Deep convolutional features for image retrieval

作者：

Highlights：

• A comprehensive study that explores deep convolutional features for CBIR.

• The paper evaluates recently proposed CNNs architectures in image retrieval tasks.

• A plug-n-play approach that uses new architectures of pre-trained CNN’s for CBIR.

• The performance of each network is evaluated using global and local descriptors.

摘要

•A comprehensive study that explores deep convolutional features for CBIR.•The paper evaluates recently proposed CNNs architectures in image retrieval tasks.•A plug-n-play approach that uses new architectures of pre-trained CNN’s for CBIR.•The performance of each network is evaluated using global and local descriptors.

论文关键词：Image retrieval,Deep convolutional features,Deep learning,CNN,Global features,Local features,CBIR

论文评审过程：Received 17 March 2020, Revised 17 February 2021, Accepted 22 March 2021, Available online 29 March 2021, Version of Record 10 April 2021.

论文官网地址：https://doi.org/10.1016/j.eswa.2021.114940