Abstract
Optical imaging systems offer a non-destructive and efficient solution for detecting defects and diseases in apples. The integration of visual and spectral imagers can enable early detection of visible and invisible post-harvest disorders such as superficial scald and infections caused by pathogens like Colletotrichum godetia. In this work, we present an innovative and automated experimental setup that collects hyperspectral images, spectral signatures, and high-quality RGB images of the entire skin of the apples. The system ensures precise fruit alignment, enabling highly reproducible measurements and consistent tracking of disease progression over time. Using this experimental setup, we acquired data from 1,754 apples across various cultivars and disease conditions, resulting in a high-quality, multi-purpose dataset. In this work, we also included a preliminary analysis on a subset of the dataset to demonstrate the system’s potential for disease assessment and monitoring, focusing on the automatic estimation of superficial scald severity and the progression of Colletotrichum godetiae in the Granny Smith variety. Ongoing data acquisition over the next two years will support the development of more robust algorithms and broaden the scope of the analysis. By integrating data from the multiple sensors we expect to be able to detect also symptoms of other diseases at early stages.