Improving on-campus security measures to ensure the well-being of students and staff results in a significant enhancement in the overall quality of life. This research proposes an Internet of Things (IoT)-based system that leverages sound recognition to detect distress screams and quickly notify a central unit. The system uses an IoT device, Arduino, to collect data from the environment, processes it, and sends the information to a central unit via a cloud IoT service over a Low Power Wide Area Network. The system uses The Things Network and Amazon Web Services platforms to enable communication. The system design considers the resource limitations of Arduino devices and low-power infrastructure. To achieve local detection within the Arduino, several Convolutional Neural Network architectures were compared and evaluated for their effectiveness in scream detection. We evaluated the performance of our models based on accuracy and F1 score, achieving our best results with an accuracy of 95
keywords = {Campus security, Internet of things, Machine learning, Scream detection