This work introduces two datasets: FTIR-Plastics-C4 (Fourier Transform Infrared Spectroscopy, in plastics, at a wavenumber spectral resolution of 4 cm⁻¹) and FTIR-Plastics-C8 (Fourier Transform Infrared Spectroscopy, in plastics, at a wavenumber spectral resolution of 8 cm⁻¹), each comprising 3,000 spectra corresponding to the most used synthetic polymers worldwide. The main contribution of this work lies in the selection and FTIR characterization of the six polymers commonly used in everyday life and industry, namely Polyethylene Terephthalate (PET), High-Density Polyethylene (HDPE), Polyvinyl Chloride (PVC), Low-Density Polyethylene (LDPE), Polypropylene (PP), and Polystyrene (PS). FTIR-Plastics-C4 consists of 3,000 spectra obtained with a configuration of 32 scans and a resolution of 4 cm⁻¹, covering a range from 4000 to 400 cm⁻¹. The FTIR-Plastics-C8 dataset also contains 3,000 spectra obtained with 32 scans and a resolution of 8 cm⁻¹ within the same range. A cleaning stage was applied to the FTIR-Plastics datasets, removing the header containing 19 lines and a footer with 34 lines from the original file. Additionally, a standardization process assigns 15 lines in the files to highlight information regarding the equipment used (based on the information provided by a Jasco spectrophotometer, model FT/IR-6700 PRO 4x, used for polymer characterization). The final dataset is in tabular .csv file format. The dataset is available on an open repository, and its application was designed to identify microplastics extracted from the environment and enable comparisons between commercial polymers.

This study examines the potential of machine learning (ML) and deep learning (DL) techniques for classifying microplastics using Fourier-transform infrared (FTIR) spectroscopy. Six commonly used industrial plastics (PET, HDPE, PVC, LDPE, PP, and PS) were analyzed. A significant contribution of this research is the use of broader and more varied spectral ranges than those typically reported in the state of the art. Furthermore, the impact of different normalization techniques (Min-Max, Max-Abs, Sum of Squares, and Z-Score) on classification accuracy was evaluated. The study assessed the performance of ML algorithms, such as k-nearest neighbors (k-NN), support vector machines (SVM), naive Bayes (NB), random forest (RF), and artificial neural networks architectures (including convolutional neural networks (CNNs) and multilayer perceptrons (MLPs)). Models were trained and validated using the FTIR-PLASTIC-c4 dataset with a 10-fold cross-validation approach to ensure robustness. The results showed that Z-score normalization significantly improved stability and generalization across most models, with CNN, MLP, and RF achieving near-perfect values in accuracy, precision, recall, and F1-score. In contrast, the sum of squares normalization was less effective, particularly for CNNs, due to its sensitivity to scale and data distribution. Notably, naive Bayes consistently underperformed because of its limitations in analyzing complex spectral data. The findings highlight the effectiveness of FTIR spectra with broad and variable ranges for the automated classification of microplastics using ML techniques, along with appropriate normalization methods.

Los microplásticos presentes en el agua se han convertido en un problema ambiental grave, por lo que ha surgido la necesidad de desarrollar técnicas y herramientas para su detección y posterior eliminación. En este estudio, se utilizó el modelo de visión por computadora YOLO (You Only Look Once) en su versión 8, para la detección de microplásticos en agua, el cual fue entrenado utilizando un conjunto de 500 imágenes obtenidas mediante microscopio estereoscópico, derivadas de muestras recolectadas en dos zonas del Estado de México: Valle de Bravo y el municipio de Malinalco. Adicionalmente a este conjunto de imágenes se le aplicaron diferentes técnicas de aumento de datos, incluyendo, Resize, Crop, Flip, Brightness and Contrast, HSV, Gamma Correction, CLAHE, RGB Shift y Gaussian Blur, con el objetivo de generar un conjunto más diverso. Los resultados obtenidos indican que YOLO, entrenado con el conjunto aumentado, alcanzó una precisión de 0.825 y un mAP@50 de 0.73, valores superiores a los logrados con el conjunto original sin aumentos. En conclusión, YOLO, en combinado con técnicas de aumento de datos, muestra ser eficaz para la detección de microplásticos, sentando las bases para su aplicación práctica en el monitoreo de cuerpos de agua.