Composed Image Retrieval using Contrastive Learning and Task-oriented CLIP-based Features

TL;DR

This paper introduces a novel composed image retrieval method leveraging contrastive learning and task-oriented fine-tuning of CLIP features, significantly improving retrieval accuracy on challenging datasets.

Contribution

It proposes a two-stage approach with task-oriented fine-tuning and a Combiner network to enhance CLIP features for composed image retrieval tasks.

Findings

Outperforms state-of-the-art methods on FashionIQ and CIRR datasets.

Task-oriented fine-tuning improves CLIP feature effectiveness.

The Combiner network effectively integrates bimodal information for retrieval.

Abstract

Given a query composed of a reference image and a relative caption, the Composed Image Retrieval goal is to retrieve images visually similar to the reference one that integrates the modifications expressed by the caption. Given that recent research has demonstrated the efficacy of large-scale vision and language pre-trained (VLP) models in various tasks, we rely on features from the OpenAI CLIP model to tackle the considered task. We initially perform a task-oriented fine-tuning of both CLIP encoders using the element-wise sum of visual and textual features. Then, in the second stage, we train a Combiner network that learns to combine the image-text features integrating the bimodal information and providing combined features used to perform the retrieval. We use contrastive learning in both stages of training. Starting from the bare CLIP features as a baseline, experimental results show that the task-oriented fine-tuning and the carefully crafted Combiner network are highly effective and outperform more complex state-of-the-art approaches on FashionIQ and CIRR, two popular and challenging datasets for composed image retrieval. Code and pre-trained models are available at https://github.com/ABaldrati/CLIP4Cir