🧠 OCR Keyword Detection API

This project is a FastAPI-based OCR (Optical Character Recognition) service that detects whether a specific keyword appears in an uploaded image.
It uses Tesseract OCR, OpenCV, and RapidFuzz to read and analyze text content from images with automatic rotation and preprocessing for better accuracy.

---

🚀 Features

• 🖼️ Image upload API with keyword checking
• 🔄 Automatic image rotation (fixes portrait/landscape orientation)
• 🎨 Multiple OCR preprocessing strategies (contrast enhancement, grayscale, etc.)
• 📊 String similarity matching using RapidFuzz
• 🧾 Outputs OCR results and confidence scores
• ⚡ Built with FastAPI + Uvicorn for high performance

---

📁 Project Structure

├── main.py               # Main FastAPI application
├── gambar/               # Folder where uploaded images are saved
├── hasil/                # Folder for OCR output and processed images
└── requirements.txt      # Python dependencies

---

🧩 Dependencies

This project uses the following main libraries:

• fastapi
• uvicorn
• opencv-python
• pytesseract
• pillow
• numpy
• rapidfuzz

---

⚙️ Installation

1. Clone this repository

   git clone https://github.com/yourusername/ocr-api.git
   cd ocr-api

2. Create a virtual environment

   python -m venv venv
   source venv/bin/activate  # On Linux/Mac
   venv\Scripts\activate     # On Windows

3. Install dependencies

   pip install -r requirements.txt

4. Install Tesseract OCR

   • Ubuntu/Debian

     sudo apt update
     sudo apt install tesseract-ocr

   • Windows
     • Download and install from: https://github.com/UB-Mannheim/tesseract/wiki

---

▶️ Running the Server

Start the FastAPI app using Uvicorn:

uvicorn main:app --host 0.0.0.0 --port 8124

or simply run:

python main.py

The server will start on:
👉 http://localhost:8124

---

🧠 API Usage

Endpoint: /ocr

Method: POST
Content-Type: multipart/form-data

Form Data:

Field	Type	Description
file	File	The image file to scan
keyword	String	The text keyword to search for

Example using curl

curl -X POST "http://localhost:8124/ocr" \
  -F "[email protected]" \
  -F "keyword=DIGITAL"

Example Response

{
  "found": true,
  "match": "DIGITAL",
  "score": 91
}

---

🧮 Processing Steps

1. Fix image orientation using EXIF metadata
2. Run raw OCR directly
3. Apply preprocessing techniques (grayscale, contrast adjustment)
4. If no match, rotate the image 180° and retry
5. Compare OCR results against the target keyword using fuzzy matching

---

📂 Output Files

All results are saved in the hasil/ directory:

• step0_raw.jpg → The processed image
• hasil_ocr.txt → The OCR extracted text

---

🧑‍💻 Example Workflow

1. Upload an image via /ocr endpoint
2. The server processes it, runs OCR, and checks similarity
3. Returns whether the keyword was found and the similarity score

---

🛠️ Customization

You can adjust:

• OCR configuration in ocr_raw() and ocr_preprocessed()
• Similarity threshold in check_similarity() (default: 85%)
• Output directory names for results and uploads

---

📜 License

This project is open-source under the MIT License.