Convolutional capture of the expansion of extra virgin olive oil droplets to quantify adulteration

Abstract

This record corresponds to the peer-reviewed journal article “Convolutional capture of the expansion of extra virgin olive oil droplets to quantify adulteration”, published in Food Chemistry (Volume 368, 2022), a Journal Citation Reports (JCR) indexed journal.

The work presents the development and validation of a computer vision–based methodology for the detection and quantification of extra virgin olive oil (EVOO) adulterations using convolutional neural networks (CNNs). The proposed approach is based on the analysis of the temporal expansion of EVOO droplets captured under controlled conditions, generating a large-scale image dataset comprising more than 302,000 images of pure and adulterated samples.

Several deep learning models were designed and evaluated for (i) the classification of different EVOO types, (ii) the detection and quantification of adulterations with sunflower and corn oils at concentrations ranging from 2.5% to 10% (w/w), and (iii) a global CNN model integrating all EVOOs and adulteration scenarios. The optimized models achieved overall classification accuracies above 96%, demonstrating a high sensitivity to subtle physicochemical differences associated with oil composition.

The results confirm that droplet image analysis combined with deep learning constitutes a rapid, non-destructive, and reliable tool for food quality control and fraud detection, with potential applicability throughout the olive oil distribution chain.