Research on garbage detection and classification based on YOLOv8

Abstract

In order to improve the identification efficiency of garbage cleaning, we improve the identification speed and accuracy by studying the data and models of garbage classification and detection, so as to achieve higher efficiency of garbage identification and cleaning. This can avoid making the robot unable to play its role due to unclear recognition. We chose YOLOv8 model for training and built a complete model system. The network supports multi-object recognition, has a simple network structure and training method, and improves the object detection speed. After 100 rounds of training, we finally got an accuracy of ${8 6 . 8 8 \%}$. The YOLOv8 network model constructed in this paper has efficient detection ability and can identify the types of garbage with high accuracy.