Performance Comparison of YOLOv5, YOLOv8, and RCNN for Waste Detection and Classification

Waste management in Indonesia, especially in Depok, has pressing issues with environmental and public health impacts. Existing waste management systems struggle to manage the volume and complexity of waste in these urban centres, so technological advancements are vital for improvement. The application of object detection algorithms such as YOLOv8 can improve the efficiency of waste categorization and management. In this study, the performance of the YOLOv8 model in waste classification was compared with other well-known models, such as YOLOv5m and RCNN. In this study, the YOLOv81 model, which was trained for 200 epochs, achieved an F1 score of 0.836 and mAP50 of 0.87, making it the most effective model in terms of detection performance compared to the other YOLOv8 models. Then, compared to RCNN and YOLOv5m, the result is that RCNN excels in accuracy and recall, and YOLOv81 has a better balance, excelling in precise object detection across different categories of waste. This performance underscores the efficiency of YOLOv81, which offers a faster and more reliable solution for real-world waste management applications. With its accuracy, speed and overall performance, YOLOv81 shows significant promise for practical implementation in waste management systems.