VidQuery - Ask Any Question From a YouTube Video!

A lightweight Streamlit app that allows users to ask questions about a YouTube video and get answers using the video's transcript. This is a Retrieval-Augmented Generation (RAG) based application leveraging embeddings for semantic search and natural language understanding.

---

Demo :

VidQuery.mp4

1. How It Works / Model Overview

• Input: YouTube video link + User query
• Processing:
  • Video transcript is fetched and split into chunks
  • Chunks are embedded using sentence-transformers/all-MiniLM-L6-v2
  • Semantic search performed using Chroma vector store
• Output: Answer to the user query using the video transcript

---

2. App Features

• Input a YouTube video link
• Ask any question related to the video
• Get a concise answer using the video's transcript
• Powered by RAG and embeddings for accurate results

---

3. Example Inputs

• YouTube Link: https://www.youtube.com/watch?v=P0N6aMczu78
• Question: "Who was Admiral Richard Byrd Jr and what did he discover in Antarctica?"
• Output: Relevant answer extracted from the transcript

---

4. Project Structure

env

│ .gitignore

│ app.py

│ LICENSE

│ README.md

│ requirements.txt

│

├───.streamlit

│ secrets.toml │

├───assets

│ gokulaix_logo.png

│
├───backend

│ │ backend.py

│ │
│ └───__pycache__

│ backend.cpython-312.pyc │ ├───chroma_db

│ │ chroma.sqlite3

│ │
│ └─── │ data_level0.bin

│ header.bin

│ index_metadata.pickle

│ length.bin

│ link_lists.bin │

├───chrome_integration

│ init.py │

├───data

│ │ transcripts.py

│ │ init.py

│ │ │ └───__pycache__

│ transcripts.cpython-312.pyc

│ init.cpython-312.pyc

│ ├───database

│ │ chromaa.py

│ │ init.py

│ │ │ └───__pycache__

│ chromaa.cpython-312.pyc

│ init.cpython-312.pyc

│ ├───main

│ │ init.py

│ │ │ └───__pycache__

│ app.cpython-312.pyc

│ init.cpython-312.pyc

│ ├───prompts

│ │ prompt.py

│ │ init.py

│ │ │ └───__pycache__

│ prompt.cpython-312.pyc

│ init.cpython-312.pyc

│ ├───retrieval

│ │ Retrieve.py

│ │ init.py

│ │ │ └───__pycache__

│ Retrieve.cpython-312.pyc

│ init.cpython-312.pyc

│ └───text_split │ text.py │ └───__pycache__ text.cpython-312.pyc

---

5. How to Run Locally

1. First, fork this repository to your own GitHub account

2. Clone the repository:

First fork and clone the backend repo

fork this repo
https://github.com/GokulAIx/VidQuery-Backend.git
clone it
git clone https://github.com/<your-username>/VidQuery-Backend.git

RUN THE BACKEND FILE , with:
python backend.py

Now fork this current Repo and clone it

git clone https://github.com/<your-username>/VidQuery.git
cd rag_app
pip install -r requirements.txt
streamlit run app.py

10. Contact Info

• Name: P Gokul Sree Chandra
• Email: polavarapugokul@gmail.com
• LinkedIn: Gokul Sree Chandra
• Portfolio: GokulAIx

11. License

This project is licensed under the MIT License.