HighlighAI - An AI Agentic System for Video Summarizing🎥🚀

A Streamlit-based application for analyzing video content and extracting actionable insights using advanced AI models. Powered by Gemini 2.0 Flash Exp, this tool enables users to upload video files and receive detailed, user-friendly analysis.

---

Features

• AI-Powered Video Analysis: Leverages Gemini 2.0 Flash Exp for multimodal AI processing.
• Supplementary Web Research: Uses DuckDuckGo for additional context and insights.
• Streamlined User Interface: Upload video files, ask questions, and get actionable results effortlessly.
• Supports Popular Formats: Works with .mp4, .mov, and .avi video files.

---

Tech Stack

• Python: Core programming language.
• Streamlit: Interactive UI and deployment.
• Phidata Library: For building and managing AI agents.
• Google Generative AI: Video file processing and Gemini model integration.
• DuckDuckGo: Supplementary web research tool.

---

Prerequisites

• Python: Version 3.8 or higher.
• Google API Key: Add your Google API key to a .env file in the project directory: GOOGLE_API_KEY=your_api_key_here

---

Installation

1. Clone the Repository:
   git clone https://github.com/your-username/video-ai-summarizer.git
   cd video-ai-summarizer

2. Install Dependencies:
   Run the following command to install required Python packages:
   pip install streamlit phi duckduckgo-python python-dotenv google-generative-ai

---

Running the App

1. Start the Streamlit app:
   streamlit run app.py

2. Access the app in your web browser at:
   http://localhost:8501

---

How to Use

1. Upload a Video: Upload a file in .mp4, .mov, or .avi format.
2. Ask a Question: Provide specific queries about the video content.
3. Analyze: Click the "Analyze Video" button to start processing.
4. View Results: Detailed insights will be displayed in the app.

---

Code Highlights

Agent Initialization:

def initialize_agent():
    return Agent(
        name="Video AI Summarizer",
        model=Gemini(id="gemini-2.0-flash-exp"),
        tools=[DuckDuckGo()],
        markdown=True,
    )

Uploading and processing the video file

processed_video = upload_file(video_path)
while processed_video.state.name == "PROCESSING":
    time.sleep(1)
    processed_video = get_file(processed_video.name)

Generating the prompt and AI analysis

analysis_prompt = (
    f"""
    Analyze the uploaded video for content and context.
    Respond to the following query using video insights and supplementary web research:
    {user_query}

    Provide a detailed, user-friendly, and actionable response.
    """
)

response = multimodal_Agent.run(analysis_prompt, videos=[processed_video])