Hydrocam Sample Data and Processed Results

This HydroShare resource provides a complete example of camera-based streamflow monitoring data collection and automated segmentation processing for the Blacksmith Fork site, demonstrated on one day of real image data. It includes both example image inputs and compute outputs generated using a containerized cloud-based inference pipeline. The processing workflow uses the Segment Anything deep learning model, deployed in a serverless environment with AWS Lambda and S3. Each image is segmented to identify regions of interest (ROIs) and calculate water-relevant pixel statistics. Ground truth comparison supports quality assurance using Intersection over Union (IoU) scores. Results are automatically uploaded and stored in a PostgreSQL database for hydrologic analysis. This dataset supports the reproducibility of the modeling approaches described in submitted manuscripts to Environmental Modelling & Software, offering transparency into the full data processing pipeline from raw image ingestion to output storage. It serves as a reference implementation for camera-based environmental monitoring at scale.