We demonstrate an occupancy detection and human activity monitoring system using low-cost motion sensing, edge computing and wireless networking devices. We design a dual Doppler sensor to achieve high signal sensitivity and wide sensing range. We develop signal filtering, detection and machine learning algorithms on an embedded computer to generate classification result in real-time. We also implement web services to enable users to access signal and room state via a wireless network. Compared with conventional occupancy sensors, the dual Doppler system has higher detection rate, and also has the capability of detecting activities even for multiple people.