Enabling Efficient Royal Quinoa Quality Inspection via Mobile-based Foreign Body Detection

Abstract

Due to its high nutritional value, quinoa has gained recognition as a valuable global superfood, particularly after the United Nations declared 2013 the "International Year of Quinoa." Royal quinoa, a variety exclusive to the Bolivian altiplano, has proven to be the richest in nutritional properties compared with other quinoa types. However, royal quinoa production has been hampered by the lack of efficient tools for separating quinoa from impurities such as straw, clods, and stones, requiring intense manual labor. To address this challenge, this paper introduces a novel inspection system for foreign body detection in quinoa. The system incorporates a conveyor belt, a hopper, a cellphone tripod and an Android app to automate the transportation and inspection of quinoa grains. The core contribution of this research lies in the development of a mobile-based foreign body detector utilizing a Xenvo Pro Lens Kit to capture minute visual features. Experimental results demonstrate that YOLOv8s achieved the highest performance with an IoU of 0.89. This research holds significant potential to revolutionize quinoa production by show-casing new possibilities for machine vision systems in this domain. The source code, device design, and dataset are available at: https://github.com/EdwinTSalcedo/QuinoaInspector