People in each frame were detected in real-time using a pre-trained You Only Look Once (YOLO) model [62] and the results were then published as an MQTT message to the AWS IoT Core.