| CPC G06V 20/188 (2022.01) [A01B 79/005 (2013.01); G06N 20/00 (2019.01); G06V 10/25 (2022.01)] | 21 Claims |
|
1. A method implemented using one or more resource-constrained edge processors at an edge of a distributed computing network that are remote from one or more central servers of the distributed computing network, the method comprising:
acquiring triage data from a sensor of an edge computing node carried through an agricultural field by agricultural equipment, wherein the triage data is acquired at a first level of detail;
locally processing the triage data at the edge using one or more machine learning models stored on or executed by the edge computing node to detect one or more targeted plant traits exhibited by one or more plants in the agricultural field;
in response to detection of the one or more targeted plant traits, establishing a region of interest (ROI) in the agricultural field and downloading, from one or more of the central servers through the distributed computing network, parameters of a targeted inference machine learning model, wherein the targeted inference machine learning model is selected from a library of machine learning model parameters based on the detected one or more targeted plant traits;
acquiring targeted inference data from the sensor at a second level of detail while the sensor of the edge computing node is carried through the ROI of the agricultural field, wherein the second level of detail is greater than the first level of detail; and
locally processing the targeted inference data at the edge using the targeted inference machine learning model to make a targeted inference about plants within the ROI of the agricultural field.
|