Designed an end-to-end transfer learning framework to sense parking availability and predict occupancy.
Worked with limited data inputs and dynamic parking availability sensor data.
Aided by heterogeneous spatio-temporal contextual data from external environment (weather, points of interest, etc) and 35 million+ parking data records from sensors in two different cities.
Framework based on adversarial domain adaptation predicts parking occupancy with accuracy similar to existing state-of-the-art methods.
© 2021 Flora Salim - CRUISE Research Group.
