A Study on YOLOv4 Algorithm for Harvesting Agriculture Robot

Smart agriculture is understood as agriculture applying high technology, production technology, product safety protection, management technology, product identification in series associated with artificial intelligence system. Key technologies used in smart agriculture applications such as: IoT technology with sensors that collect accurate data on climate, growing conditions, and health of crops/animals; intelligent automation technology with robots and unmanned aerial vehicles gradually replacing humans in farming activities; image processing, artificial intelligence (AI). This paper proposes an improved YOLOv4 algorithm to real-time detect tomato for harvesting agriculture robot in Robot Operating System (ROS). The advantage of this model is the use of the combination of data tomato detection and real experiment system to increase the accuracy. To verify the effectiveness of this method, an actual applications, tomato with different ripeness of tomato in the same area can be detected with the accuracy of 89%.