Tools: Python, OpenCV, NumPy, Folium, Onshape, Excel

After seeing the scale and impact of Ontario’s wildfires this year, Paanini and I were motivated to explore how technology could support faster and more accessible wildfire reconnaissance. Together, we developed AeroScout, a Python and OpenCV prototype that analyzes prerecorded footage and flags suspected fire regions using HSV colour filtering. The project combines computer vision, simulated detection mapping, a lightweight 3D-printed camera hood, and a proposed agency deployment model. AeroScout is currently a software prototype, with live telemetry and physical drone integration planned for future development. 

Video frames are converted to HSV, filtered for fire-coloured regions, cleaned to reduce noise, and logged as JSON detections. 

Detection timestamps are grouped and plotted along a simulated camera route. Live GPS integration remains a future development goal. 

A friction-fit camera hood was modelled in Onshape, 3D printed, and evaluated for fit. Its estimated CAD mass is 4.1 grams. 

An illustrative first-year model estimates $85,000 in revenue, $31,520 in direct delivery costs, and a 62.9% gross margin per agency. 

A scenario analysis compares AeroScout’s estimated cost per check with commercial plane and helicopter benchmarks. 

Future development progresses from detection testing and drone integration to an agency pilot and commercial validation.