AI4EO Mapping and Monitoring Services

Lynker Analytics, a Wellington-based data science company that describes itself as an AI centre of excellence for Lynker, builds custom machine-learning (computer vision) models that turn satellite and aerial imagery into GIS datasets and monitoring systems. It serves government agencies, councils and land managers that need current, consistent maps of land cover, land use, buildings, roads, wetlands and vegetation, which are slow and costly to capture by hand.

Models are built in Python with TensorFlow/Keras, trained with active ('human in the loop') learning, often through ArcGIS Online labelling apps shared with the client, and fine-tuned on the target imagery. Outputs are vectorised, quality-checked and delivered as topologically correct feature classes with data quality and accuracy reports. Satellite inputs in published projects include 4-band pan-sharpened Airbus SPOT imagery (1.5 m) for Bay of Plenty Regional Council's 2021 land-use layer, Maxar imagery for the Ministry for the Environment's riparian planting survival assessment, Planet and Maxar imagery for Auckland Council's post-Cyclone Gabrielle bare-earth mapping, and satellite imagery for a NZ Defence Force trial that extracted buildings (F1 0.90) and roads (F1 0.93) over Fiji, Samoa, Tonga and the Marshall Islands. With the Meat Industry Association, Lynker Analytics built an AI Earth-observation system that identifies beef-farming land converted from forest, so European importers can show compliance with the EU Deforestation Regulation (EUDR). Many other projects use aerial photography and LiDAR, including more than 3 million building outlines captured for Land Information New Zealand between 2019 and 2024.