Design and implementation of a portable IOT device for assessing quality of green coffee beans using deep learning

Abstract

The existing system for assessing the quality of green coffee beans faces significant challenges. Currently, manual methods introduce delays and subjectivity, delaying the certification process for coffee beans exporters. These delays affect trade flow, and the variability in quality determination undermines the credibility of Ugandan coffee beans in international coffee markets. This project addresses these issues by developing a portable IoT device with deep learning capabilities. This device automates the assessment process, reducing reliance on human judgment. By quickly certifying bean quality, it streamlines trade operations and ensures consistent grading. A Raspberry Pi 4, a DHT11 sensor and a Picamera were integrated to create the portable IoT device. The methodology involves designing the schematic diagram, developing scripts, and interfacing the sensors. Additionally, a user-friendly web app interface is created using User-interface scripts ensuring that the interface is intuitive, efficient, and accessible for users during coffee bean quality assessment.