Integrating Listwise Ranking into Pairwise-based Image-Text Retrieval

TL;DR

This paper introduces a listwise ranking method for image-text retrieval that enhances existing pairwise models by optimizing the entire relevance list, leading to improved retrieval performance and more user-friendly results.

Contribution

It proposes a novel listwise ranking approach with a differentiable NDCG loss, integrated into pairwise ITR models to better reflect relevance among negative samples.

Findings

Improved retrieval accuracy on benchmark datasets.

Enhanced ranking quality with relevance-aware listwise optimization.

Plug-and-play integration with existing models.

Abstract

Image-Text Retrieval (ITR) is essentially a ranking problem. Given a query caption, the goal is to rank candidate images by relevance, from large to small. The current ITR datasets are constructed in a pairwise manner. Image-text pairs are annotated as positive or negative. Correspondingly, ITR models mainly use pairwise losses, such as triplet loss, to learn to rank. Pairwise-based ITR increases positive pair similarity while decreasing negative pair similarity indiscriminately. However, the relevance between dissimilar negative pairs is different. Pairwise annotations cannot reflect this difference in relevance. In the current datasets, pairwise annotations miss many correlations. There are many potential positive pairs among the pairs labeled as negative. Pairwise-based ITR can only rank positive samples before negative samples, but cannot rank negative samples by relevance. In this paper, we integrate listwise ranking into conventional pairwise-based ITR. Listwise ranking optimizes the entire ranking list based on relevance scores. Specifically, we first propose a Relevance Score Calculation (RSC) module to calculate the relevance score of the entire ranked list. Then we choose the ranking metric, Normalized Discounted Cumulative Gain (NDCG), as the optimization objective. We transform the non-differentiable NDCG into a differentiable listwise loss, named Smooth-NDCG (S-NDCG). Our listwise ranking approach can be plug-and-play integrated into current pairwise-based ITR models. Experiments on ITR benchmarks show that integrating listwise ranking can improve the performance of current ITR models and provide more user-friendly retrieval results. The code is available at https://github.com/AAA-Zheng/Listwise_ITR.