Tag: geoai

Photogrammetry rebuilds a surface from overlapping photos. LiDAR measures it with a laser. One gives colour and a cheap flight, the other sees under the trees.

Multispectral imagery records a few wide bands of light. Hyperspectral records hundreds of narrow ones. What each can tell you about a field, a roof or a mine.

An orthomosaic is one map-corrected image stitched from many aerial photos. An orthophoto is one corrected photo. Which you have, and where seams go wrong.

Semantic segmentation labels every pixel with a class: building or not. Instance segmentation draws every building as its own object.

A DSM is the top of everything: roofs, treetops, ground. A DTM is bare ground alone, and DEM usually means the same. Which one to use for flood, slope or 3D.

Five ways to show a street rebuilt, from hand-drawn vector layers to a generated aerial view. Only one gives back a picture that stays on the map.

From the official hazard map to a generated aerial view of the water. Which method gives a number, which gives a picture, why an elevation model changes both.

For a classified raster, use Raster > Conversion > Polygonize. Here are the five methods, the fix for staircase edges, and what to do with a photograph.

A building footprint is the ground area inside a building's outer walls. What counts, how it differs from floor area, and how to calculate it from a plan.

GeoAI is machine learning run on geographic data. The five things it extracts today, what it costs in hardware and time, and how to run it in QGIS.