Vessel names for the radar detections inside closed fishing areas
A fisheries officer guarding a closed spawning ground gets a free radar picture of it most weeks, and most boats in that picture have no name.
One pass can show thirty dots inside the line, three of them carrying AIS. The patrol, a day away, can board two or three; the rest are filed as suspected.
Nameboard turns each radar pass over an agency's named closed areas into a list its patrol can act on. Every detection inside the line is traced against nearby AIS tracks, the license register, VMS and patrol logs where shared, and the area's past seasons. Within six hours of the scene, each has a proposed vessel and its signs, or no name and the reason.
Two steps in Nameboard's trace are too noisy for rules: telling a boat from sea clutter, and saying which boat a dark detection is. A convolutional model we trained on radar chips decides which bright targets are vessels and corrects the length a raw outline overstates. A ranking model then scores each candidate on six signs, such as an AIS track gone quiet near the line.
No detection gets a name from Nameboard on one sign alone. A name needs two independent signs, a score of 0.80 and a lead of 0.20 over the next vessel; anything less comes back nameless with one of six reasons. An agency officer accepts or rejects every name before the patrol sees it. Targets the radar model cannot settle go to our analysts, whose labels on public radar chips build a corpus we own.
The founders
Iago Vilariño
CEO
Spent six seasons as an onboard fisheries observer on trawlers out of Vigo, logging unlit boats at the edge of closed grounds that no season report ever put a name to.
Antía Lamas
CDO
Built ship detection on free radar scenes for a marine research group in Vigo, and after every run was asked the one thing the pipeline could not say: which boat.
Uxía Mosquera
Chief Strategy Officer
Planned inspection campaigns for a fisheries control service in Galicia, where each season's report closed on a column of detections filed as suspected; left in 2023 to change that without leaving Vigo.
Where it runs
Radar scenes reach Nameboard in bursts set by satellite orbits and closure calendars. Passes over a region's areas land within one hour, then nothing for half a day; a new area starts with a backfill of two seasons; spawning closures open together each spring. License registers and VMS positions may not go to another company's hosted model, so every step of the trace runs on cloud infrastructure in N. Virginia.
Nameboard's radar and ranking models now run on cloud GPU capacity we control in N. Virginia, billed by the hour and idle between passes. By April 2027 the length model is retrained on the northern spring's analyst labels. By October 2027 each area's past seasons become an embedding index of track shapes, and the pass windows move to GPUs reserved for a year in N. Virginia.
Nameboard S.L. was registered in October 2023 in Vigo, Europe's largest fishing port and the seat of the European Fisheries Control Agency, where Uxía Mosquera had planned inspection campaigns. She founded it with Iago Vilariño, who had logged the nameless dot from an observer's deck, and Antía Lamas, who had met it in a radar lab. The first years went into the radar model and the trace, built on public scenes of the Galician coast and the Bay of Biscay. Nameboard is built for the agencies that patrol closed fishing areas: Spain's Secretaría General de Pesca, the Xunta de Galicia's coastguard service, and their counterparts in other EU states and beyond.
- N. Virginia
- Where the trace runs
- Vigo
- Nameboard S.L., Rúa de Policarpo Sanz 22, Planta 3, 36202 Vigo, Spain