Agribot: Artificial Intelligence of Things System to Classify and Predict the Quality of Produce in Smart Agriculture

Abstract

Manual inspection of agricultural produce is labour-intensive, costly, and prone to human error. In this project Agribot, an autonomous produce sorting and remote monitoring system that combines Artificial Intelligence (AI) and Internet of Things (IoT) technologies, is developed to classify produce quality, automate sorting, and collect data for crop performance insights. The system comprises a sorting machine and a computer vision system integrated with a deep learning model. A YOLOv5 object detection model was trained for autonomous produce inspection. The backend system employs serverless architecture, with inspection data stored in cloud storage and presented to users via a mobile application. Results show that Agribot's inspection model achieved accuracies of 85.2%, 79.8%, 80.6%, and 79.8% for grades A, B, C, and rejected guava, respectively. Agribot successfully sorts guavas into different containers based on their grades and displays inspection results on a mobile application. This system aims to replace manual inspection with an autonomous process, potentially reducing labor costs and inconsistencies in sorting while providing valuable data to farmers.

Access to Document

Other files and links

• Link to publication in Scopus

Fingerprint