Deep learning based approach for automated characterization of large marine microplastic particles.

The rapidly growing concern of marine microplastic pollution has drawn attentions globally. Microplastic particles are normally subjected to visual characterization prior to more sophisticated chemical analyses. However, the misidentification rate of current visual inspection approaches remains high. This study proposed a state-of-the-art deep learning-based approach, Mask R-CNN, to locate, classify, and segment large marine microplastic particles with various shapes (fiber, fragment, pellet, and rod). A microplastic dataset including 3000 images was established to train and validate this Mask R-CNN algorithm, which was backboned by a Resnet 101 architecture and could be tuned in less than 8 h. The fully trained Mask R-CNN algorithm was compared with U-Net in characterizing microplastics against various backgrounds. The results showed that the algorithm could achieve Precision = 93.30%, Recall = 95.40%, F1 score = 94.34%, APbb (Average precision of bounding box) = 92.7%, and APm (Average precision of mask) = 82.6% in a 250 images test dataset. The algorithm could also achieve a processing speed of 12.5 FPS. The results obtained in this study implied that the Mask R-CNN algorithm is a promising microplastic characterization method that can be potentially used in the future for large-scale surveys.

Spatiotemporal variations of surface water microplastics near Kyushu, Japan: A quali-quantitative analysis.

Microplastics in the ocean are threatening marine ecosystems. Although plastic contaminants are ubiquitous, their distribution is thought to be heterogeneous. Here, we elucidate the spatial and temporal variations in the quanti-qualitative characteristics of microplastics near Kyushu, Japan in the East China Sea. Six surveys across nine stations were conducted over a 14-month period, and a total of 6131 plastic items were identified. The average microplastic abundance and size were 0.49 ± 0.92 (items·m-3 ± S.D.), and 1.71 ± 0.93 (mm ± S.D.), respectively. Differences between the highest and lowest abundances were 50-fold among monthly means, and 550-fold across all net tows. With respect to colour, polymer type, and shape, white and transparent polyethylene fragments were the dominant composition. There were significant differences for each of the analytical microplastic parameters among the survey months. Our results provide baseline data and lead to a more comprehensive understanding of the spatiotemporal characteristics of microplastic pollution.