Similar picture searching method and device based on deep learning, equipment and medium

Abstract

The invention provides a similar picture searching method and device based on deep learning, equipment and a medium, and the method comprises the steps: extracting feature information of a to-be-searched picture based on a deep learning model; performing dimension reduction processing on the feature information to obtain feature dimension reduction information; performing partition indexing according to the feature dimension reduction information to obtain a feature ID; and querying in a database according to the feature ID to obtain a similar picture corresponding to the picture to be searched. The method is based on a deep learning model, so the efficiency and comprehensiveness of feature extraction are greatly improved; dimension reduction processing is carried out on extracted features, so a data set is easier to use, the calculation overhead of an algorithm is reduced, noise is removed, a result is easy to understand, and the subsequent partition indexing speed is increased; through a partition indexing method, the search speed is greatly improved on the premise of ensuring the search accuracy; the file ID and the picture name are stored through the database, so that the picture positioning process is accelerated, and new picture information can be stored easily.

Description

technical field [0001] The present invention relates to the technical field of computer vision, in particular to a deep learning-based similar image search method, device, equipment and medium. Background technique [0002] With the continuous development of the field of artificial intelligence, especially the continuous improvement of deep learning algorithms and their models, using deep learning models to solve the difficulties and pain points in the field of computer vision is an important development trend in recent years and in the future. With the rapid development of Internet technology, the information generated is growing explosively. The traditional file system cannot meet the needs of massive files for dynamic expansion and efficient access. [0003] Compared with traditional computer vision algorithms, deep learning models can solve practical problems more efficiently and intelligently. However, in the field of similar image search, due to problems such as insuff...

AI Technical Summary

Problems solved by technology

[0004] The present invention provides a similar image search method, device, equipment and medium based on deep learning, which is used to solve the defect of low search efficiency due to insufficient data of similar image search engines in the prior art, thereby improving search efficiency and search accuracy

Images