Microplastic pollution assessment with digital holography and zero-shot learning

Abstract

Microplastic pollution poses severe environmental problems. Developing effective imaging tools for the identification and analysis of microplastics is a critical step to curtail their proliferation. Digital holographic imaging can record the morphological and refractive index information of such small plastic fragments, yet due to the heterogeneous sampling environments and variations in the microplastic shapes, traditional supervised learning methods are of limited use. In this work, we pioneer a zero-shot learning method that combines the holographic images with their semantic attributes to identify the microplastics in heterogeneous samples, even if they have not appeared in the training dataset. It makes use of the attention mechanism for image feature extraction, and the Kullback-Leibler divergence both to alleviate the domain shift problem and to guide the training of the mapping function. Experimental results demonstrate the effectiveness of our approach, and the potential use in a wide variety of environmental pollution assessment.

Snapshot Polarization-Sensitive Holography for Detecting Microplastics in Turbid Water

“…Hence, only a few useful characteristics of the MP particles could be estimated, which is also a drawback of traditional intensity-based imaging devices working underwater. For instance, it is challenging to extract the fractal parameters or visual features if only the degraded intensity maps are available. However, due to the capability of resolving polarization responses, additional DoLP and AoP features can be obtained using the proposed computational method.…”

“…More distinguishable features , could be extracted from the polarization-resolved holograms which contain abundant information about MP particles. In addition to manually selecting useful features, artificial intelligent methods , could be utilized to extract intrinsic features from the images. To enhance the practical applicability of our technique, it is crucial to explore the influence of various nonmaterial factors, such as the color, size, thickness, and surface characteristics of MPs, on the classification metrics.…”

“…They showed that the holographic fingerprint is informative and can offer new possibilities for environment monitoring. Meanwhile, Zhu et al ,, explored the feasibility of deep learning-based holographic methods for identifying different types of MPs. They demonstrated that the discriminative information about the MP samples encoded in the holograms can be extracted using deep learning approaches.…”

Holographic and polarization features analysis for microplastics characterization and water monitoring

“…Polarization states characteristics, including phase retardation, optics axis orientation, angle of polarization (AoP), and degree of linear polarization (DoLP), reveal the anisotropy, birefringence, and structural information of the materials. 1,2 Digital holography [3][4][5] (DH) recently demonstrates its outstanding performance in microplastic (MP) identification [6][7][8] and material analysis. 9,10 Compared with the structural and morphological characteristics, i.e., circularity and roughness, polarization and holographic features 11 are closely related to the material molecular structures.…”

“…1,2 Digital holography [3][4][5] (DH) recently demonstrates its outstanding performance in microplastic (MP) identification [6][7][8] and material analysis. 9,10 Compared with the structural and morphological characteristics, i.e., circularity and roughness, polarization and holographic features 11 are closely related to the material molecular structures. 12 In this work, we propose an intelligent polarization holographic system that can capture informative sample images for their discriminative analysis and identification.…”