Abstract: A method of radar detection of targets in a region of interest, comprises storing boundaries of the region of interest, represented in a world reference system of coordinates. A camera fixed to the radar sensor captures an image of the environment, where a calibration object like a chessboard is identified, and based on its position in the captured image, the position and orientation of the radar sensor are determined in the world reference system. The environment is scanned, to determine target positions in a radar reference system of coordinates, centered in the radar sensor. Based on this, the target positions are transferred from the radar reference system to the world reference system, and are compared with the stored boundaries of the region of interest.

Abstract: A radar sensor scans an environment to identify, for each target, a candidate target volume. A camera, fixed to the radar sensor, captures an image of the environment, and runs an image recognition algorithm to identify target boundaries for each target. The radar sensor determines a refined target volume, by selecting and keeping a first part of the candidate target volume located inside the target boundaries and removing a second part of the candidate target volume located outside the target boundaries.