A preliminary evaluation of a low-cost multispectral sensor for non-destructive evaluation of olive fruits’ fat content

This study presents a preliminary evaluation of a low-cost multispectral device for the non-destructive assessment of olive fruits’ fat content. The developed device integrates a multispectral sensor, with a spectral response of 18 channels falling in a range from 410 to 940 nm, a calibrated light source, and a programmable board, in a ‘gun’-shaped device whose trigger activates sample reading. The device was used to measure 50 intact olive samples, which were subsequently chemically analysed to determine their actual fat content. Then, the multispectral readings from the 18 channels were used as input variables to train a neural network, using the actual fat content registers as reference data. The measured results, in terms of root-mean-square-error and coefficient of determination, shows promising capabilities of the developed low-cost device in the prediction of fat content of intact olives, what stands up for further development and experimentation.