{"ID":5551907,"CreatedAt":"2026-07-02T01:54:51.863792489Z","UpdatedAt":"2026-07-04T01:45:22.703757252Z","DeletedAt":null,"paper_url":"https://arxiv.org/abs/2607.00374","arxiv_id":"2607.00374","title":"Learning to Compose: Revisiting Proxy Task Design for Zero-Shot Composed Image Retrieval","abstract":"Composed Image Retrieval (CIR) retrieves a target image from a reference image and a textual modification. While supervised CIR relies on costly triplets, Zero-Shot CIR (ZS-CIR) alleviates this reliance through proxy tasks trained on image-text pairs. However, existing proxy tasks primarily enhance visual and textual representations to accommodate a predefined composition mechanism such as pseudo-word injection into a frozen text encoder or linear feature arithmetic. As a result, the composition function itself remains unlearned, limiting the model's ability to express diverse and fine-grained semantic modifications. To address this, we propose FoCo, which models composition as two coordinated stages: focusing on modification-relevant visual content, and then completing the target semantics. We realize these through two proxy tasks: text-anchored visual aggregation to selectively gather visual content guided by localized textual semantics, and context-conditioned semantic completion to transform these aggregated visuals with the remaining scene context into a coherent composed representation. The tasks are trained jointly with a cross-instance contrastive objective, encouraging semantic diversity and discouraging shortcut composition strategies. Extensive experiments on four ZS-CIR benchmarks show FoCo's state-of-the-art performance and improved generalization.","short_abstract":"Composed Image Retrieval (CIR) retrieves a target image from a reference image and a textual modification. While supervised CIR relies on costly triplets, Zero-Shot CIR (ZS-CIR) alleviates this reliance through proxy tasks trained on image-text pairs. However, existing proxy tasks primarily enhance visual and textual r...","url_abs":"https://arxiv.org/abs/2607.00374","url_pdf":"https://arxiv.org/pdf/2607.00374v1","authors":"[\"Jingjing Zhang\",\"Lei Zhang\",\"Zheren Fu\",\"Zhendong Mao\"]","published":"2026-07-01T03:20:06Z","proceeding":"cs.CV","tasks":"[\"cs.CV\",\"cs.AI\",\"cs.CL\",\"cs.IR\",\"cs.MM\"]","methods":"[]","has_code":false}