{"url":"/sota/zero-shot-composed-image-retrieval-zs-cir-on-1","task":{"name":"Zero-Shot Composed Image Retrieval (ZS-CIR)","url":"/task/zero-shot-composed-image-retrieval-zs-cir","note":null},"dataset":{"name":"CIRR","url":"/dataset/cirr"},"category":"Computer Vision","categories":["Computer Vision"],"category_note":null,"description":"Given a query composed of a reference image and a relative caption, Composed Image Retrieval (CIR) aims to retrieve target images that are visually similar to the reference one but incorporate the changes specified in the relative caption. The bi-modality of the query provides users with more precise control over the characteristics of the desired image, as some features are more easily described with language, while others can be better expressed visually. \r\n\r\n **Zero-Shot Composed Image Retrieval (ZS-CIR)** is a subtask of CIR that aims to design an approach that manages to combine the reference image and the relative caption without the need for supervised learning.","description_from":"task","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","rank":"the archive's row order at snapshot; not re-ranked","rows_end_at":"2025-07-28","rows_withheld_as_spam":0,"metric_values":"the archive's strings, untouched"},"metrics":["R@1","R@5","R@10","R@50","Rsubset@1"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"R@1":null,"R@5":null,"R@10":null,"R@50":null,"Rsubset@1":null}},"counts":{"rows":47,"rows_with_code":44,"rows_with_paper_page":47,"rows_dated":47,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"CoLLM (finetuned - BLIP-L/16)","metrics":{"R@1":"45.8","R@10":"84.7","R@50":"95.8"},"uses_additional_data":false,"paper_date":"2025-03-25","paper":"/paper/collm-a-large-language-model-for-composed","paper_url":"https://arxiv.org/abs/2503.19910v1","paper_title":"CoLLM: A Large Language Model for Composed Image Retrieval","code":"https://github.com/hmchuong/CoLLM","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"CoVR-BLIP-2","metrics":{"R@1":"43.74","R@10":"83.95","R@5":"73.61","R@50":"96.1"},"uses_additional_data":false,"paper_date":"2023-08-28","paper":"/paper/covr-learning-composed-video-retrieval-from","paper_url":"https://arxiv.org/abs/2308.14746v4","paper_title":"CoVR-2: Automatic Data Construction for Composed Video Retrieval","code":"https://github.com/lucas-ventura/CoVR","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":3,"model":"ImageScope (CLIP-ViT-L/14)","metrics":{"R@1":"39.37","R@10":"78.05","R@5":"67.54","R@50":"92.94"},"uses_additional_data":false,"paper_date":"2025-03-13","paper":"/paper/imagescope-unifying-language-guided-image-1","paper_url":"https://arxiv.org/abs/2503.10166v1","paper_title":"ImageScope: Unifying Language-Guided Image Retrieval via Large Multimodal Model Collective Reasoning","code":"https://github.com/pengfei-luo/ImageScope","n_code_links":1,"syntology":null},{"rank_in_archive_order":4,"model":"IP-CIR + LDRE (CLIP G/14)","metrics":{"R@1":"39.25","R@10":"80","R@5":"70.07","R@50":"94.89"},"uses_additional_data":false,"paper_date":"2024-11-24","paper":"/paper/imagine-and-seek-improving-composed-image","paper_url":"https://arxiv.org/abs/2411.16752v2","paper_title":"Imagine and Seek: Improving Composed Image Retrieval with an Imagined Proxy","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"SEIZE (CLIP G/14)","metrics":{"R@1":"38.87","R@10":"79.42","R@5":"69.42"},"uses_additional_data":false,"paper_date":"2024-10-28","paper":"/paper/semantic-editing-increment-benefits-zero-shot","paper_url":"https://dl.acm.org/doi/10.1145/3664647.3681649","paper_title":"Semantic Editing Increment Benefits Zero-Shot Composed Image Retrieval","code":"https://github.com/yzy-bupt/LDRE","n_code_links":2,"syntology":null},{"rank_in_archive_order":6,"model":"TransAgg (Laion-CIR-Combined)","metrics":{"R@1":"37.87","R@5":"68.88","R@50":"93.86"},"uses_additional_data":false,"paper_date":"2023-06-12","paper":"/paper/zero-shot-composed-text-image-retrieval","paper_url":"https://arxiv.org/abs/2306.07272v2","paper_title":"Zero-shot Composed Text-Image