WM 2025 Final Project Group 16

• 資工三 詹挹辰 B11902057
• 資工三 巫俋霆 B11902051
• 資工三 施廣霖 B11902072

DEMO SITE: https://wm2025.iceylemon.net/

Please note that the demo site is hosted on a server without GPU acceleration, so the response time may be slow. (Live demo on 6/6 had the site hosted on laptop with GPU.) For faster performance, please run the system locally with GPU.

The main components of the system are as follows (only the important files are listed):

• textbook_retrieval_system/: Contains the core implementation of the retrieval system.

  • frontend/: Contains the frontend implementation for the system. We used React+TypeScript+Vite for the frontend.
    • src/: Contains the source code for the frontend.
      • App.tsx: The main React application file that renders the frontend.
      • App.css: The main CSS file for styling the frontend.
    • index.html: The main HTML file for the frontend.
    • package.json: Lists the dependencies and scripts for the frontend.
    • vite.config.ts/: Configuration file for Vite, the build tool used for the frontend.
  • backend/: Contains the backend implementation for the system. We used Flask server for our backend.
    • embeddings_cache/: Directory for storing computed embeddings.
    • parsed_output/: Directory for storing parsed output from textbooks.
    • app.py: The main Flask application file that serves the API endpoints. The retrieval model is also initialized here.
    • ocr.py: Script for performing OCR on user query image to extract text.
    • utils.py: Utility functions used in the backend.
    • requirements.txt: Lists the Python dependencies required to run the backend.

• scripts/: Contains auxiliary scripts for downloading models and other resources.

• demo/: Contains the demo problems and images used for testing the retrieval system.

  • problems.txt: A text file containing example problems for testing the retrieval system (an example problem each line).
  • images/: Directory containing images of the problems for testing the retrieval system.

• README.md: This file, providing an overview and instructions for the system.

To get run the retrieval system, follow these steps:

We will be working on textbook_retrieval_system/ directory, so please make sure you are in the correct directory.

cd textbook_retrieval_system

Install Node.js and npm (Node Package Manager) if you haven't already. You can download them from Node.js official website.

Then, install the required packages and start the development server:

cd frontend
npm install
npm run dev

We use conda for environment management, please ensure you have it installed. If you don't have conda, you can install it from Anaconda's official website.

Create a virtual environment and install the required packages:

cd backend
conda create -n mineru 'python=3.12' -y
conda activate mineru
python -m pip install -r requirements.txt

# Download the models needed by MinerU
python -m pip install huggingface_hub
wget https://github.com/opendatalab/MinerU/raw/master/scripts/download_models_hf.py -O download_models_hf.py
python download_models_hf.py