Garbage classification detection system based on the YOLOv8 algorithm

To solve the problem of diverse types of household garbage that are difficult to classify accurately, a neural convolutional network-based garbage classification system that can be used for validation on unmanned vehicles is proposed. First, the garbage is classified into 44 categories by examining public datasets, which were combined with the actual collected data to build a comprehensive dataset that generated and labeled 15 000 images. Subsequently, the YOLOv8 algorithm is studied, applied to garbage detection and classification, and used for practical verification in unmanned vehicles. Experimental results show that the trained model has a speed of ∼35 f/s and a recognition accuracy of 90%. Finally, due to the real-time performance requirements of unmanned vehicles, the dataset was retrained, and the YOLOv8 weight model was investigated. The model was deployed on the Jetson Xavier NX development platform and integrated into unmanned vehicles to enable automatic garbage recognition and tracking.