Low-Cost Multispectral Computational-Optical System in the Near Infrared for the Estimation of Physical-Chemical Properties of Dry Cocoa Beans Using Deep Learning
Description: The project involved the development of a low-cost multispectral computational-optical system in the near-infrared to estimate the physical-chemical properties of dry cocoa beans. It began with the identification of the spectral bands most relevant for cocoa, followed by the design and construction of the multispectral acquisition system.
Personal contribution: I participated in the achievement of the specific objective: “computational estimation of cocoa’s physical-chemical properties using selected near-infrared spectral bands and deep learning algorithms.” This included leading the methodological conception of the evaluation methods, co-directing part of the research team, and supervising the scientific writing and methodology of the results. The approach provided a faster, cost-effective alternative to traditional techniques, enabling efficient cocoa characterization in collaboration with local farmers and the university.