# DiCE-CIR: Direct Composition Learning for Efficient Zero-Shot Composed Image Retrieval

arXiv: https://arxiv.org/abs/2607.04665 · HTML: https://mlsanity.com/paper/2607.04665/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 11 of its 23 references that are in the map.
- First posted on arXiv 6 Jul 2026.

## Builds on (most cited first, 11 of 11)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [Image Retrieval on Real-life Images with Pre-trained Vision-and-Language Models](https://mlsanity.com/paper/2108.04024/) (2021)
- [Composing Text and Image for Image Retrieval - An Empirical Odyssey](https://mlsanity.com/paper/1812.07119/) (2018)
- [Zero-Shot Composed Image Retrieval with Textual Inversion](https://mlsanity.com/paper/2303.15247/) (2023)
- [Pic2Word: Mapping Pictures to Words for Zero-shot Composed Image Retrieval](https://mlsanity.com/paper/2302.03084/) (2023)
- [Missing Target-Relevant Information Prediction with World Model for Accurate Zero-Shot Composed Image Retrieval](https://mlsanity.com/paper/2503.17109/) (2025)
- [Fine-grained Textual Inversion Network for Zero-Shot Composed Image Retrieval](https://mlsanity.com/paper/2503.19296/) (2025)
- [Few-Shot Learning with Geometric Constraints](https://mlsanity.com/paper/2003.09151/) (2020)
- [Self-Augmentation: Generalizing Deep Networks to Unseen Classes for Few-Shot Learning](https://mlsanity.com/paper/2004.00251/) (2020)
- [Spatial Reasoning for Few-Shot Object Detection](https://mlsanity.com/paper/2211.01080/) (2022)
- [FAR-Net: Multi-Stage Fusion Network with Enhanced Semantic Alignment and Adaptive Reconciliation for Composed Image Retrieval](https://mlsanity.com/paper/2507.12823/) (2025)

## Abstract (arXiv)

Zero-shot composed image retrieval (ZS-CIR) aims to retrieve a target image from a multimodal query consisting of a reference image and an edit text describing the desired modification. Recent ZS-CIR studies have relied on projection-based methods that map a reference image into pseudo-word tokens in the text embedding space. However, such methods require additional projection and re-encoding steps, increasing training complexity, reducing efficiency, and introducing a discrepancy between training and inference. In this paper, we propose DiCE-CIR, a direct composition learning method that predicts composed query representations by directly composing a reference image and an edit text. To enable scalable training without manually annotated triplets, we automatically construct compositional training samples from large-scale image-caption pairs using a large language model. Based on these samples, we train a lightweight composition module with objectives that promote alignment with the target, edit-consistent semantic transformation, and retrieval discriminability. We conduct extensive experiments on ZS-CIR benchmarks and show that DiCE-CIR achieves state-of-the-art performance on CIRCO and competitive performance on CIRR while maintaining high computational efficiency.