Image Retrieval Algorithm Based on Distribution Entropy Gain Loss Function

Abstract

The invention discloses an image retrieval algorithm based on a distribution entropy gain loss function. The algorithm uses a pre-training network for initialization, trains the network according to the requirements of an image retrieval task, and uses a self-designed distribution entropy gain loss function when training the network. The accuracy of image retrieval is improved; the distribution entropy gain loss function combines the contrast loss function and relative entropy to enhance the accuracy of image similarity measurement when training the network; the contrast loss function calculates the similarity between features through Euclidean distance, The relative entropy can be used to measure the distribution difference between feature vectors, and the relative entropy is added to the contrast loss function to improve the feature vector similarity measure; use the distribution entropy gain loss function to train the network model, and adjust the network parameters to obtain a more suitable image for the image. The network model of the retrieval task, the trained network model achieved better retrieval effect in the image retrieval experiment.

Description

technical field [0001] The invention belongs to the technical field of image retrieval, and relates to an image retrieval algorithm for training a network through a distributed entropy gain loss function. Background technique [0002] With the vigorous development of Internet technology, social software is rich and diverse, and various forms of multimedia information are flooding our lives. How to quickly and accurately capture and effectively utilize multimedia information has become an important research topic, which has attracted extensive attention in the academic circles. Under this trend, image retrieval technology has been fully and comprehensively developed. [0003] In recent years, with the successful application of neural networks to image classification, researchers have paid more and more attention to the application of neural networks in the field of image retrieval. A large number of studies have shown that the features output by the convolutional layers of n...

AI Technical Summary

Problems solved by technology

Although the pre-trained network has achieved amazing retrieval performance, it often does not have network parameters that match the image retrieval task, so image retrieval network fine-tuning has become a hot research topic

Images