<![CDATA[The implementation of smoke-free area regulation in campus environment faces challenges in terms of supervision and enforcement. This study aims to determine the performance of deep learning-based object detection algorithm, namely Faster R-CNN, in detecting smoking activity in the campus area of Universitas Islam Sultan Agung (UNISSULA). The dataset used consists of 1935 annotated images of smoking activity obtained from Roboflow, with data division of 85% training and 15% validation. The model was trained using Google Colab and tested based on evaluation metrics such as accuracy, precision, recall, and f1-score. The results show that Faster R-CNN has superior performance with the best evaluation value reaching 100% at a threshold of 0.5. These findings conclude that Faster R-CNN is suitable for use in a smoking activity detection system in a campus environment, especially in the context of detection accuracy and consistency.]]>
Copyrights © 2025
