tinyML Smart Weather Station: Development of Low-Cost Observation Networks to Support Weather and…

inyML Smart Weather Station: Development of Low-Cost Observation Networks to Support Weather and Hydrology Applications

Paul A. Kucera, Ph.D
Assistant Director
University Corporation for Atmospheric Research

Jenny Plunkett
Senior Developer Relations Engineer
Edge Impulse

Accurate and reliable real-time monitoring and dissemination of observations of atmospheric and hydrologic conditions in general is critical for a variety of research and decision support applications including flood monitoring, agriculture applications, water resource monitoring, and improving numerical weather prediction. A combination of automatic weather stations, stream gauge, and soil moisture/temperature observations can provide critical information about the atmospheric, hydrological, and soil conditions that is critical for many societal applications. The University Corporation for Atmospheric Research (UCAR) have developed a low-cost, innovate environmental sensors. The sensors that have been developed include atmospheric temperature, pressure, humidity, wind speed and direction, solar light, precipitation, stream level, and soil moisture and temperature. The sensors have been developed using innovative new technologies such as 3D printers, single board computing systems, and wireless communications. The presentation will provide an overview of the new observation technology and experiences with recent network deployments.

During this section of the talk, we will walk you through the easiest way to develop a TinyML sound recognition solution. With the help of Edge Impulse Studio we will cover all the steps required to develop a successful solution for your weather station challenge, from data set creation, to developing and training a model,to deployment on any tiny device of your choice.

During the talk we will provide all the resources necessary alongside a reference project for you to develop the best TinyML solution.

source