Non-destructive internal disorder detection of Conference pears by semantic segmentation of X-ray CT scans using deep learning

Long term storage is required to deliver high quality pear fruit year-round. Under suboptimal conditions, internal disorders, such as browning and cavity formation, can develop are often invisible from the outside. We present a non-destructive inspection method quantify disorders in X-ray CT scans of using deep neural network for semantic segmentation. Herein, U-net based model was trained automatically indicate healthy tissue, core regions affected by i.e., formation browning. The quantitative data resulting segmentations used measure severity disorders. Excellent classification accuracies 99.4 92.2% were obtained “consumable” vs “non-consumable” on one hand “healthy” “defect but consumable” other hand. identification showed be most difficult.