Image Dataset of Water Samples with Suspended Solids under High-Intensity White Light for Deep Learning Classification

This dataset contains labeled images of water samples with varying concentrations of total suspended solids (TSS), acquired under controlled experimental conditions using a high-intensity white light illumination system. Water samples were prepared using distilled water and controlled concentrations of suspended solids ranging from 40 to 6000 mg/L, allowing classification into four water quality levels: good, acceptable, contaminated, and strongly contaminated. Images were obtained under optimized illumination conditions, which demonstrated the highest classification performance in the associated study, achieving near-perfect accuracy in deep learning models based on MobileNetV2 . The dataset is structured for machine learning applications and supports the development, training, and validation of convolutional neural networks for automated water quality monitoring. This dataset is particularly relevant for real-world applications due to its improved image quality and model performance.