Method and system for fast retrieval of target image based on artificial intelligence

Abstract

The invention discloses an artificial intelligence-based fast retrieval method and system for a target image, which acquires a template image and several known labels corresponding to the template image; extracts an image to be detected from a target image database; and combines the image to be detected and the template image Input into the trained convolutional neural network, output the hash code of the image to be detected and the hash code of the template image; based on the Hamming distance between the hash code of the image to be detected and the hash code of the template image, get The similarity between the image to be detected and the template image, select one or more images to be detected whose similarity is higher than the set threshold as the retrieval result output. Through the use of artificial intelligence technology, the image samples in complex scenes collected by the robot vision platform are based on the convolutional neural network, and the hash method is used to extract image features, and the introduction of distinguishing confusing entities, optimizing similarity relations and distinguishing sample attention, Better respond to item retrieval in complex scenarios.

Description

technical field [0001] The invention relates to the technical field of image retrieval, in particular to an artificial intelligence-based fast retrieval method and system for target images. Background technique [0002] The statements in this section merely mention the background technology related to the present invention and do not necessarily constitute the prior art. [0003] Item retrieval aims to use computers or robots to process, analyze and understand images captured by cameras to identify targets and objects in various modes. It is an important research topic in the field of computer vision. [0004] Nowadays, robots can be used to collect images of real environments. For simple images, it is easy to learn a suitable feature representation to distinguish them from samples with different semantics. In complex scenes, images need more attention to get a proper feature representation. For complex scenarios, for example, in multi-label learning (images contain multip...

AI Technical Summary

Problems solved by technology

Therefore, popular item retrieval methods treat all samples equally, resulting in relatively poor generalization performance of item retrieval methods in complex scenarios.

[0005] (1) Item retrieval in complex scenarios contains a large number of confusing entities. These entities generally have similar feature representations, and popular item retrieval methods cannot distinguish them (without considering confusing features);

[0006] (2) Item retrieval in complex scenes requires more accurate image similarity in order to mine the real similarity relationship of images to guide the generation of image features. Existing item retrieval methods do not consider the similarity relationship of mining images;

[0007] (3) Item retrieval in complex scenes needs to provide more attention to complex samples and divide attention reasonably, but existing item retrieval methods treat all samples equally

Images