Sensing Comes Before the Model
Three published earth observation architectures decide what to watch from law, revisit and ownership, never from model accuracy. Two of the three ship with an empty model register. First published on CodeNinja Atoms: https://codeatoms.ai/blog/sensing-comes-before-the-model/ . The usual way to start an earth observation project is to pick the model. Someone has a segmentation network that works on…
When designing an earth observation project, the order in which the model is chosen versus the sensing decision is crucial. Three designs in the field—agriculture and earth observation systems in Pakistan, Saudi Arabia, and the United States—demonstrate that sensing should come before model selection. In two of the three systems, no model is committed at all, while in the third, two models are chosen.
The sensing decision is made based on constraints such as legal restrictions, sensor frequency, and ownership. In some cases, models are not verified or committed, yet ownership runs through the services agreement. In Fodder Watch, two models are selected for vegetation analysis, and in Baseline, the sensing decision relies on the requirement to exclude unmanned data.
The hardware used for sensing is determined by baselines, correction sources, and test evidence. Owning the hardware is often more cost-effective than renting, but in cases where sovereignty is a requirement, renting may be the only viable option. The sensing decision ultimately determines the objects that the model must analyze and the license under which it operates.
Written by urgent.news from Dev.to's reporting — not their text. Machine-written — may contain errors; check the original before relying on it.
