Efficient Microplastic Detection in Water Using ResNet50 and Fluorescence Imaging

Microplastics are prominent contaminants in aquatic environments, which present significant risk to environmental and human health. The capacity to identify microplastics from microscopy images is a vital aspect of environmental microbiology work. In this paper, fluorescence microscopy, which is a novel imaging technique and capable of producing high-quality images, has been used to more efficiently and accurately identify microplastics compared with traditional microscopy techniques. To facilitate automated detection of microplastics, Convolutional Neural Networks, which can classify and segment images of microplastic particles, have been implemented to work with complex and multiscale imaging datasets. The proposed method, which uses deep learning architectures when working with fluorescence microscopy images, improves detection ability and usability of methods for identifying microplastics in variable image quality and structure. Despite inconsistent datasets and limited access to imaging equipment, the paper suggests deep learning with fluorescence microscopy is a promising approach toward scalable deep tissue microplastic detection.