Draft Design of Fruit Object Recognition using Transfer Learning in Smart Farm

Abstract

Agriculture can save labor and production costs by automatically recognizing and growing fruit. And the technology that can complete this process is in AI. Using AI technology, we designed a Fruit Object Detection and Monitoring to increase the efficiency of fruit cultivation management, an important task in the agricultural industry. For this, Yolo, transfer learning algorithms and ROS were studied. After that, a Fruit Object Detection and Monitoring was designed by linking a Raspberry Pi 4 equipped with a camera and Arduino, a micro cloud storage cluster and a micro cloud AI cluster. Until now, the design has been tested except for real-time object recognition monitoring, and is planned to be completed through future research.