| CPC G06V 10/751 (2022.01) [G06F 16/532 (2019.01); G06V 10/7715 (2022.01); G06V 20/52 (2022.01); G06V 40/10 (2022.01)] | 12 Claims |
|
1. A person retrieval device for improving person search accuracy based on input images from a plurality of cameras, comprising:
a memory containing program instructions; and
a processor coupled to the memory, wherein the processor is configured to execute the program instructions to:
extract, from an input image, a query person and carrying object image, as a query person-item image, that is an image including the entire body of a person to be retrieved including an image of a carrying object;
extract an image of the carrying object from the query person and carrying object image from the query person-item image, the image of the carrying object being one of a backpack or a rucksack;
calculate a query person image which is the query person-item image after applying processing to a portion, of an image of a query item in the query person-item image so that significant features are not extracted from the image of the query item, by replacing pixel values of the portion with zero;
extract a query person feature amount including a feature amount of the entire body from the query person image; and
compare the query person feature amount with a registered person feature amount of a registered person previously registered in a database based on a pre-trained machine learning model, and determine whether or not the person to be retrieved is same as the registered person.
|