Computer Vision · Video Intelligence
PublicSmart Video Analytics
All-in-one video intelligence system detecting people, vehicles, smoke, fire, and violence in a single Streamlit pipeline.
Achievement
Built an integrated computer vision pipeline with live overlays, scenario detection, event logs, and an interactive analytics dashboard.
Media gallery
Overview
Smart Video Analytics combines multiple detection models in one application. It analyzes uploaded video streams, draws real-time overlays, classifies scenarios, and surfaces results through a Streamlit interface and summary dashboard. Red bounding boxes indicate violence-related detections, while other overlays cover people, vehicles, smoke, and fire.
Problem
Manual video monitoring is difficult to scale and can miss critical events across different scenarios.
Solution
Unify person, vehicle, smoke, fire, and violence detection into one workflow with configurable thresholds, frame-level overlays, structured logging, and dashboard reporting.
My role
Model integration, multi-scenario video pipeline, overlay logic, Streamlit UI, and dashboard design.
Key features
- All models in one Streamlit app
- Person and vehicle detection
- Smoke and fire detection
- Violence detection with red overlays
- Scenario classification and event logging
- Interactive analytics dashboard