Session: ASME Undergraduate Student Design Expo

Paper Number: 173081

Investigating Microplastic Settling Velocity Using Computer Vision 

Microplastics, small plastic particles less than or equal to 5 mm in diameter, can enter freshwater and marine systems through direct release from industrial processes or the degradation of larger plastic debris. Microplastic pollution poses threats to ecosystems and human health, and the development of technologies for quantifying microplastics and evaluating their dynamics is crucial for addressing these issues. Settling velocity is an especially critical parameter for predicting the fate of microplastics in aquatic environments as it dictates whether a particle remains in suspension or is deposited. Material properties, including shape, size, and density, as well as environmental factors, such as water density and microorganism concentration, can affect settling velocity. Calculating settling velocity, particularly in the field, poses several challenges, but recent advances in artificial intelligence pose more accurate and efficient alternatives to traditional laboratory methods. The main objective of this research is to develop a computer vision system for real-time and offline detection of microplastics and calculation of their settling velocities. In the laboratory, a high-resolution optical camera interfaced with a computer captured the trajectories of the microplastics as they sank, one at a time, through a controlled water column. A set of training videos was collected and the frames were extracted and annotated in order to develop a YOLOv12n-based object detection model. A Python-based analysis calculated the size and settling velocity of each microplastic. The computer vision system was tested with three sizes of microplastics (3 mm, 4 mm, and 5 mm) as well as three types of water (distilled water, river water, and saltwater). For each microplastic, the time taken to traverse the vertical distance of the camera frame was recorded with a stopwatch and used to manually calculate settling velocity. When comparing the velocities calculated using the object detection model to the stopwatch ground truth data, the average relative error across all water types was 5.97% for the 3 mm microplastics, 7.14% for the 4 mm microplastics, and 6.15% for the 5 mm microplastics. The real-time analysis was consistent with the precision of the offline calculations as the average relative error of all microplastic sizes in distilled water was 4.62%. On average, settling velocity was 1.31 times faster in distilled water compared to saltwater, indicating how microplastic settling velocity can vary substantially under different conditions. This computer vision-based approach not only enhances laboratory studies of settling behavior but also lays the groundwork for future monitoring of microplastic transport in the field. In doing so, it will enable the research community to rapidly track microplastics throughout aquatic systems, refine predictive models, and implement more timely strategies for mitigating pollution.

Presenting Author: Catherine Stacy Macalester College

Presenting Author Biography: My name is Cate Stacy, and I am a junior at Macalester College majoring in Geology and Environmental Studies with a minor in Data Science. I am passionate about studying Earth’s surface processes and hydrology, particularly through the lens of climate change. My prior research has involved developing methods for measuring variations in ice rafted debris using an ocean sediment core. Most recently, I participated in the Aquatic Science, Engineering, and Technology Research Experiences for Undergraduates Program (ASET-REU) at Clarkson University. I plan on attending graduate school and hope to pursue a career that blends my interests in Earth science, computer programming, and data analysis.

Authors:

Catherine Stacy Macalester College