We used transfer learning to teach a model to take advantage of multi-band imagery without discarding the original RGB pre-training. This resulted in significant performance improvement.
We used transfer learning to teach a model to take advantage of multi-band imagery without discarding the original RGB pre-training. This resulted in significant performance improvement.
Idris is a pure, statically-typed functional language with a powerful type system. Let’s explore writing code in Idris using type-driven development and interactive editing.
TileJSON.io is an open source project by Azavea. It is an easy way to view and share raster tile sets using slippy map endpoints.
Without outside funding and given our commitment to openly sharing our knowledge, people often ask: How do you fund your innovation? We outline 6 main ways.
Use this open source data processing pipeline to convert geodatabase files to vector tiles for use in a web application.
This Azavea 10% Time project defines a process for converting big data files to vector tiles and allows a user to identify habitat areas in need of protection.
Follow this guide to set up a geoprocessing web application using GeoTrellis, Akka HTTP, and React.
This post describes how to use deep learning to do semantic segmentation on aerial and satellite imagery, experiments on the ISPRS Potsdam dataset, and how to visualize model predictions on a map.