Method, measurement and the occasional unglamorous finding. Written by the people who did the engineering, not by a marketing department describing it afterwards.
Written by
The engineering and research teams
Topics
Applied AI, data, earth observation, public systems
An imported crop classification model reported 94 per cent accuracy and was wrong about a third of the fields we checked. The gap was not the model. It was the ground truth.
Institutions ask for a dashboard and discover they have bought a question. Here is the order of work that actually produces a number somebody is willing to sign.
Most exploration licences in the region sit on more geophysical and satellite coverage than anybody has looked at carefully. Fusing what exists is usually cheaper than acquiring more.
Digitization projects fail at indexing, not at scanning. A searchable archive requires decisions about classification that only the institution can make.
Yield, soil condition and water use on most farms here are estimated rather than measured. The first season of real measurement rarely tells you what you expected.
Read
These notes come out of real engagements. If one of them describes a problem you have, say so.
