Towards a Calibration-free Approach to Deep Learning based Single-incidence Inverse Scattering

State-of-the-art Deep Learning based methods for inverse scattering assume the knowledge of transmitter and receiver locations. However, this requires a calibration stage which involves the careful placement of the transmitters and receivers at specific known locations or placing the transmitters and receivers at arbitrary locations and using a system to calculate their respective positions. This can be tedious and reduces the ease of usability of the system. In this work, we consider single-incidence inverse scattering where the transmitter location is unknown. We will demonstrate that even when the transmitter location is unknown, simultaneous contrast reconstruction and transmitter localization can be achieved using Deep Learning models.