Since 2020 Helicopters of DC has been revealing Washington, D.C.’s helicopters to residents using crowd-sourced and open-source data. This talk will start by briefly outlining the Twitter data submission system we designed to track D.C.'s helicopters, before explaining how Musk's changes to the platform invalidated it (some reasons, like dropping FourSquare API, might surprise you). We dive deep into the Telegram bot we designed with a free low/no code platform Make.com that allows residents to submit "helicopter spots" and returns them instant answers from ADS-B and computer vision analysis of their photos. I'll demonstrate how set up your own computer vision program with a free Roboflow account. This platform helps you manage enormous datasets of photos, and collaboratively annotate, train, and deploy models to infer on their hardware without a line of code. Last we will look at ATC radio, and how we are remotely collecting it with a waterproof Raspberry Pi, an RTL-SDR, and open-source RTL-SDR Airband software. Helicopter radio shares a channel with Washington National Airport's ground traffic (responsible for some 5 million transmissions a month). I will show the audience how despite aviation-specific terminology we are using OpenAI-Whisper to filter these transmissions for helicopter-specific keywords. Finally we will look at how our Telegram users can review the relevant radio transmissions for accuracy, and tag them with terms to build an annotated map with attached audio that is shared with the public.
SANS Open-Source Intelligence Summit 2024
Identifying the Secret Military Helicopters of Washington, D.C.
Andrew Logan, Investigative Developer, Helicopters of DC
View upcoming Summits: www.sans.org/u/DuS