SpectraSentinel

Lightweight dual-stream real-time drone detection and tracking system

SpectraSentinel: Real-Time Drone Detection

Lightweight Dual-Stream System for Drone Detection, Tracking & Payload Identification

🥈 Global 2nd Runner-up - IEEE SPS Video & Image Processing (VIP) Cup 2025

Project Overview

SpectraSentinel is a novel dual-stream architecture that combines spectral and spatial fusion cues for real-time drone detection, tracking, and payload identification. The system is designed to be lightweight and efficient for deployment in resource-constrained environments.

Key Features

🎯 Dual-Stream Architecture: Combines spectral and spatial information for robust detection
Real-Time Performance: Optimized for low-latency processing
🪶 Lightweight Design: Minimal computational requirements
📦 Payload Identification: Novel approach to classify drone payloads
🎬 Video Processing: Efficient tracking across video frames

Technical Innovation

  • Spectral-Spatial Fusion: Novel fusion strategy combining complementary features
  • Lightweight CNN: Optimized convolutional architectures for speed
  • Tracking Algorithm: Robust multi-object tracking for drone trajectories
  • Real-Time Pipeline: End-to-end system from detection to classification

Competition Success

IEEE SPS VIP Cup 2025: Achieved Global 2nd Runner-up position in the international video and image processing competition [Competition Link]

Applications

  • Security and surveillance systems
  • Airspace monitoring
  • Counter-drone systems
  • Infrastructure protection
  • Event security

Research Team

Lead Researchers: Shahriar Kabir, Istiak Ahmmed Rifti
Contributors: H.M. Shadman Tabib, Mushfiqur Rahman, Sadatul Islam Sadi, Hasnaen Adil, Ahmed Mahir Sultan Rumi
Supervisor: Prof. Dr. Ch. Md. Rakin Haider, BUET

Publication

arXiv: “SpectraSentinel: Lightweight Dual-Stream Real-Time Drone Detection, Tracking and Payload Identification” [PDF]


SpectraSentinel represents a significant advancement in practical, deployable drone detection systems.