Post-harvest quality assessment for leek with hyperspectral imaging and synthetically created data

Food security and food waste remain a worldwide challenge. Hyperspectral imaging can provide non-destructive quality measurements for high-value perishable crops. With digital technology and artificial intelligence, we can assess quality in a transparent and reliable way and test best practises for resource optimization. During the 2021 growing season, the project monitored a fertilisation and green manure trail with Remotely-piloted aircraft system (RPAS) -mounted multi- and hyperspectral sensors. Post-harvest, a trained expert scored the leeks and we used Specim FX10 and FX17 hyperspectral imaging sensors to analyse spectral reflectance of the leeks. Followed with a storage experiment in which we assessed the shelf life quality of the leeks repeatedly. With the combination of pre- and postharvest hyperspectral imaging, we were able to predict at harvest the quality of deterioration of the leeks during storage and provide a quality estimation on a field level. Further we investigate the use of synthetically created hyperspectral data to improve the accuracy of the predictions.