The future of food safety will be heavily impacted by technology. A relevant example is presented in a recent publication that describes the use of an artificial intelligence (AI)-based method to rapidly detect trace levels of gluten or nuts in lentil flour samples.
The method was developed using mathematical models based on convolutional neural networks and transfer learning (viz., ResNet34) trained to identify lentil flour samples that contain trace levels of wheat flour or ground pistachio nuts.
The technique is based on the analysis of photographs taken by a simple reflex camera and further classification into groups assigned to the type and amount (up to 50 ppm) of trace material present. Two different algorithms were trained (for wheat flour and pistachio, respectively), using a total of 2200 images for each neural network.
The performance of the models was tested using blind sets of data comprising 10 per cent of the collected images. Results indicated 99.1 per cent of lentil flour samples containing ground pistachio were correctly classified, while the method was 96.4 per cent accurate in classifying the samples containing wheat flour.
Reference: Pradana-López et al. 2022. Journal of Food Chemistry. DOI: 10.1016/j.foodchem.2022.132832
