Search papers, labs, and topics across Lattice.
This paper introduces a hierarchical multi-agent framework (PDF) for Zero-Shot Compositional Image Retrieval (ZS-CIR) that addresses perception myopia and logic drift in existing methods. PDF uses an Intent Routing Manager to create a candidate pool and a Decision Manager with a Training-free Reasoning Policy Distillation mechanism and Tournament-style Test-Time Scaling (TTS) for self-evolving reasoning. Experiments on CIRR, CIRCO, and FashionIQ datasets show that PDF achieves state-of-the-art performance, demonstrating the effectiveness of experience-driven self-evolution and TTS.
Zero-shot image retrieval gets a major boost from a hierarchical multi-agent system that evolves its reasoning at test time, outperforming previous methods on standard benchmarks.
Zero-Shot Compositional Image Retrieval (ZS-CIR) requires both preserving the visual continuity of the reference image and faithfully executing the semantic variables specified in the modification text, which constitutes the core challenge of the task. Existing methods often suffer from Perception Myopia in a single space, or fall into Logic Drift in iterative collaboration due to the perception ceiling of the underlying retriever. To address this issue, we propose a one-stop hierarchical Perception-to-Deliberation Framework (PDF), which, to the best of our knowledge, is the first to introduce experience self-evolution and Test-Time Scaling Law (TTS) into ZS-CIR. Relying on a hierarchical multi-agent architecture, PDF first utilizes an Intent Routing Manager to dynamically dispatch multi-view Worker perception signals based on modification intents to construct a high-recall candidate pool. Subsequently, the Decision Manager combines a Training-free Reasoning Policy Distillation mechanism with a Tournament-style TTS strategy to achieve self-evolving fine-grained reasoning, yielding the final retrieval results. Experimental results demonstrate that PDF achieves SOTA performance on three benchmark datasets: CIRR, CIRCO, and FashionIQ. This study indicates that experience-driven self-evolution and TTS represent a highly promising and scalable path for achieving zero-shot fine-grained multimedia retrieval. The code will be made publicly available upon acceptance.
