REAL TIME RAINFALL INTENSITY QUANTIFICATION using COMPUTER VISION AND MACHINE LEARNING

In January 2020, a partnership involving H2i Balkans and Aqua Procon (Brno, Czech Republic) won a support of the UNDP Challenge Fund for project on “Quantifying real time rainfall intensity based on computer vision and machine learning.

Through collection and processing of video footage into instantaneous rainfall intensities, the project enables authorities to use widely-available CCTV recordings to better manage local floods, help transportation authorities to improve road safety, farmers to better manage their crops and energy authorities to manage production of hydro power and plan maintenance of energy installations.  The ultimate project beneficiary of this project is Federal Hydro Meteorological Institute (FHMI) of Bosnia and Herzegovina, an Agency responsible for hydrological and meteorological data collection, processing and dissemination.

Image below illustrates accuracy and advantages of the AI-based computer vision technology developed by H2i compared to more traditional rainfall gauge observations. In addition to significantly lower cost of implementation, the novel AI-based approach deployed in the project introduces significant advantages compared to conventional rainfall monitoring approaches.