I’ve been pondering this question for some time. The promise of artificial intelligence (AI) is to automate and replace mundane tasks. As separate disciplines, GIS and remote sensing have plenty that is tedious: Data processing and (core elements) of analysis to name but two.
I’ve spent most of my career in the vector based GIS world. That is points, lines and polygons. As a student I struggled with the concept. Finally realising GIS vectors were in many ways outlines (in the case of polygons) or centroids (points and lines) of actual features; buildings, parks, rivers, traffic lights etc Tangible, and concrete if but markers. These were markers with coordinates and attributes: the secret sauce of GIS data. There is much one can do with vectors in terms of analysis (network analysis etc). Also, as I mentioned in my most recent Geospatial 2.0 Friday Talk, visualizing 2D vector GIS data helps us see patterns. One of the many powers of maps.
Remote sensing has long been focused on earth observation (EO) imagery. Not my area of experience, but my understanding is that experts process and analyse images (SAR or radar similarly falls under this remote sensing category, you’ll need to ask Joe Morrison, Roi Shilo or Steven Ramage more on this topic). One key area is feature identification and change. Much used by the military, EO data is now seeing broader availability and use. The release of new micro satellites has led to a big drive to launch new satellite constellations. We are about to experience a tsunami of new EO data.
You can tell I am thinking out loud as I write this post. Vectors are concrete while imagery is abstract. That is, imagery needs processing to be useful in analysis. I’m seeing many in the Geospatial 2.0 world now using AI to process imagery: Often to identify, and extract features respectively from imagery. Attentive provide a good example. They are identifying residential and commercial property features (parking , driveways etc) and automating the measurement process.
So AI is processing the abstract (imagery) and turning it into the concrete (vector) or data that can be used in a GIS and beyond. There are big implications here.
I’ll need to ponder on this topic more. No rocket science here, just me trying to connect dots (as ever). If you can add to this conversation, I’d welcome your thoughts.