🎥 Video to Narrative (Crime Scene Caption + Summary AI)

A Flask-based AI application that:

• Extracts frames from uploaded videos
• Captions frames using ViT-GPT2 (transformers)
• Transcribes audio using Whisper
• Summarizes visuals + audio using Groq LLM (llama3-8b-8192)

📁 Project Structure

video_to_narrative/
├── app.py                 # Main Flask backend
├── logs/
│   └── app.log            # Runtime logs
├── uploads/               # Temporary video + frames (ignored in Git)
├── audio/                 # Temporary audio files (ignored in Git)
├── templates/
│   └── index.html         # Frontend upload page
├── v2t/                   # Virtual environment (ignored in Git)
├── .gitignore
├── requirements.txt
├── README.md

✅ Features

• ✅ Upload crime/dashcam/bodycam video
• ✅ Frame-by-frame captioning (ViT-GPT2)
• ✅ Audio transcription (Whisper)
• ✅ Groq-powered dual summary:
  • General Summary
  • Law Enforcement Report

🚀 Setup Instructions

1. Clone the repo

git clone https://github.com/ChaituRajSagar/video_to_narrative.git

cd video_to_narrative

2. Create Python environment

python -m venv v2t

v2t\Scripts\activate # On Windows

Or

source v2t/bin/activate # On macOS/Linux

3. Install dependencies

pip install -r requirements.txt

4. Set your Groq API Key

Create a .env file (or set directly in app.py):

GROQ_API_KEY=your_groq_api_key_here

▶️ Run the App

python app.py

Visit http://127.0.0.1:8009

🛑 Notes

• Uses CPU by default (Torch auto-detects)

• Auto-cleanup of:

  • Extracted frames
  • Uploaded video
  • Audio files

• Logs available in logs/app.log