Build custom AI models for object detection

January 25th, 2019, Published in Articles: PositionIT

Picterra introduced the custom detector, a tool that allows users of its online platform to build their own artificial intelligence model to detect objects in aerial and satellite images. Without any previous machine learning experience, the new tool allows users to create their own AI detection model and train it to identify their specific objects of interest in just a few steps. It is possible to export these detections and incorporate them into any geospatial analytics workflow such as a .csv file or in a variety of georeferenced formats (KML, GeoJSON, and Shapefile) for mapping applications. The tool has a variety of uses, among other for detecting buildings, tracking farming plots, counting cattle, and identifying solar panels. The tool works best with sharp image, and it is recommended when drawing annotations around objects that it is best to keep polygons tight. In 2018 the company launched a self-serve online platform that makes the combined power of earth observation and artificial intelligence accessible to general users. It also offers AI as a service (AIaaS) to test and implement large-scale AI-powered geospatial analytics.

Contact Veronica Alonso, Picterra, veronica.alonso@picterra.ch

Related Articles

  • PositionIT Inbox, Jan/Feb 2019
  • Online tool to simplify environmental sensitivity screening
  • Surveyors talk tariffs, productivity and private practice principles
  • Lessons from a digital mine
  • Geospatial-based climate adaptation guide