Multi-resonance and transfer-learning based tracking system

A method comprising:

• positioning a sensor array under a coverage area, wherein the sensor array comprises a plurality of sensing resonators tuned at different resonance frequencies, where an output frequency response of the sensor array varies as a function of a location of a target object or a shape of the target object within the coverage area;
• generating a continuous wave signal over a frequency range;
• electromagnetically coupling a reading coil element to the sensor array, wherein the reading coil element comprises a conductive loop;
• applying the continuous wave signal to the reading coil element;
• acquiring reflection coefficient data of the reading coil element showing changes in the output frequency response of the sensor array from the reading coil element;
• and
• predicting, by a control unit device using machine learning, a location of the target object within the coverage area or a behavior of the target object based on the frequency spectrum data acquired from the reflection coefficient data of the reading coil element,
• wherein the machine learning comprises a transfer learning technique with dual data inputs in a training phase and only frequency spectrum data in a prediction phase, wherein the dual data inputs comprise the frequency spectrum data and video data.

A non-transitory computer readable storage medium having instructions stored therein, wherein the instructions, when executed by a processor, cause the processor to:

• generate a continuous wave signal over a frequency range;
• apply the continuous wave signal to a reading coil element that is electromagnetically coupled to a sensor array, wherein the sensor array comprises a plurality of sensing resonators tuned at different resonance frequencies, where an output frequency response of the sensor array varies as a function of a location of a target object or a shape of the target object within a coverage area of the sensor array;
• acquire reflection coefficient data of the reading coil element showing changes in the output frequency response of the sensor array from the reading coil element;
• and
• predict, by a control unit device using machine learning, a location of the target object within the coverage area or a behavior of the target object based on frequency spectrum data acquired from the reflection coefficient data of the reading coil element,
• wherein the machine learning comprises a transfer learning technique with dual data inputs in a training phase and only frequency spectrum data in a prediction phase, wherein the dual data inputs comprise the frequency spectrum data and video data.