GenAI Image Captioning

This project demonstrates an AI-powered image captioning application built with Hugging Face Transformers and Gradio.
It uses the BLIP (Bootstrapping Language-Image Pretraining) model to generate natural language captions for uploaded images.

✨ Features #

Upload any image and generate descriptive captions automatically.
Powered by the Salesforce BLIP image captioning model.
Clean and interactive Gradio web interface.
Optional text prompt support for guided captioning.
Extensible for batch processing (media, SEO, accessibility).

📂 Project Structure #

image-captioning/
├── image_captioning.py     # Gradio app
├── image_cap.py            # Simple CLI version
├── requirements.txt        # Dependencies
├── captions.txt            # Example output (URL: caption pairs)
└── README.md               # Project documentation

⚙️ Installation #

Clone the repository:

git clone https://github.com/<your-username>/image-captioning-ai.git
cd image-captioning-ai

Create a virtual environment (recommended):

python -m venv .venv
source .venv/bin/activate     # On Windows: .venv\Scripts\activate

Install dependencies:

pip install -r requirements.txt

🚀 Usage #

Run the Gradio app #

python image_captioning.py

A local URL will appear (e.g. http://127.0.0.1:7860).
Open it in your browser.
Upload an image → AI generates a caption.

Example #

Input:
IBM HQ

Output:

"a large glass building with trees in front of it"

📊 Business Use Case #

This application can be adapted for news and media agencies:

Automates image captioning for hundreds of daily articles.
Saves editorial time while ensuring consistency.
Provides alt-text for accessibility.
Improves SEO performance by supplying meaningful image descriptions.

☁️ Deployment #

You can deploy this app to the cloud:

IBM Code Engine – containerized deployment with a free Kubernetes cluster.
Hugging Face Spaces – quick demo hosting with Gradio support.
Docker – containerize locally and push to your preferred registry.

🛠️ Tech Stack #

Python 3.9+
Transformers (Hugging Face)
Gradio
PIL (Pillow)
PyTorch

📌 Next Steps #

Add batch captioning (folder → captions.csv).
Support for captions via webpage URL (scrape and caption).
Export captions directly into CMS for media/news workflows.
Integrate speech-to-text (STT) or text-to-speech (TTS) for accessibility.

👤 Author #

Developed as part of IBM Generative AI learning modules.
Author: Ahmad Tawil
Guided project by Sina Nazeri, IBM Developer Skills Network