HEPHA: A Mixed-Initiative Image Labeling Tool for Specialized Domains

• Paper Link (arXiv): https://arxiv.org/abs/2503.03094

📌 Overview

HEPHA is a mixed-initiative tool that supports image labeling by eliciting labeling knowledge from domain experts.

This project is built with:

• Frontend: ReactJS, Vite
• Backend: ExpressJS, Flask

🚀 Setup Instructions

Pre-requisites:

• Git(with LFS installed)
• MongoDB
• Python 3.9
• NodeJS 19
• Tmux (for local setup)

Local Setup

1. Clone the repository

   git clone https://github.com/Neural-Symbolic-Image-Labeling/HEPHA.git
   cd HEPHA
   git lfs pull

2. Install dependencies
   This will install all required packages:

   make setup

3. Configure environment
   Create and modify your environment variables in .env file.

4. Launch the project
   The project runs in a tmux session for process management:

   make run

   To access the tmux session:

   tmux a  # Use Ctrl+B followed by D to detach

5. Stop the project

   make stop

Docker Setup

1. Clone the repository

   git clone https://github.com/Neural-Symbolic-Image-Labeling/HEPHA.git
   cd HEPHA
   git lfs pull

2. Configure environment
   Copy and modify the environment file:

   cp .env.example .env

3. Start containers
   Launch services in detached mode:

   docker compose up -d --build

4. Stop and clean up

   docker compose down

⚖️ License

This project is licensed under the Apache License 2.0 - see the LICENSE for details.

📬 Contact

Please email szhou20@uci.edu for any questions. Thanks!

Reference

@misc{zhou2025hephamixedinitiativeimagelabeling,
    title={HEPHA: A Mixed-Initiative Image Labeling Tool for Specialized Domains}, 
    author={Shiyuan Zhou and Bingxuan Li and Xiyuan Chen and Zhi Tu and Yifeng Wang and Yiwen Xiang and Tianyi Zhang},
    year={2025},
    eprint={2503.03094},
    archivePrefix={arXiv},
    primaryClass={cs.HC},
    url={https://arxiv.org/abs/2503.03094}, 
}