Retrieval","code":"https://github.com/Code-kunkun/ZS-CIR","n_code_links":1,"syntology":null},{"rank_in_archive_order":7,"model":"OSrCIR (CLIP G/14)","metrics":{"R@1":"37.26","R@10":"77.33","R@5":"67.25"},"uses_additional_data":false,"paper_date":"2024-12-15","paper":"/paper/reason-before-retrieve-one-stage-reflective","paper_url":"https://arxiv.org/abs/2412.11077v3","paper_title":"Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval","code":"https://github.com/Pter61/osrcir","n_code_links":1,"syntology":null},{"rank_in_archive_order":8,"model":"SCOT (WACV 2025)","metrics":{"R@1":"36.82","R@10":"74.48","R@5":"64.34","R@50":"93.42"},"uses_additional_data":false,"paper_date":"2025-01-12","paper":"/paper/scot-self-supervised-contrastive-pretraining","paper_url":"https://arxiv.org/abs/2501.08347v1","paper_title":"SCOT: Self-Supervised Contrastive Pretraining For Zero-Shot Compositional Retrieval","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":9,"model":"CoLLM (Pretrained - BLIP-L/16)","metrics":{"R@1":"35.00","R@10":"78.6","R@50":"94.2"},"uses_additional_data":false,"paper_date":"2025-03-25","paper":"/paper/collm-a-large-language-model-for-composed","paper_url":"https://arxiv.org/abs/2503.19910v1","paper_title":"CoLLM: A Large Language Model for Composed Image Retrieval","code":"https://github.com/hmchuong/CoLLM","n_code_links":1,"syntology":null},{"rank_in_archive_order":10,"model":"CIReVL (CLIP G/14)","metrics":{"R@1":"34.65","R@5":"64.29"},"uses_additional_data":false,"paper_date":"2023-10-13","paper":"/paper/vision-by-language-for-training-free","paper_url":"https://arxiv.org/abs/2310.09291v2","paper_title":"Vision-by-Language for Training-Free Compositional Image Retrieval","code":"https://github.com/explainableml/vision_by_language","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":4,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":11,"model":"MagicLens (CoCa L)","metrics":{"R@1":"33.3","R@10":"77.9","R@5":"67.0","R@50":"94.4"},"uses_additional_data":false,"paper_date":"2024-03-28","paper":"/paper/magiclens-self-supervised-image-retrieval","paper_url":"https://arxiv.org/abs/2403.19651v2","paper_title":"MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions","code":"https://github.com/google-deepmind/magiclens","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":12,"model":"WeiMoCIR (CLIP G/14)","metrics":{"R@1":"31.04","R@5":"60.41"},"uses_additional_data":false,"paper_date":"2024-09-07","paper":"/paper/training-free-zs-cir-via-weighted-modality","paper_url":"https://arxiv.org/abs/2409.04918v4","paper_title":"Training-free Zero-shot Composed Image Retrieval via Weighted Modality Fusion and Similarity","code":"https://github.com/whats2000/WeiMoCIR","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":13,"model":"WeiMoCIR (CLIP L/14)","metrics":{"R@1":"30.94","R@5":"60.87"},"uses_additional_data":false,"paper_date":"2024-09-07","paper":"/paper/training-free-zs-cir-via-weighted-modality","paper_url":"https://arxiv.org/abs/2409.04918v4","paper_title":"Training-free Zero-shot Composed Image Retrieval via Weighted Modality Fusion and Similarity","code":"https://github.com/whats2000/WeiMoCIR","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":14,"model":"MagicLens (CLIP L)","metrics":{"R@1":"30.1","R@10":"74.4","R@5":"61.7","R@50":"92.6"},"uses_additional_data":false,"paper_date":"2024-03-28","paper":"/paper/magiclens-self-supervised-image-retrieval","paper_url":"https://arxiv.org/abs/2403.19651v2","paper_title":"MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions","code":"https://github.com/google-deepmind/magiclens","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":15,"model":"IP-CIR + LDRE (CLIP L/14)","metrics":{"R@1":"29.76","R@10":"71.21","R@5":"58.82","R@50":"90.41"},"uses_additional_data":false,"paper_date":"2024-11-24","paper":"/paper/imagine-and-seek-improving-composed-image","paper_url":"https://arxiv.org/abs/2411.16752v2","paper_title":"Imagine and Seek: Improving Composed Image Retrieval with an Imagined Proxy","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":16,"model":"CoLLM (Pretrained - CLIP-L/14)","metrics":{"R@1":"29.7","R@10":"72.8","R@50":"91.5"},"uses_additional_data":false,"paper_date":"2025-03-25","paper":"/paper/collm-a-large-language-model-for-composed","paper_url":"https://arxiv.org/abs/2503.19910v1","paper_title":"CoLLM: A Large Language Model for Composed Image Retrieval","code":"https://github.com/hmchuong/CoLLM","n_code_links":1,"syntology":null},{"rank_in_archive_order":17,"model":"OSrCIR (CLIP L/14)","metrics":{"R@1":"29.45","R@10":"69.86","R@5":"57.68"},"uses_additional_data":false,"paper_date":"2024-12-15","paper":"/paper/reason-before-retrieve-one-stage-reflective","paper_url":"https://arxiv.org/abs/2412.11077v3","paper_title":"Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval","code":"https://github.com/Pter61/osrcir","n_code_links":1,"syntology":null},{"rank_in_archive_order":18,"model":"WeiMoCIR (CLIP H/14)","metrics":{"R@1":"29.11","R@5":"59.76"},"uses_additional_data":false,"paper_date":"2024-09-07","paper":"/paper/training-free-zs-cir-via-weighted-modality","paper_url":"https://arxiv.org/abs/2409.04918v4","paper_title":"Training-free Zero-shot Composed Image Retrieval via Weighted Modality Fusion and Similarity","code":"https://github.com/whats2000/WeiMoCIR","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":19,"model":"WeiMoCIR (CLIP B/32)","metrics":{"R@1":"26.31","R@5":"57.69"},"uses_additional_data":false,"paper_date":"2024-09-07","paper":"/paper/training-free-zs-cir-via-weighted-modality","paper_url":"https://arxiv.org/abs/2409.04918v4","paper_title":"Training-free Zero-shot Composed Image Retrieval via Weighted Modality Fusion and Similarity","code":"https://github.com/whats2000/WeiMoCIR","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":20,"model":"OSrCIR (CLIP B/32)","metrics":{"R@1":"25.42","R@10":"68.19","R@5":"54.54"},"uses_additional_data":false,"paper_date":"2024-12-15","paper":"/paper/reason-before-retrieve-one-stage-reflective","paper_url":"https://arxiv.org/abs/2412.11077v3","paper_title":"Reason-before-Retrieve: One-Stage Reflective Chain-of-Thoughts for Training-Free Zero-Shot Composed Image Retrieval","code":"https://github.com/Pter61/osrcir","n_code_links":1,"syntology":null},{"rank_in_archive_order":21,"model":"CIReVL (CLIP L/14)","metrics":{"R@1":"24.55","R@5":"52.31"},"uses_additional_data":false,"paper_date":"2023-10-13","paper":"/paper/vision-by-language-for-training-free","paper_url":"https://arxiv.org/abs/2310.09291v2","paper_title":"Vision-by-Language for Training-Free Compositional Image Retrieval","code":"https://github.com/explainableml/vision_by_language","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":4,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":22,"model":"CIReVL (CLIP B/32)","metrics":{"R@1":"23.94","R@5":"52.51"},"uses_additional_data":false,"paper_date":"2023-10-13","paper":"/paper/vision-by-language-for-training-free","paper_url":"https://arxiv.org/abs/2310.09291v2","paper_title":"Vision-by-Language for Training-Free Compositional Image Retrieval","code":"https://github.com/explainableml/vision_by_language","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":4,"n_samples":8,"n_pointer_only_licence":0}},{"rank_in_archive_order":23,"model":"RTD + LinCIR (CLIP G/14)","metrics":{"R@5":"67.47"},"uses_additional_data":false,"paper_date":"2024-06-13","paper":"/paper/reducing-task-discrepancy-of-text-encoders","paper_url":"https://arxiv.org/abs/2406.09188v2","paper_title":"An Efficient Post-hoc Framework for Reducing Task Discrepancy of Text Encoders for Composed Image Retrieval","code":"https://github.com/navervision/lincir","n_code_links":1,"syntology":null},{"rank_in_archive_order":24,"model":"LDRE (CLIP G/14)","metrics":{"R@5":"66.39"},"uses_additional_data":false,"paper_date":"2024-07-11","paper":"/paper/ldre-llm-based-divergent-reasoning-and","paper_url":"https://dl.acm.org/doi/10.1145/3626772.3657740","paper_title":"LDRE: LLM-based Divergent Reasoning and Ensemble for Zero-Shot Composed Image Retrieval","code":"https://github.com/yzy-bupt/LDRE","n_code_links":2,"syntology":null},{"rank_in_archive_order":25,"model":"SPN4CIR (SPN-CC)","metrics":{"R@5":"65.42","Rsubset@1":"64.87"},"uses_additional_data":false,"paper_date":"2024-04-17","paper":"/paper/improving-composed-image-retrieval-via","paper_url":"https://arxiv.org/abs/2404.11317v2","paper_title":"Improving Composed Image Retrieval via Contrastive Learning with Scaling Positives and Negatives","code":"https://github.com/BUAADreamer/SPN4CIR","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":0}},{"rank_in_archive_order":26,"model":"LinCIR (CLIP G/14)","metrics":{"R@5":"64.72"},"uses_additional_data":false,"paper_date":"2023-12-04","paper":"/paper/language-only-efficient-training-of-zero-shot","paper_url":"https://arxiv.org/abs/2312.01998v2","paper_title":"Language-only Efficient Training of Zero-shot Composed Image Retrieval","code":"https://github.com/navervision/lincir","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":5,"n_samples":8,"n_pointer_only_licence":8}},{"rank_in_archive_order":27,"model":"MagicLens (CoCa B)","metrics":{"R@5":"64.0"},"uses_additional_data":false,"paper_date":"2024-03-28","paper":"/paper/magiclens-self-supervised-image-retrieval","paper_url":"https://arxiv.org/abs/2403.19651v2","paper_title":"MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions","code":"https://github.com/google-deepmind/magiclens","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":28,"model":"MTCIR (BLIP B/16)","metrics":{"R@5":"58.87"},"uses_additional_data":false,"paper_date":"2023-11-13","paper":"/paper/pretrain-like-you-inference-masked-tuning","paper_url":"https://arxiv.org/abs/2311.07622v3","paper_title":"Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval","code":"https://github.com/Chen-Junyang-cn/PLI","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":29,"model":"MagicLens (CLIP B)","metrics":{"R@5":"58.0"},"uses_additional_data":false,"paper_date":"2024-03-28","paper":"/paper/magiclens-self-supervised-image-retrieval","paper_url":"https://arxiv.org/abs/2403.19651v2","paper_title":"MagicLens: Self-Supervised Image Retrieval with Open-Ended Instructions","code":"https://github.com/google-deepmind/magiclens","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":30,"model":"CompoDiff (CLIP G/14)","metrics":{"R@5":"57.61"},"uses_additional_data":false,"paper_date":"2023-03-21","paper":"/paper/compodiff-versatile-composed-image-retrieval","paper_url":"https://arxiv.org/abs/2303.11916v4","paper_title":"CompoDiff: Versatile Composed Image Retrieval With Latent Diffusion","code":"https://github.com/navervision/compodiff","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":2,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":31,"model":"SEIZE (CLIP B/32)","metrics":{"R@5":"57.42"},"uses_additional_data":false,"paper_date":"2024-10-28","paper":"/paper/semantic-editing-increment-benefits-zero-shot","paper_url":"https://dl.acm.org/doi/10.1145/3664647.3681649","paper_title":"Semantic Editing Increment Benefits Zero-Shot Composed Image Retrieval","code":"https://github.com/yzy-bupt/LDRE","n_code_links":2,"syntology":null},{"rank_in_archive_order":32,"model":"SEIZE (CLIP L/14)","metrics":{"R@5":"57.16"},"uses_additional_data":false,"paper_date":"2024-10-28","paper":"/paper/semantic-editing-increment-benefits-zero-shot","paper_url":"https://dl.acm.org/doi/10.1145/3664647.3681649","paper_title":"Semantic Editing Increment Benefits Zero-Shot Composed Image Retrieval","code":"https://github.com/yzy-bupt/LDRE","n_code_links":2,"syntology":null},{"rank_in_archive_order":33,"model":"RTD + LinCIR (CLIP L/14)","metrics":{"R@5":"56.17"},"uses_additional_data":false,"paper_date":"2024-06-13","paper":"/paper/reducing-task-discrepancy-of-text-encoders","paper_url":"https://arxiv.org/abs/2406.09188v2","paper_title":"An Efficient Post-hoc Framework for Reducing Task Discrepancy of Text Encoders for Composed Image Retrieval","code":"https://github.com/navervision/lincir","n_code_links":1,"syntology":null},{"rank_in_archive_order":34,"model":"iSEARLE (CLIP