Custom GeoAI Solutions

Mapizy, a GeoAI company based in Western Australia, builds custom computer-vision workflows for organisations that need current, quantified information about built and natural environments, where field surveys are labour-intensive and costly and existing map data is often out of date. A client supplies the project area, the analytics it needs and any data of its own; Mapizy sources suitable imagery, applies deep-learning processing workflows tailored to the task, and delivers maps, statistics and geospatial analysis.

Satellite imagery is one of several inputs, alongside aerial, drone, mobile-mapping and ground-level images. Published satellite projects include a mine-rehabilitation assessment for BHP, in which an experiment used Airbus Pléiades imagery to map plant lifeforms over 23 km² at a fraction of the cost of aerial imagery and LiDAR, and a sandalwood plantation inventory of more than 3,000 hectares from Pléiades images, produced twice a year with stocking, crown-size, tree-health and change-detection maps that the client's field staff verified on the ground. A forest-inventory solution built on Maxar WorldView-3 imagery won first place in the agriculture category of the DigitalGlobe (Maxar) 2018 Australia Sustainability Hackathon.

Applications Mapizy lists include mining and energy (remote tracking of mine-rehabilitation parameters, carbon sources and sinks, mineral-deposit detection from multi-source datasets), agriculture and forestry (forest inventory, deforestation, land-use classification), infrastructure (road safety features in aerial and satellite images, vegetation encroachment on power lines, water leaks and blockages from satellite images) and defence (object detection, off-nadir image analysis, anomaly detection and quantifying change to built and natural environments).