Sentinel
Advance UAV detection software
The problem Sentinel solves
The Rising Threat of Malicious Drones
Malicious Use: There is a rapid rise in the use of small , low-cost drones for malicious activities like surveillance and smuggling.
Detection Gap:Traditional radar and GPS systems fail to detect these low-altitude and micro-drones.
High Security Risk: This creates significant security risks for defense installations, borders, airports, and critical infrastructure.
Cost Barrier: Existing counter-drone systems are often prohibitively expensive and heavily dependent on specialized hardware.
Challenges we ran into
1. Lack of Specialized Data (Thermal & RF)
The Problem: While standard drone images are easy to find, Thermal (Infrared) drone footage and Raw RF Drone Signals are extremely rare in open-source datasets.
Our Struggle: We couldn't just download a "ready-to-use" dataset for the night-vision and signal models.
How we solved it: We had to manually curate a small custom dataset and use Data Augmentation (flipping, rotating, changing brightness) to artificially increase our training data size.
2. Challenge : Converting raw high-frequency radio signals into 2D Spectrogram images requires heavy mathematical computation (Fast Fourier Transform), which initially caused a 300ms lag in our system.
Why it mattered: This delay meant the "Alert" would pop up after the drone had already moved, making the tracking feel unresponsive.
Tracks Applied (1)
MongoDB Atlas
Major League Hacking