B/32)","metrics":{"R@5":"55.69"},"uses_additional_data":false,"paper_date":"2024-05-05","paper":"/paper/isearle-improving-textual-inversion-for-zero","paper_url":"https://arxiv.org/abs/2405.02951v1","paper_title":"iSEARLE: Improving Textual Inversion for Zero-Shot Composed Image Retrieval","code":"https://github.com/miccunifi/searle","n_code_links":2,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":35,"model":"LDRE (CLIP L/14)","metrics":{"R@5":"55.57"},"uses_additional_data":false,"paper_date":"2024-07-11","paper":"/paper/ldre-llm-based-divergent-reasoning-and","paper_url":"https://dl.acm.org/doi/10.1145/3626772.3657740","paper_title":"LDRE: LLM-based Divergent Reasoning and Ensemble for Zero-Shot Composed Image Retrieval","code":"https://github.com/yzy-bupt/LDRE","n_code_links":2,"syntology":null},{"rank_in_archive_order":36,"model":"iSEARLE-OTI (CLIP B/32)","metrics":{"R@5":"55.18"},"uses_additional_data":false,"paper_date":"2024-05-05","paper":"/paper/isearle-improving-textual-inversion-for-zero","paper_url":"https://arxiv.org/abs/2405.02951v1","paper_title":"iSEARLE: Improving Textual Inversion for Zero-Shot Composed Image Retrieval","code":"https://github.com/miccunifi/searle","n_code_links":2,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":37,"model":"LDRE (CLIP B/32)","metrics":{"R@5":"55.13"},"uses_additional_data":false,"paper_date":"2024-07-11","paper":"/paper/ldre-llm-based-divergent-reasoning-and","paper_url":"https://dl.acm.org/doi/10.1145/3626772.3657740","paper_title":"LDRE: LLM-based Divergent Reasoning and Ensemble for Zero-Shot Composed Image Retrieval","code":"https://github.com/yzy-bupt/LDRE","n_code_links":2,"syntology":null},{"rank_in_archive_order":38,"model":"Context-I2W (CLIP L/14)","metrics":{"R@5":"55.1"},"uses_additional_data":false,"paper_date":"2023-09-28","paper":"/paper/context-i2w-mapping-images-to-context","paper_url":"https://arxiv.org/abs/2309.16137v2","paper_title":"Context-I2W: Mapping Images to Context-dependent Words for Accurate Zero-Shot Composed Image Retrieval","code":"https://github.com/pter61/context-i2w","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":39,"model":"MTCIR (CLIP L/14)","metrics":{"R@5":"54.58"},"uses_additional_data":false,"paper_date":"2023-11-13","paper":"/paper/pretrain-like-you-inference-masked-tuning","paper_url":"https://arxiv.org/abs/2311.07622v3","paper_title":"Pretrain like Your Inference: Masked Tuning Improves Zero-Shot Composed Image Retrieval","code":"https://github.com/Chen-Junyang-cn/PLI","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":40,"model":"CompoDiff (CLIP L/14)","metrics":{"R@5":"54.36"},"uses_additional_data":false,"paper_date":"2023-03-21","paper":"/paper/compodiff-versatile-composed-image-retrieval","paper_url":"https://arxiv.org/abs/2303.11916v4","paper_title":"CompoDiff: Versatile Composed Image Retrieval With Latent Diffusion","code":"https://github.com/navervision/compodiff","n_code_links":1,"syntology":{"n_ran":4,"n_unverified":2,"n_samples":6,"n_pointer_only_licence":0}},{"rank_in_archive_order":41,"model":"iSEARLE-XL-OTI (CLIP L/14)","metrics":{"R@5":"54.05"},"uses_additional_data":false,"paper_date":"2024-05-05","paper":"/paper/isearle-improving-textual-inversion-for-zero","paper_url":"https://arxiv.org/abs/2405.02951v1","paper_title":"iSEARLE: Improving Textual Inversion for Zero-Shot Composed Image Retrieval","code":"https://github.com/miccunifi/searle","n_code_links":2,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":42,"model":"iSEARLE-XL (CLIP L/14)","metrics":{"R@5":"54.00"},"uses_additional_data":false,"paper_date":"2024-05-05","paper":"/paper/isearle-improving-textual-inversion-for-zero","paper_url":"https://arxiv.org/abs/2405.02951v1","paper_title":"iSEARLE: Improving Textual Inversion for Zero-Shot Composed Image Retrieval","code":"https://github.com/miccunifi/searle","n_code_links":2,"syntology":{"n_ran":3,"n_unverified":1,"n_samples":4,"n_pointer_only_licence":4}},{"rank_in_archive_order":43,"model":"SEARLE","metrics":{"R@5":"53.42"},"uses_additional_data":false,"paper_date":"2023-03-27","paper":"/paper/zero-shot-composed-image-retrieval-with","paper_url":"https://arxiv.org/abs/2303.15247v2","paper_title":"Zero-Shot