🎥 Semantic Video Retrieval System

A production-ready, multimodal video retrieval system that matches videos to natural language queries using powerful vision-language models (BLIP) and Vector Databases (ChromaDB).

This version includes a clean UI, speed optimizations, and robust error handling.

---

✨ Key Features

• Natural Language Search: "Find the clip where a dog is running on grass."
• Optimized Performance: Adjustable frame sampling (doesn't process every single frame, making it 5-10x faster).
• Video Summary Generation: Automatically captions video content.
• Clean UI: Tabbed interface for Search, Upload, and Library management.
• Caching: Models load once, preventing slow reloads.

---

🧠 Tech Stack

Component	Technology
Language	Python
Vision Model	BLIP (Salesforce)
Vector DB	ChromaDB
Audio Processing	ffmpeg, pydub
Frame Extraction	OpenCV
UI / Deployment	Streamlit

---

🛠️ Installation

1. Clone the repository (or download the files).

2. Create a virtual environment (recommended):

   python -m venv venv
   
   # Windows
   venv\Scripts\activate
   
   # Mac/Linux
   source venv/bin/activate

3. Install dependencies

   pip install -r requirements.txt
   
   #start the demo
   
   streamlit run run.py