AI Object Tracking
YOLOv8n + DeepSort delivers persistent, multi-class tracking with stable track IDs across frames — drones, persons, vehicles.
From Saudi Engineers to the World
عــــــــقــــــــاب
Autonomous AI-Powered Object Tracking Drone for Smart Surveillance
Edge AI · Real-time tracking · Identification Friend or Foe (IFF) · Built for the field.
01 / About the Project
An elite autonomous drone system, engineered end-to-end.
Eqaab is a graduation capstone fusing autonomous flight, edge AI vision, IFF classification, and a real-time ground control station — all designed around a deliberate, metadata-only architecture for production-grade reliability.
Smart Surveillance
Continuous, autonomous patrol of restricted zones with structured, low-bandwidth alerts — built for the field, not the lab.
AI Tracking
On-board YOLO inference with persistent multi-object tracking — every target keeps a track ID across frames.
Autonomous Drone
Mission planning, geofencing, and PX4 failsafes — with hardware-level RC override always available to the pilot.
02 / Capabilities
Four pillars, one autonomous system.
Every capability was chosen for resilience over flash — the system stays useful when the network drops, the GPS shakes, and the operator looks away.
Autonomous Navigation
Mission waypoints, geofencing, and PX4 autopilot deliver hands-off patrols with deterministic safety failsafes.
Drone Interception
IFF classification flags hostile UAVs in real time — friendly drones authenticate; unknowns escalate to HIGH threat.
Real-time Monitoring
Structured telemetry over WebSocket — 2 Hz position updates, event-driven detections, and a live mission timeline.
03 / System Architecture
From sensor edge to operator screen.
A three-tier pipeline: detect at the edge, process in the cloud, decide at the dashboard. Structured JSON the whole way — no video streams, no surprises.
Drone
Pi 5 + Pixhawk · Edge
AI Model
YOLO + DeepSort + IFF
Cloud
FastAPI · WebSocket Gateway
Dashboard
React GCS · Live Map
~ 5 KB/s
Avg. uplink (metadata only)
8–12 FPS
On-device inference
< 200 ms
Edge → operator latency
04 / The Team
Seven engineers. One mission.
Cross-disciplinary team spanning AI, embedded systems, flight controls, and full-stack engineering.
Project Supervisor
DR. Ahmed Bin Mahfoudh
Faculty of Computer & Information Technology
Guiding the technical direction, scope, and academic rigor of Eqaab — from system architecture through to final defense.