๐ซ CAMEL: The first and the best multi-agent framework. Finding the Scaling Law of Agents. https://www.camel-ai.org
Python agentai-societiesartificial-intelligencecommunicative-aicooperative-aideep-learninglarge-language-modelsmulti-agent-systemsnatural-language-processing
๐ฅ YouTube Video Q&A with CAMEL (Audio + Visual OCR)
This project enables users to input a YouTube URL and ask questions about the video's spoken content and on-screen text (e.g., PPT slides). It integrates:
- โ Audio transcription via Whisper
- ๐ผ๏ธ Visual OCR from extracted video frames
- ๐ง CAMEL AI multi-agent reasoning
- ๐ Streamlit interface for interaction
๐ Features
- Downloads YouTube videos via CAMEL toolkit
- Transcribes audio using OpenAI Whisper
- Extracts key video frames at regular intervals
- Performs OCR on frames to extract visible text
- Combines transcript and OCR data as knowledge
- Allows users to ask questions answered by CAMEL agent
๐ฆ Requirements
Install Python packages with specific versions:
pip install -r requirements.txt
๐งฐ Also install system dependencies:
Tesseract OCR:
- Windows: Download here
- macOS:
brew install tesseract - Ubuntu:
sudo apt-get install tesseract-ocr
FFmpeg (required by Whisper and for frame extraction):
- Windows: Download here (add
/binto PATH) - macOS:
brew install ffmpeg - Ubuntu:
sudo apt-get install ffmpeg
๐ ๏ธ Setup for Windows
If Tesseract is not found, add this in your Python code:
import pytesseract
pytesseract.pytesseract.tesseract_cmd = r"C:\Program Files\Tesseract-OCR\tesseract.exe"
Ensure the path is correct based on your installation.
๐งช Run the App
streamlit run app.py
Then, open the Streamlit interface in your browser.
๐ File Structure
app.py # Main Streamlit app
README.md # This file
requirements.txt # Python dependencies
๐ง Powered By
๐ฌ Example Usage
- Paste a YouTube link of a tech tutorial with slides.
- Ask: "What is the title of the slide shown at 5 minutes?"
- The app returns relevant text from OCR and transcript.