Tomato ripeness prediction using low resolution portable spectrometer and machine learning

Tomato ripeness assessment is critical to ensure optimal product quality. This study proposes a novel approach to predict total soluble solids (TSS) and firmness, and classify tomato ripeness using a low-resolution AS7265x portable spectrometer combined with machine learning techniques. Prediction models for TSS and firmness were developed using Partial Least Squares Regression (PLSR) and Artificial Neural Networks (ANN), while ripeness classification used Naive Bayes, Support Vector Machine (SVM) and K-Nearest Neighbors (KNN). The ANN outperformed PLSR, achieving an R² of 0.95 for firmness prediction, and an R² of 0.82 for TSS prediction. The SVM classification model showed strong performance, achieving 96 % accuracy in categorizing the maturity level. These findings highlight the potential of integrating portable spectrometers with advanced machine learning algorithms for real-time, non-destructive tomato ripeness assessment, providing an efficient tool for precision agriculture and supply chain optimization.