from Indian tractor invoices.
+
+## Features
+
+- 🔍 **Text Extraction:** Dealer name, model name, horse power, asset cost
+- 🖊️ **Signature Detection:** Automatic signature detection with bounding boxes
+- 📋 **Stamp Detection:** Stamp detection with confidence scores
+- 🎯 **Validation:** Smart validation and error correction
+- ⚡ **Fast Processing:** ~3-5 seconds per invoice
+
+## Models
+
+- **YOLO**: Custom-trained signature/stamp detector
+- **Qwen2.5-VL-7B**: Vision-language model for text extraction (4-bit quantized)
+
+## API Documentation
+
+Visit `/docs` for interactive API documentation.
+
+## Quick Start
+
+### Python
+```python
+import requests
+
+url = "https://YOUR_USERNAME-invoice-extractor.hf.space/extract"
+files = {"file": open("invoice.png", "rb")}
+
+response = requests.post(url, files=files)
+print(response.json())
+```
+
+### cURL
+```bash
+curl -X POST "https://YOUR_USERNAME-invoice-extractor.hf.space/extract" \
+  -F "file=@invoice.png"
+```
+
+## Hardware
+
+Requires GPU: T4 minimum (8GB VRAM recommended)