Method, apparatus and robot for out-of-stock detection of shelf commodities

Abstract

The disclosure discloses a method, an apparatus and a robot for out-of-stock detection of shelf commodity. The method includes: acquiring a plurality of shelf RGB images and a plurality of shelf depth images; executing an out-of-stock area detection step to obtain all vacant areas in the shelf for shelf RGB images and shelf depth images; stitching the shelf RGB images into a complete shelf image, merging and de-duplicating on all vacant areas and detection boxes of commodities, shelf label and layer lattice, and mapping them to the complete shelf image; generating a shed grid area corresponding to each commodity for the complete shelf image, and dividing all vacant areas in the shelf based on the shed grid area; and determining an out-of-stock detection result of each commodity based on the position information and the depth information of the vacant area and the shed grid area.