AI Image Caption Generator

An AI-powered image captioning application that generates natural language descriptions for images using a pretrained BLIP (Bootstrapped Language-Image Pretraining) model from Hugging Face.

This project provides both:

• A Command-Line Interface (CLI) for quick caption generation
• A Web-based interface using Gradio for interactive usage

---

Features

• Automatically generates captions for images
• Uses a pretrained Vision–Language Transformer (BLIP)
• Supports both CLI and Web App usage
• Simple and readable codebase, suitable for learning and experimentation

---

Model Details

• Model Name: Salesforce/blip-image-captioning-base
• Task: Image Captioning
• Framework: PyTorch
• Library: Hugging Face Transformers

BLIP models are trained jointly on images and text, allowing them to understand visual content and generate accurate natural-language descriptions.

---

Project Structure

├── CLI-App.py # Command-line image captioning script

├── web-app.py # Gradio-based web application

├── requirements.txt # Python dependencies

├── cat.jpeg # Sample image (replace with your own)

└── README.md

---

Installation

1️⃣ Clone the Repository

git clone https://github.com/adithya-umesh/ai-image-captioner.git
cd ai-image-captioner

2️⃣ (Optional but Recommended) Create a Virtual Environment

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

3️⃣ Install Dependencies

pip install -r requirements.txt

Note: The first run will download the BLIP model weights, which may take a few minutes.

CLI Application Usage: Run the CLI app

python CLI-App.py

How it works:

Loads an image (cat.jpeg by default)

Converts it to RGB format

Processes it using the BLIP processor

Generates and prints a caption in the terminal

To caption a different image, edit the following line in CLI-App.py: img_path = "your_image.jpg"

Web Application Usage:

Run the web app

python web-app.py

Opens a Gradio interface in your browser Upload any image Instantly receive an AI-generated caption

---

🧩 Technologies Used:

• Python
• PyTorch
• Hugging Face Transformers
• Gradio
• Pillow (PIL)
• NumPy

📌 Example Output

• Input Image: A cat sitting on a sofa
• Generated Caption: A photo of a cat sitting on a couch in a living room

🔮 Future Improvements

• Upgrade to blip-image-captioning-large for higher-quality captions
• Add GPU (CUDA) support
• Batch image captioning
• Deploy as a public web service (Hugging Face Spaces)
• Improve UI and add caption confidence scoring

📜 License

This project is intended for educational and experimental purposes.

Model rights belong to Salesforce Research and Hugging Face.

🙌 Acknowledgements

Salesforce Research for the BLIP model

Hugging Face for Transformers and model hosting

Gradio for rapid web UI development

---