Hi all,
@supscientist and I just wanted to share our recent work on creating a device that uses edge-deployed machine learning to detect flooding on roadways. We also just wanted to give a major shoutout to the Particle Community for their continued troubleshooting help and support - thank you!!!
Our device is called the TinyCamML, or Tiny Camera with Machine Learning. It uses the Boron and an OpenMV microcontroller to take a photo of a roadway, classify it as containing a flood or no flood, and then transmit only that binary classification to a website in real-time (no image transfer to help preserve privacy). During deployments, it successfully identifies floods 90% of the time.
More details can be found in any of the links below:
Published Open Access Paper: https://doi.org/10.1029/2025WR042023
Preprint: https://eartharxiv.org/repository/view/10278/
Code : https://github.com/TinyCamML/TinyCamML
ML code: https://github.com/TinyCamML/Flood-Model
Data: zenodo.org/records/16740295
Thank you again to the Particle Community!