DEPLOY_GUIDE.md

Deployment Guide for CaptionIQ

This document explains how to deploy your Image Captioning application to the public.

1. Hugging Face Spaces (Streamlit SDK)

Hugging Face Spaces allows you to quickly host Machine Learning demos. I have prepared an automated script to deploy it.

Automatic Method (Recommended)

You can deploy directly from your local terminal using the script provided:

python3 deploy_hf.py --token YOUR_HUGGING_FACE_TOKEN --repo-name CaptionIQ

Note: Make sure your token has Write permissions. You can generate one at huggingface.co/settings/tokens.

The script will:

1. Automatically create a Streamlit Space named CaptionIQ on your Hugging Face account.
2. Filter out unnecessary large files (like dataset images in data/Images).
3. Upload your models and code.
4. Set up Git LFS (Large File Storage) for your .h5 model files automatically.

2. Streamlit Community Cloud

Streamlit provides a free community cloud that integrates directly with GitHub. If you prefer to deploy there:

1. Push your code to GitHub:

   • Create a GitHub repository for your project.
   • Commit and push your code. Because the .h5 models are large, ensure you have Git LFS installed (git lfs install) before pushing, or exclude them and load models from an external URL. I have added .gitattributes to track .h5 files with LFS.

2. Deploy via Streamlit Cloud:

   • Go to share.streamlit.io.
   • Click "New app".
   • Select your GitHub repository, choose the main branch, and specify app.py as your Main file path.
   • Click Deploy.

Note on Free Tiers: Streamlit Community Cloud has a 1GB memory limit which might be tight when loading two large VGG models simultaneously. Hugging Face Spaces (Free Tier) typically offers 16GB of RAM, making it a better fit for this project!