---
title: AI Video Analysis System
emoji: 🎬
colorFrom: green
colorTo: blue
sdk: gradio
sdk_version: 6.12.0
python_version: 3.11
app_file: app.py
pinned: false
license: mit
---

# 🎬 Agentic AI Video Analysis System — Gradio Edition

An end-to-end video analysis pipeline powered by YOLOv8, Groq LLaMA-3, and Gradio for Hugging Face Spaces.

## What it does
- Extracts frames from an uploaded video using OpenCV
- Detects objects in each frame using YOLOv8 nano
- Generates a natural-language summary using Groq LLaMA-3 or mock mode
- Produces an analysis report with insights, risk flags, and recommendations
- Lets users download JSON and text reports

## How to use
1. Upload a video in MP4, AVI, MOV, or MKV format
2. Adjust frame extraction and detection settings
3. Click Analyze Video
4. Review extracted frames, annotated frames, summaries, and reports

## API Key Setup
1. Get a Groq API key
2. Open Space Settings → Secrets
3. Add:
   - Name: GROQ_API_KEY
   - Value: your key
4. Restart the Space

Without a key, the app still works in mock mode.