UniCVR: From Alignment to Reranking for Unified Zero-Shot Composed Visual Retrieval

Fuente: arXiv
Salvato in:
Dettagli Bibliografici
Autori principali: Wen, Haokun, Song, Xuemeng, Zhang, Haoyu, Zhao, Xiangyu, Guan, Weili, Nie, Liqiang
Natura: Preprint
Pubblicazione: 2026
Soggetti:
Accesso online:
Tags: Aggiungi Tag
Nessun Tag, puoi essere il primo ad aggiungerne!!
_version_ 1866911614149591040
author Wen, Haokun
Song, Xuemeng
Zhang, Haoyu
Zhao, Xiangyu
Guan, Weili
Nie, Liqiang
author_facet Wen, Haokun
Song, Xuemeng
Zhang, Haoyu
Zhao, Xiangyu
Guan, Weili
Nie, Liqiang
contents Composed image retrieval, multi-turn composed image retrieval, and composed video retrieval all share a common paradigm: composing the reference visual with modification text to retrieve the desired target. Despite this shared structure, the three tasks have been studied in isolation, with no prior work proposing a unified framework, let alone a zero-shot solution. In this paper, we propose UniCVR, the first unified zero-shot composed visual retrieval framework that jointly addresses all three tasks without any task-specific human-annotated data. UniCVR strategically combines two complementary strengths: Multimodal Large Language Models (MLLMs) for compositional query understanding and Vision-Language Pre-trained (VLP) models for structured visual retrieval. Concretely, UniCVR operates in two stages. In Stage I, we train the MLLM as a compositional query embedder via contrastive learning on a curated multi-source dataset of approximately 3.5M samples, bridging the heterogeneous embedding spaces between the MLLM and the frozen VLP gallery encoder. A cluster-based hard negative sampling strategy is proposed to strengthen contrastive supervision. In Stage II, we introduce an MLLM-guided dual-level reranking mechanism that applies adaptive budgeted subset scoring to a small number of top-ranked candidates, and then exploits the resulting relevance signals through a dual-level re-scoring scheme, producing more accurate final rankings with minimal computational overhead. Extensive experiments across five benchmarks covering all three tasks demonstrate that UniCVR achieves cutting-edge performance, validating its effectiveness and generalizability. Our data and code will be released upon acceptance.
format Preprint
id arxiv_https___arxiv_org_abs_2604_20318
institution arXiv
publishDate 2026
record_format arxiv
spellingShingle UniCVR: From Alignment to Reranking for Unified Zero-Shot Composed Visual Retrieval
Wen, Haokun
Song, Xuemeng
Zhang, Haoyu
Zhao, Xiangyu
Guan, Weili
Nie, Liqiang
Computer Vision and Pattern Recognition
Multimedia
Composed image retrieval, multi-turn composed image retrieval, and composed video retrieval all share a common paradigm: composing the reference visual with modification text to retrieve the desired target. Despite this shared structure, the three tasks have been studied in isolation, with no prior work proposing a unified framework, let alone a zero-shot solution. In this paper, we propose UniCVR, the first unified zero-shot composed visual retrieval framework that jointly addresses all three tasks without any task-specific human-annotated data. UniCVR strategically combines two complementary strengths: Multimodal Large Language Models (MLLMs) for compositional query understanding and Vision-Language Pre-trained (VLP) models for structured visual retrieval. Concretely, UniCVR operates in two stages. In Stage I, we train the MLLM as a compositional query embedder via contrastive learning on a curated multi-source dataset of approximately 3.5M samples, bridging the heterogeneous embedding spaces between the MLLM and the frozen VLP gallery encoder. A cluster-based hard negative sampling strategy is proposed to strengthen contrastive supervision. In Stage II, we introduce an MLLM-guided dual-level reranking mechanism that applies adaptive budgeted subset scoring to a small number of top-ranked candidates, and then exploits the resulting relevance signals through a dual-level re-scoring scheme, producing more accurate final rankings with minimal computational overhead. Extensive experiments across five benchmarks covering all three tasks demonstrate that UniCVR achieves cutting-edge performance, validating its effectiveness and generalizability. Our data and code will be released upon acceptance.
title UniCVR: From Alignment to Reranking for Unified Zero-Shot Composed Visual Retrieval
topic Computer Vision and Pattern Recognition
Multimedia
url https://arxiv.org/abs/2604.20318