Feature extraction model training method and device, sample retrieval method and device and computer equipment

Abstract

The invention relates to a feature extraction model training method and device, a sample retrieval method and device, computer equipment and a storage medium. The method comprises the following steps: inputting each sample in a first sample group into an initial feature extraction model to obtain an initial classification feature, an initial semantic feature and an initial fusion feature; the first sample group comprises a target sample, a corresponding reference sample and a category label of the sample, and the initial feature extraction model comprises a sample classification network, a non-semantic feature extraction network and a feature fusion network; obtaining classification loss based on the initial classification features and the category labels corresponding to the same sample; obtaining feature loss based on other features corresponding to the target sample and the reference sample; and model parameters of the initial feature extraction model are adjusted based on the feature loss and the classification loss until convergence conditions are met, a target feature extraction model used for extracting sample features of the input sample is obtained, and the sample features are used for sample retrieval. By adopting the method, the model training efficiency can be improved.

Description

Technical field [0001] The present application relates to the field of computer technology, in particular to a feature extraction model training, a sample retrieval method, a device, a computer device, and a storage medium. Background technique [0002] With the development of computer technology, retrieval technology, for example, image retrieval technology. The retrieval technology is a sample feature stored in the sample search library by extracting the characteristics of the query sample, which matches the sample features stored in the sample search library to retrieve the samples similar to the query sample in the sample search library. [0003] In the conventional technique, the sample input model is typically extracted, and one model is used to extract a feature, so that different features need to be trained different models, training consumption. Inventive content [0004] Based on this, it is necessary for the above technical problems to provide a feature extraction mod...

AI Technical Summary

Problems solved by technology

[0003] In the traditional technology, the sample input model is usually used for feature extraction, and one model is used to extract a feature. In this way, different models need to be trained for different features, and the training is time-consuming.

Images