analiza aiqbench
This paper addresses the instability in zero-shot composed image retrieval by introducing PEC-CIR, a framework that replaces single-pass generation with a multi-stage Planner-Executor-Critic pipeline. By explicitly evaluating candidate queries against constraints before final selection, the method effectively mitigates semantic distortions and improves precision without requiring additional training data. This approach demonstrates how structured reasoning can enhance reliability in vision-language model applications where exact attribute preservation is critical.
Thinking Before Retrieving: Robust Zero-Shot Composed Image Retrieval via Strategic Planning and Self-Criticism
arXiv:2606.31222v2 Announce Type: replace Abstract: Composed image retrieval requires identifying a target image from a gallery by integrating a reference image with a textual modification instruction. In a training-free zero-shot setting, this task relies on constructing a retrieval-oriented textual query within a frozen vision--language embedding space at inference time. Existing approaches predominantly rely on a single-pass generation strategy that fuses the reference context and modification text into a unified description. This strategy makes it difficult to detect or correct semantic distortions and omissions during generation. Consequently, the preservation of reference attributes and the integration of textual requirements interfere with each other, which degrades retrieval precision. To address these challenges, we introduce PEC-CIR, a training-free framework that structures query construction as a multi-stage reasoning pipeline. The framework operates through a Planner--Executor--Critic architecture where the Planner extracts explicit constraints, the Executor generates multiple candidate target descriptions, and the Critic evaluates these candidates based on constraint compliance. By reframing query construction as a staged inference process instead of a single-pass output, PEC-CIR reduces the propagation of generative errors by explicitly evaluating candidate queries before retrieval, thereby improving retrieval stability.
Abstract:Composed image retrieval requires identifying a target image from a gallery by integrating a reference image with a textual modification instruction. In a training-free zero-shot setting, this task relies on constructing a retrieval-oriented textual query within a frozen vision--language embedding space at inference time. Existing approaches predominantly rely on a single-pass generation strategy that fuses the reference context and modification text into a unified description. This strategy makes it difficult to detect or correct semantic distortions and omissions during generation. Consequently, the preservation of reference attributes and the integration of textual requirements interfere with each other, which degrades retrieval precision. To address these challenges, we introduce PEC-CIR, a training-free framework that structures query construction as a multi-stage reasoning pipeline. The framework operates through a Planner--Executor--Critic architecture where the Planner extracts explicit constraints, the Executor generates multiple candidate target descriptions, and the Critic evaluates these candidates based on constraint compliance. By reframing query construction as a staged inference process instead of a single-pass output, PEC-CIR reduces the propagation of generative errors by explicitly evaluating candidate queries before retrieval, thereby improving retrieval stability.
Artificial Intelligence (cs.AI)
arXiv:2606.31222 [cs.AI]
(or arXiv:2606.31222v2 [cs.AI] for this version)
https://doi.org/10.48550/arXiv.2606.31222 Focus to learn more arXiv-issued DOI via DataCite
arXivLabs is a framework that allows collaborators to develop and share new arXiv features directly on our website.
Both individuals and organizations that work with arXivLabs have embraced and accepted our values of openness, community, excellence, and user data privacy. arXiv is committed to these values and only works with partners that adhere to them.
Have an idea for a project that will add value for arXiv's community? Learn more about arXivLabs.
Text integral sub licența deschisă a publicației · 273 cuvinte
Subiecte conexe
[AINews] Opus 5.5 is good at explainer videos
Latent Space · 29 sept. 2026
What's Continual Learning, and Why Might We Expect To See It In Advanced LLM Agents?
LessWrong · 12 iun. 2026
Does preservation make sense before we know how to revive?
LessWrong · 15 iun. 2026
[AINews] The Future of Latent Space
Latent Space · 25 sept. 2026