Object Detection and 3D Estimation via an FMCW Radar Using a Fully Convolutional Network

July 2019

tl;dr: Sensor fusion method using radar to estimate the range, doppler, and x and y position of the object in camera.

Overall impression

Two steps in the pipeline: preprocessing (2D FFT and phase normalization) and CNN for detection and angle estimation.

However no physical extent was obtained. This is one area we could focus on.

The paper provides interesting ideas, but not very practical, as there is only one object in the experiment setup.

Key ideas

Technical details