table_extraction README.md

document_intelligence/table_extractor/README.md

@@ -0,0 +1,109 @@
+# Table Detection OCR Service
+
+This service provides a FastAPI-based API for detecting tables in PDF documents. It uses a YOLO-based object detection model to locate tables in images and return cropped table images for further processing.
+
+## Features
+
+- **Table Detection**: Automatically detects tables in PDF documents using the YOLO model. Outputs them in a human-readable format with appropriate visual cues.
+- **Image Cropping**: Crops out detected tables from the PDF pages and returns them as base64-encoded PNG images.
+- **FastAPI-based**: Easy to integrate into existing workflows via API.
+
+
+## Installation
+
+1. **Clone the repository**:
+   ```bash
+   git clone https://github.com/yourusername/table-detection-ocr.git
+   cd table-detection-ocr
+   ```
+
+2. **Set up a virtual environment** (optional but recommended):
+   ```bash
+   python3 -m venv venv
+   source venv/bin/activate  # On Windows, use `venv\Scripts\activate`
+   ```
+
+3. **Install dependencies**:
+   ```bash
+   pip install -r requirements.txt
+   ```
+
+4. **Run the service**:
+   ```bash
+   uvicorn main:app --reload --port 8002
+   ```
+
+The service will be available at `http://127.0.0.1:8002`.
+
+## API Endpoints
+
+### **POST /predict**
+Upload a PDF file and receive base64-encoded images of detected tables in JSON format.
+
+#### Request
+- **Method**: POST
+- **Endpoint**: `/predict`
+- **Parameters**: 
+  - `file`: The PDF file to be uploaded.
+  - `padding` (optional): Integer value for padding around detected tables (default is 10 pixels).
+
+#### Example CURL Command
+
+```bash
+curl -X 'POST' \
+  'http://127.0.0.1:8002/predict' \
+  -H 'accept: application/json' \
+  -H 'Content-Type: multipart/form-data' \
+  -F '[email protected]' \
+  -F 'padding=10'
+```
+
+#### Response
+The response will include the detected tables as base64-encoded PNG images in a JSON structure:
+
+```json
+{
+  "0": [
+    {
+      "table_text": "Extracted table text here...",
+      "score": 9
+    },
+    {
+      "table_text": "Another table's text...",
+      "score": 8
+    }
+  ]
+}
+```
+
+### **GET /health**
+Check the health status of the service.
+
+#### Request
+- **Method**: GET
+- **Endpoint**: `/health`
+
+#### Response
+```json
+{
+  "status": "OK"
+}
+```
+
+## Usage Example
+
+1. Upload a PDF using the `/predict` endpoint.
+2. The service detects tables, crops them from the images, and returns them as base64-encoded PNG images in JSON format.
+3. You can process the images or decode the base64 strings to display them.
+
+## Environment Variables
+
+- **ASKEM_DOC_AI_API_KEY**: This environment variable is required for accessing the OpenAI API used for further table processing. Be sure to set it in your environment.
+
+```bash
+export ASKEM_DOC_AI_API_KEY="your-openai-api-key"
+```
+
+## License
+
+This project is licensed under the [Apache License 2.0](LICENSE).