| CPC G06V 10/22 (2022.01) [G06F 16/9014 (2019.01); G06V 2201/07 (2022.01)] | 13 Claims |
|
1. An object recognition apparatus comprising:
a memory storing instructions; and
one or more processors configured to execute the instructions to:
store a table in which, for each of a plurality of first objects, hash values for feature amounts of a plurality of first feature points for the first object are associated with the first object;
detect object regions of a plurality of second objects arranged on a shelf, from an input image;
extract, for each object region, feature amounts of second feature points within the object region;
determine, for each object region, hash values for the feature amounts of the second feature points within the object region;
determine, for each object region, a number of the feature amounts of the second feature points within the object region for which the hash values are associated with the first objects within the table;
identify which of the second objects are candidate objects, based on the determined numbers for the object regions of the second objects;
recognize, each second object that is a candidate object by matching the second feature points within the object region of the second object with the first feature points of the first objects;
discard, as an erroneous detection, a recognition result acquired for any candidate object for which the object region differs in area by a predetermined value or more from another object region on a same stage of the shelf; and
output the recognition result for each candidate object that has not been discarded.
|