# From Transformation to Target State: Rethinking Query Representation for Zero-Shot Composed Image Retrieval

arXiv: https://arxiv.org/abs/2610.05993 · HTML: https://mlsanity.com/paper/2610.05993/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Placed in Multimodal reasoning, one of 17 directions in LLM reasoning.
- Builds on 32 of its 63 references that are in the map.
- First posted on arXiv 5 Oct 2026.

## Placed in

- LLM reasoning / [Multimodal reasoning](https://mlsanity.com/lr/multimodal-reasoning/)

## Builds on (most cited first, 30 of 32)

- [Learning Transferable Visual Models From Natural Language Supervision](https://mlsanity.com/paper/2103.00020/) (2021)
- [WinCLIP: Zero-/Few-Shot Anomaly Classification and Segmentation](https://mlsanity.com/paper/2303.14814/) (2023)
- [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)
- [Automatic Spatially-aware Fashion Concept Discovery](https://mlsanity.com/paper/1708.01311/) (2017)
- [From Images to Textual Prompts: Zero-shot VQA with Frozen Large Language Models](https://mlsanity.com/paper/2212.10846/) (2022)
- [Vision-by-Language for Training-Free Compositional Image Retrieval](https://mlsanity.com/paper/2310.09291/) (2023)
- [On the test-time zero-shot generalization of vision-language models: Do we really need prompt learning?](https://mlsanity.com/paper/2405.02266/) (2024)
- [CLIPPO: Image-and-Language Understanding from Pixels Only](https://mlsanity.com/paper/2212.08045/) (2022)
- [Improved Zero-Shot Classification by Adapting VLMs with Text Descriptions](https://mlsanity.com/paper/2401.02460/) (2024)
- [A Hard-to-Beat Baseline for Training-free CLIP-based Adaptation](https://mlsanity.com/paper/2402.04087/) (2024)
- [Improving Zero-shot Generalization and Robustness of Multi-modal Models](https://mlsanity.com/paper/2212.01758/) (2022)
- [FAME-ViL: Multi-Tasking Vision-Language Model for Heterogeneous Fashion Tasks](https://mlsanity.com/paper/2303.02483/) (2023)
- [Decomposed Soft Prompt Guided Fusion Enhancing for Compositional Zero-Shot Learning](https://mlsanity.com/paper/2211.10681/) (2022)
- [Transductive Zero-Shot and Few-Shot CLIP](https://mlsanity.com/paper/2405.18437/) (2024)
- [Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval](https://mlsanity.com/paper/2412.11077/) (2024)
- [Troika: Multi-Path Cross-Modal Traction for Compositional Zero-Shot Learning](https://mlsanity.com/paper/2303.15230/) (2023)
- [Missing Target-Relevant Information Prediction with World Model for Accurate Zero-Shot Composed Image Retrieval](https://mlsanity.com/paper/2503.17109/) (2025)
- [Air-Know: Arbiter-Calibrated Knowledge-Internalizing Robust Network for Composed Image Retrieval](https://mlsanity.com/paper/2604.19386/) (2026)
- [PerceptionCLIP: Visual Classification by Inferring and Conditioning on Contexts](https://mlsanity.com/paper/2308.01313/) (2023)
- [CoTMR: Chain-of-Thought Multi-Scale Reasoning for Training-Free Zero-Shot Composed Image Retrieval](https://mlsanity.com/paper/2502.20826/) (2025)
- [CLIPure: Purification in Latent Space via CLIP for Adversarially Robust Zero-Shot Classification](https://mlsanity.com/paper/2502.18176/) (2025)
- [SyncMask: Synchronized Attentional Masking for Fashion-centric Vision-Language Pretraining](https://mlsanity.com/paper/2404.01156/) (2024)
- [VTQA: Visual Text Question Answering via Entity Alignment and Cross-Media Reasoning](https://mlsanity.com/paper/2303.02635/) (2023)
- [WISER: Wider Search, Deeper Thinking, and Adaptive Fusion for Training-Free Zero-Shot Composed Image Retrieval](https://mlsanity.com/paper/2602.23029/) (2026)
- [Data-Efficient Generalization for Zero-shot Composed Image Retrieval](https://mlsanity.com/paper/2503.05204/) (2025)
- [G-MIXER: Geodesic Mixup-based Implicit Semantic Expansion and Explicit Semantic Re-ranking for Zero-Shot Composed Image Retrieval](https://mlsanity.com/paper/2604.14710/) (2026)
- [STiTch: Semantic Transition and Transportation in Collaboration for Training-Free Zero-Shot Composed Image Retrieval](https://mlsanity.com/paper/2605.21261/) (2026)

## Abstract (arXiv)

Composed image retrieval (CIR) aims to retrieve a desired target image from a query consisting of a reference image and a modification text. This task exhibits an unusual representational asymmetry: the modification text specifies a transition from the reference state, whereas retrieval candidates depict completed target states. This creates a representation mismatch for zero-shot methods that query pretrained vision-language spaces directly with transformation-oriented language. We study this mismatch and reformulate zero-shot composed image retrieval as target-state reconstruction followed by retrieval. We instantiate this formulation with ASAP-CIR, a training-free framework that reconstructs a static target representation using a frozen multimodal large language model (MLLM). The representation combines multiple holistic descriptions with a variable set of importance-weighted atomic semantics, thereby preserving both overall target identity and fine-grained visual constraints. Retrieval then integrates holistic state alignment, atomic constraint grounding, and calibrated target-state evidence aggregation. A controlled text-only diagnostic shows that target-side static query formulations achieve more reliable retrieval than dynamic composed query formulations, particularly when source-state semantics must be suppressed or transformed. Experiments on FashionIQ, CIRR, and CIRCO further characterize the effectiveness and limitations of this representation principle, with the clearest gains on the multi-target CIRCO benchmark. These results show that how composed intent is represented before retrieval is a consequential design choice, distinct from the choice of retrieval backbone itself.