Sharks flagged on the live feed, while the drone is still overhead.
A neural network watches the drone's live video and marks sharks the moment they appear, on the pilot's own laptop or tablet, with no upload and no cloud. Run it below on sample drone footage, your own clip, or a live controller feed.
Your own video, photos, camera or controller window are analysed in this browser tab, frame by frame, on this device. The sample flight replays detections computed ahead of time with the full analysis, which runs live on computers with a fast graphics chip; phones and slower computers run a lighter analysis. The tracker and alerts run live either way. A sighting becomes an alert only when it persists across consecutive frames, which filters out glints and whitecaps that flash for a single frame.
Choose a source belowLoading the network…
“Drone controller window” shares a window such as DJI Fly or a capture card showing the remote controller's HDMI feed, so the detector watches exactly what the pilot sees.
Benchmark results
Spotting conditions, beach by beach
Sun glint, wind chop and swell decide how far down a pilot can see. Pick a beach for the current forecast, where the sun sits, and which way to point the camera to keep the glint out of frame.
…
Forecast: Open-Meteo weather and marine models, fetched live for the chosen beach. Sun position computed in the page. The glint meter in the live detector measures the same thing on the actual video.
Sharks off the Western Australian coast
Recorded incidents in WA by species
Shark reports on the WA map, last 30 days
Where each one turns up
A network of cleared, qualified pilots
The software is half of it. Every pilot on a beach roster needs current clearances. The register flags anything expiring and links to the official check for each one. Illustrative records, not real pilots or clubs.
Same flights, fisheries answers
The aircraft and the network that watch for sharks can also count, map and log what fisheries managers need.