(ML) YouTube Caption Summarization Program

GitHub Link

📢 Demo Video

📖 Presentation Material

https://www.miricanvas.com/v/124idbc

⭐️ Project Overview

• A service that takes a YouTube video link, extracts the connected captions, summarizes them, and answers questions based on the summarized content.
• Utilized Summarization Model and Answering Model and fine-tuned the results to display input-output forms in the Front and Back end.
• Ranked in the top 30 on the KorQuAD 1.0 leaderboard.
• Awarded 3rd place at the department's presentation on Department Day in 2023.

👬 Team Composition

ML Team (4 members) - Front (1), Back (1)

🔨 Role and Function

• Model selection and study of related papers
• UI design for the models

💡 Reflections

• By developing a service that may be useful on the widely-used video platform YouTube and fine-tuning existing models, I gained confidence that with proper fine-tuning of models, I can develop AI-powered innovative features for existing services or domains that already utilize AI.