Back to Home

SkySentry

A low-latency object detection and tracking system built in 48 hours with distributed services for capture, inference, tracking, and live analytics.

GoTypeScriptPythonReactNext.jsWebSockets

Overview

SkySentry is a real-time computer vision platform designed to stream, detect, track, and visualize moving objects with end-to-end latency constraints.

What was built

  • A Go WebSocket backend for ingesting and distributing live stream events.
  • A Python Ultralytics inference service for object detection on incoming frames.
  • TypeScript orchestration and tracking logic with persistent IDs and smoothing.
  • Dashboard components for live metrics, trajectory views, and operator monitoring.

Impact

  • Built in 48 hours with roughly 69 commits and about 15k LOC.
  • Designed for sub-100ms response behavior with real-time frame throughput up to 60 FPS.
  • Demonstrated a production-style multi-service architecture under hackathon constraints.

Media

SkySentry overview

SkySentry object map

SkySentry tracking

Links