Izquierdo, ManuelLastra Mejías, MiguelGonzález Flores, EsterCancilla, John C.Aroca Santos, ReginaTorrecilla Velasco, José Santiago2026-01-122026-01-122020-04-01anuel Izquierdo, Miguel Lastra-Mejías, Ester González-Flores, John C. Cancilla, Regina Aroca-Santos, José S. Torrecilla, Deep thermal imaging to compute the adulteration state of extra virgin olive oil, Computers and Electronics in Agriculture, Volume 171, 2020, 105290, ISSN 0168-1699, https://doi.org/10.1016/j.compag.2020.105290.10.1016/j.compag.2020.105290https://hdl.handle.net/20.500.14352/129845A new and reliable method to detect adulterations is presented in this work by comparing the thermal evolution of pure and adulterated extra virgin olive oil (EVOO) during its cooling process. For this purpose, algorithms based on convolutional neural networks have been tested for the analysis and classification of different thermographic images of pure and adulterated EVOO containing amounts ≤ 8% in weight of different adulterants (refined olive oil, olive pomace oil, and sunflower oil). Eight convolutional neural network models have been designed for the classification of EVOO samples and for the estimation of adulterant concentrations. Finally, other similar models were developed that included all the samples, containing any of the adulterants, to reach a global more versatile tool. The statistical performance of the convolutional neural networks in terms of classification accuracy range from 97 to 100%.engjournal articlehttps://doi.org/10.1016/j.compag.2020.105290https://www.sciencedirect.com/science/article/pii/S0168169919314620restricted access66Convolutional neural networksDeep learningInfrared thermographyExtra virgin olive oil Edible oilsAdulteration detectionIngeniería química3309 Tecnología de Los Alimentos