<![CDATA[Waste management remains a major environmental challenge in Indonesia, particularly due to the low level of public awareness in sorting waste based on its type. This research aims to design and implement an image-based automatic waste sorting system using the YOLOv5s algorithm with TensorFlow Lite conversion on a Raspberry Pi 3B+. The research was conducted through a system development approach without involving human respondents, focusing on performance evaluation using an image dataset consisting of three categories: paper and tissue, plastic bottles, and cans. The proposed system integrates hardware components, including a camera, servo motors, an ultrasonic sensor, and an LCD, with software components such as YOLOv5s, OpenCV, and TensorFlow Lite. The model performance was evaluated using precision, recall, and mean Average Precision (mAP), while system functionality was assessed through hardware testing. The results show that the model achieved a precision of 0.986, recall of 0.978, and mAP@0.5 of 0.99, indicating excellent detection performance. In addition, the implementation of TensorFlow Lite significantly improved computational efficiency, with the system achieving a processing speed of 173.9 frames per second (FPS). These results demonstrate that the proposed system is capable of performing accurate and efficient real-time waste classification on resource-constrained devices. This research contributes by providing an efficient and practical implementation of real-time waste sorting using a lightweight deep learning model on embedded hardware.]]>
