Computer vision-based monitoring of stormwater discharge: proof-of-concept for stage measurements

• Stormflow stage in culverts captured using Machine Learning and math in images. • Finding culverts using ML, its shape and precise size through ellipse fitting. • Flowing culvert unoccupied area shape extraction through Mask R-CNN model. • Using homography and a reference object for image-to-real-world transformation. • Stages: ±0.5 cm (calm) and ± 1.5 cm (turbulent) for over 80% of measurements. Stormwater systems serve as critical urban drainage infrastructure, maintaining municipal functionality and playing key hydrological and environmental roles. Yet, stormwater outlets are not designed for stormflow monitoring, making traditional contact-based sensors unsuitable for gauging highly variable and turbulent flows. To address this, we propose a computer vision approach that quantifies discharge from near-range images and videos of the outlet. Given the challenges of applying traditional computer vision in natural environments with variable lighting and environmental noise, our method employs a combination of machine learning and computer vision (CV-ML) techniques, leveraging the recognizable geometrical shape of culverts in otherwise very noisy images. Specifically, water stage is obtained through subtracting the extracted shape of the round culvert from the height of empty area above water. Subsequently, image-based measurements are transformed into real-world units by applying a homography transformation calibrated using a checkerboard reference object placed parallel to the outlet. Performance evaluation on a culvert of known dimensions demonstrated ± 1 cm accuracy in approximately 80% of measurements. For water stages evaluated under calm, moderately and turbulent flows, the method estimated, respectively, more than 80% of stages within ± 0.5 cm, more than 80% of stage values within ± 0.75 cm, and 63% of values within ± 1 cm (96% within ± 2 cm). These results show great promise in the use of image-based techniques in difficult and turbulent conditions where no traditional techniques are applicable. They are also the prerequisite for estimations of discharge, for which more work is necessary, and on-going.