Raman spectra for plastics identification (RaSPI) and Raman maps for plastics identification (RaMPI) datasets.

Area of Science:

• Analytical Chemistry
• Environmental Science
• Data Science

Background:

• Accurate identification of environmental plastics pollution is crucial.
• Machine learning (ML) techniques show promise for accelerating this identification process.
• High-quality, diverse datasets are essential for developing robust ML models.

Purpose of the Study:

• To introduce two novel Raman spectroscopy datasets: RaSPI and RaMPI.
• To support the development and validation of next-generation ML methods for plastic identification.
• To provide standardized, high-quality data for researchers in environmental plastics analysis.

Main Methods:

• The Raman spectra for plastics identification (RaSPI) dataset includes 402 Raman spectra across 14 plastic types.
• The Raman maps for plastics identification (RaMPI) dataset comprises 34 2D spectroscopic maps (33,119 spectra).
• Both datasets feature high spectral resolution (<1 cm⁻¹) and include data from pristine and polluted environmental samples, with manual classification of 14 plastic types.

Main Results:

• The RaSPI dataset offers detailed spectral information for 14 plastic types, including variations in additives.
• The RaMPI dataset provides extensive 2D spectroscopic data with varying signal-to-noise ratios, suitable for testing algorithms.
• Both datasets are characterized by consistency and quality, making them valuable for ML model training and methodology validation.

Conclusions:

• The RaSPI and RaMPI datasets represent significant resources for the scientific community.
• These datasets will accelerate advancements in ML-driven environmental plastics identification and microplastics research.
• Researchers can utilize these datasets for training ML models, developing spectroscopic processing algorithms, and validating new methodologies.