Composed Image Retrieval with Textual Inversion","code":"https://github.com/miccunifi/searle","n_code_links":2,"syntology":null},{"rank_in_archive_order":44,"model":"LinCIR (CLIP L/14)","metrics":{"R@5":"53.25"},"uses_additional_data":false,"paper_date":"2023-12-04","paper":"/paper/language-only-efficient-training-of-zero-shot","paper_url":"https://arxiv.org/abs/2312.01998v2","paper_title":"Language-only Efficient Training of Zero-shot Composed Image Retrieval","code":"https://github.com/navervision/lincir","n_code_links":1,"syntology":{"n_ran":3,"n_unverified":5,"n_samples":8,"n_pointer_only_licence":8}},{"rank_in_archive_order":45,"model":"SEARLE-XL","metrics":{"R@5":"52.48"},"uses_additional_data":false,"paper_date":"2023-03-27","paper":"/paper/zero-shot-composed-image-retrieval-with","paper_url":"https://arxiv.org/abs/2303.15247v2","paper_title":"Zero-Shot Composed Image Retrieval with Textual Inversion","code":"https://github.com/miccunifi/searle","n_code_links":2,"syntology":null},{"rank_in_archive_order":46,"model":"Pic2Word","metrics":{"R@5":"51.70"},"uses_additional_data":false,"paper_date":"2023-02-06","paper":"/paper/pic2word-mapping-pictures-to-words-for-zero","paper_url":"https://arxiv.org/abs/2302.03084v2","paper_title":"Pic2Word: Mapping Pictures to Words for Zero-shot Composed Image Retrieval","code":"https://github.com/google-research/composed_image_retrieval","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":47,"model":"PALAVRA","metrics":{"R@5":"43.49"},"uses_additional_data":false,"paper_date":"2022-04-04","paper":"/paper/this-is-my-unicorn-fluffy-personalizing","paper_url":"https://arxiv.org/abs/2204.01694v3","paper_title":"\"This is my unicorn, Fluffy\": Personalizing frozen vision-language representations","code":"https://github.com/nvlabs/palavra","n_code_links":2,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":1}}],"since_archive":{"claim":"Results that newer papers report for their own method, placed here by Syntology. A model pointed at the cell in the paper's own table; the number was read from that cell and checked against this leaderboard's metric, dataset, split and scale; an independent check that saw this leaderboard's other rows and every other leaderboard on the same dataset accepted it. Not reviewed by the paper's authors or by the archive's editors, and not ranked against the archive rows.","extraction_file_present":true,"measurement":{"test_papers":883,"papers_with_output":881,"judged_true":108,"judged":110,"wilson95_lower":0.9361,"measured_on":"2026-09-24","frozen_commit":"0e3de0df94"},"measurement_note":"blind adjudication of accepted entries on a held-out split of archive papers, rules frozen before the test","coverage":{"sentence":"Syntology has checked 7,081 of the 9,623 papers on this site that are newer than the archive; results from the others appear after they are checked.","complete":false,"papers_newer_than_archive":9623,"papers_checked":7081,"papers_extracted_not_yet_verified":217,"boards_without_verdict":27,"papers_not_yet_extracted":2325},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-25T09:33:49+00:00","claim":"Per row: N of M harvested code samples from that row's paper executed on a synthesized fixture; the other M-N are unverified. Not a reproduction of the row's number; not a correctness claim. n_pointer_only_licence counts samples the site points at rather than redistributes (a licence axis, independent of ran/unverified).","rows_with_graph_line":26,"rows_with_any_sample_ran":26,"distinct_papers_with_graph_line":12,"distinct_papers_with_any_sample_ran":12,"samples_over_distinct_papers":{"n_ran":27,"n_unverified":14,"n_samples":41,"n_pointer_only_licence":17,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":63,"n_unverified":33,"n_samples":96,"n_pointer_only_licence":41,"note":"row-weighted: a paper behind several rows is counted once per row; inflated relative to samples_over_distinct_papers by design, kept for readers summing the per-row syntology blocks"}}}