Self-supervised Product Quantization for Deep Unsupervised Image Retrieval

TL;DR

This paper introduces Self-supervised Product Quantization (SPQ), a label-free deep learning approach for image retrieval that leverages self-supervised contrastive learning to achieve state-of-the-art results without labeled data.

Contribution

It presents the first deep unsupervised image retrieval method using self-supervised learning and a novel cross-quantized contrastive strategy for joint codeword and descriptor learning.

Findings

Achieves state-of-the-art retrieval performance without labels

Effective in extracting descriptive features for images

Outperforms existing supervised methods on benchmarks

Abstract

Supervised deep learning-based hash and vector quantization are enabling fast and large-scale image retrieval systems. By fully exploiting label annotations, they are achieving outstanding retrieval performances compared to the conventional methods. However, it is painstaking to assign labels precisely for a vast amount of training data, and also, the annotation process is error-prone. To tackle these issues, we propose the first deep unsupervised image retrieval method dubbed Self-supervised Product Quantization (SPQ) network, which is label-free and trained in a self-supervised manner. We design a Cross Quantized Contrastive learning strategy that jointly learns codewords and deep visual descriptors by comparing individually transformed images (views). Our method analyzes the image contents to extract descriptive features, allowing us to understand image representations for accurate retrieval. By conducting extensive experiments on benchmarks, we demonstrate that the proposed method yields state-of-the-art results even without supervised pretraining.