We can't find the internet
Attempting to reconnect
Something went wrong!
Hang in there while we get back on track
Development of a real-time plastic waste detection system based on deep learning to support the automation of industrial waste sorting processes
Summary
A YOLO11-Nano deep learning model trained on the WasteIn dataset achieved 91.67% accuracy in real-time plastic waste detection via smartphone cameras with under 100ms inference time, enabling mobile-based waste sorting support. This technology directly addresses microplastic pollution at the source — automated identification and sorting of plastic waste streams at scale can prevent degradation of macroplastics into the microplastic particles that ultimately contaminate ecosystems and food chains.
The accumulation of plastic waste has become one of the major environmental issues in Indonesia, where conventional waste management systems are still limited in handling and classifying various types of waste. This research aims to develop an automatic waste detection system using Artificial Intelligence (AI) and implement it in a mobile application capable of identifying plastic waste in real time. The model was trained using the WasteIn dataset, which contains annotated images of different waste categories, including plastic, paper, glass, metal, organic, and electronic waste. The YOLO11-Nano architecture was applied due to its lightweight structure and efficiency for mobile-based deployment. The trained model was then converted into TensorFlow Lite (TFLite) format and integrated into an Android Studio environment to enable real-time inference through smartphone cameras. Based on the evaluation of 36 test images, the system achieved an accuracy of 91.67%, with consistent performance in detecting plastic, paper, and organic waste. The inference time of less than 100 milliseconds per frame demonstrates the system’s feasibility for real-time mobile applications. The results indicate that the integration of deep learning and computer vision technologies can effectively support waste classification processes and contribute to sustainable waste management practices.