Devices and methods using machine learning for surveillance and granting of privileges

This invention describes a system that uses two levels of sensing to save power. A first sensor collects basic, low-detail information, like motion data from a gyro or accelerometer, or a general signal, using minimal resources. A machine learning program then analyzes this coarse data to decide if something interesting is happening. If it detects a potential event, a second sensor is activated to collect much more detailed, high-resolution data, specifically optical information, to confirm or further investigate.