About this Open-Source Computer Vision Settling Velocity Method

This method automates the measurement of microplastic settling velocity in water columns using computer vision and deep learning object detection. Settling velocity is a critical parameter for predicting microplastic fate and transport in aquatic environments, yet existing measurement approaches rely on manual observation with stopwatches, introducing subjective error and inconsistency. Stacy, Sarker, Baki, and Imtiaz developed this open-source system to replace manual timing with objective, automated trajectory tracking.

The method records video of microplastics sinking through a water column, then uses a YOLOv12n neural network trained for object detection to identify and track particle positions across frames. Custom software calculates settling velocity from the tracked trajectories. Validation testing against manual ground truth (stopwatch timing) showed average errors of 5.97% for 3 mm spheres and 7.14% for 4 mm spheres across three water types. The approach eliminates observer bias, processes multiple particles simultaneously, and produces reproducible numeric outputs suitable for fate and transport modeling.

Key Features

• YOLOv12n object detection model trained for spherical microplastic identification and position tracking in video frames
• Automated trajectory analysis calculates settling velocity without manual stopwatch measurement
• Validated against ground truth with 5.97–7.14% error across 3 mm and 4 mm particle sizes
• Tested in three water types, supporting freshwater and marine fate and transport applications
• Open-source code and method published with full transparency for reproducibility and adaptation

Development and Validation

The method was developed by Catherine L. Stacy, Md Abdul Baset Sarker, Abul B. M. Baki, and Masudul H. Imtiaz and published in the peer-reviewed journal Microplastics in 202, “An Open-Source Computer-Vision-Based Method for Spherical Microplastic Settling Velocity Calculation:. The full paper describes the YOLOv12n model architecture, training dataset, laboratory apparatus, and validation protocol. Ground truth settling velocities were recorded manually with a stopwatch for direct comparison, providing a quantified accuracy baseline. Testing covered three classes of spherical microplastics and three water types, establishing performance across common experimental conditions.

Access and Data Availability

The method is published open-access at https://doi.org/10.3390/microplastics4040075. Code and supplementary data are available at https://ieee-dataport.org/documents/computer-vision-based-method-microplastic-settling-velocity-calculation. The resource is licensed under MDPI’s open-access policy, allowing researchers to download, adapt, and use the method and code in their own studies. The paper includes sufficient detail for reproduction: camera specifications, water tank dimensions, YOLOv12n model configuration, and post-processing scripts for trajectory calculation.

Why This Resource Belongs on Plastiverse

This resource directly addresses the need for objective, standardized microplastic settling velocity measurement—a critical input for fate and transport models and aquatic compartment risk assessment. Settling velocity determines how quickly microplastics sink through the water column, affecting residence time, bioavailability, and long-range dispersal. Unlike manual methods, this computer vision approach eliminates observer bias and scales to handle multiple particles per experiment, supporting both field-based and laboratory monitoring. The open-source code and validation data enable researchers to adopt or adapt the method for their own site-specific conditions, advancing reproducibility in microplastics environmental fate research.