{"url":"/sota/cross-modal-retrieval-on-flickr30k","task":{"name":"Cross-Modal Retrieval","url":"/task/cross-modal-retrieval","note":null},"dataset":{"name":"Flickr30k","url":"/dataset/flickr30k"},"category":"Computer Vision","categories":["Computer Vision","Natural Language Processing"],"category_note":null,"description":"**Cross-Modal Retrieval (CMR)** is a task of retrieving items across different modalities, such as image, text, video, and audio. The core challenge of CMR is the *heterogeneity gap*, which arises because data from different modalities have distinct representations, making direct comparison difficult.  To address this, most CMR methods focus on learning a *shared latent embedding space*. In this space, concepts from different modalities are projected, allowing their similarity to be measured using a distance metric.\r\n\r\n\r\n<span class=\"description-source\">[Scene-centric vs. Object-centric Image-Text Cross-modal Retrieval: A Reproducibility Study](https://arxiv.org/abs/2301.05174)</span>","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":["Image-to-text R@1","Image-to-text R@5","Image-to-text R@10","Text-to-image R@1","Text-to-image R@5","Text-to-image R@10"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"Image-to-text R@1":null,"Image-to-text R@5":null,"Image-to-text R@10":null,"Text-to-image R@1":null,"Text-to-image R@5":null,"Text-to-image R@10":null}},"counts":{"rows":27,"rows_with_code":22,"rows_with_paper_page":25,"rows_dated":25,"rows_using_additional_data":13},"rows":[{"rank_in_archive_order":1,"model":"X2-VLM (large)","metrics":{"Image-to-text R@1":"98.8","Image-to-text R@10":"100","Image-to-text R@5":"100","Text-to-image R@1":"91.8","Text-to-image R@10":"99.5","Text-to-image R@5":"98.6"},"uses_additional_data":true,"paper_date":"2022-11-22","paper":"/paper/x-2-vlm-all-in-one-pre-trained-model-for","paper_url":"https://arxiv.org/abs/2211.12402v2","paper_title":"X$^2$-VLM: All-In-One Pre-trained Model For Vision-Language Tasks","code":"https://github.com/zengyan-97/x-vlm","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":4,"n_samples":6,"n_pointer_only_licence":6}},{"rank_in_archive_order":2,"model":"X2-VLM (base)","metrics":{"Image-to-text R@1":"98.5","Image-to-text R@10":"100","Image-to-text R@5":"100","Text-to-image R@1":"90.4","Text-to-image R@10":"99.3","Text-to-image R@5":"98.2"},"uses_additional_data":true,"paper_date":"2022-11-22","paper":"/paper/x-2-vlm-all-in-one-pre-trained-model-for","paper_url":"https://arxiv.org/abs/2211.12402v2","paper_title":"X$^2$-VLM: All-In-One Pre-trained Model For Vision-Language Tasks","code":"https://github.com/zengyan-97/x-vlm","n_code_links":2,"syntology":{"n_ran":2,"n_unverified":4,"n_samples":6,"n_pointer_only_licence":6}},{"rank_in_archive_order":3,"model":"BEiT-3","metrics":{"Image-to-text R@1":"98.0","Image-to-text R@10":"100.0","Image-to-text R@5":"100.0","Text-to-image R@1":"90.3","Text-to-image R@10":"99.5","Text-to-image R@5":"98.7"},"uses_additional_data":true,"paper_date":"2022-08-22","paper":"/paper/image-as-a-foreign-language-beit-pretraining","paper_url":"https://arxiv.org/abs/2208.10442v2","paper_title":"Image as a Foreign Language: BEiT Pretraining for All Vision and Vision-Language Tasks","code":"https://github.com/microsoft/unilm/tree/master/beit","n_code_links":2,"syntology":null},{"rank_in_archive_order":4,"model":"OmniVL (14M)","metrics":{"Image-to-text R@1":"97.3","Image-to-text R@10":"100","Image-to-text R@5":"99.9","Text-to-image R@1":"87.9","Text-to-image R@10":"99.1","Text-to-image R@5":"97.8"},"uses_additional_data":true,"paper_date":"2022-09-15","paper":"/paper/omnivl-one-foundation-model-for-image","paper_url":"https://arxiv.org/abs/2209.07526v2","paper_title":"OmniVL:One Foundation Model for Image-Language and Video-Language Tasks","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"ERNIE-ViL 2.0","metrics":{"Image-to-text R@1":"97.2","Image-to-text R@10":"100.0","Image-to-text R@5":"100.0","Text-to-image R@1":"93.3","Text-to-image R@10":"99.8","Text-to-image R@5":"99.4"},"uses_additional_data":true,"paper_date":"2022-09-30","paper":"/paper/ernie-vil-2-0-multi-view-contrastive-learning","paper_url":"https://arxiv.org/abs/2209.15270v1","paper_title":"ERNIE-ViL 2.0: Multi-view Contrastive Learning for Image-Text Pre-training","code":"https://github.com/PaddlePaddle/ERNIE","n_code_links":1,"syntology":null},{"rank_in_archive_order":6,"model":"Aurora (ours, r=128)","metrics":{"Image-to-text R@1":"97.2","Image-to-text R@10":"100","Image-to-text R@5":"100","Text-to-image R@1":"86.8","Text-to-image R@10":"98.9","Text-to-image R@5":"97.6"},"uses_additional_data":true,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":7,"model":"X-VLM (base)","metrics":{"Image-to-text R@1":"97.1","Image-to-text R@10":"100.0","Image-to-text R@5":"100.0","Text-to-image R@1":"86.9","Text-to-image R@10":"98.7","Text-to-image R@5":"97.3"},"uses_additional_data":true,"paper_date":"2021-11-16","paper":"/paper/multi-grained-vision-language-pre-training","paper_url":"https://arxiv.org/abs/2111.08276v3","paper_title":"Multi-Grained Vision Language Pre-Training: Aligning Texts with Visual Concepts","code":"https://github.com/zengyan-97/x-vlm","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":0,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":8,"model":"VSE-Gradient","metrics":{"Image-to-text R@1":"97.0","Image-to-text R@10":"100","Image-to-text R@5":"99.6","Text-to-image R@1":"86.3","Text-to-image R@10":"99.0","Text-to-image R@5":"97.4"},"uses_additional_data":true,"paper_date":"2022-10-21","paper":"/paper/dissecting-deep-metric-learning-losses-for","paper_url":"https://arxiv.org/abs/2210.13188v1","paper_title":"Dissecting Deep Metric Learning Losses for Image-Text Retrieval","code":"https://github.com/microsoft/VSE_Gradient","n_code_links":2,"syntology":null},{"rank_in_archive_order":9,"model":"ALIGN","metrics":{"Image-to-text R@1":"95.3","Image-to-text R@10":"100","Image-to-text R@5":"99.8","Text-to-image R@1":"84.9","Text-to-image R@10":"98.6","Text-to-image R@5":"97.4"},"uses_additional_data":true,"paper_date":"2021-02-11","paper":"/paper/scaling-up-visual-and-vision-language","paper_url":"https://arxiv.org/abs/2102.05918v2","paper_title":"Scaling Up Visual and Vision-Language Representation Learning With Noisy Text Supervision","code":"https://github.com/facebookresearch/metaclip","n_code_links":5,"syntology":{"n_ran":8,"n_unverified":2,"n_samples":10,"n_pointer_only_licence":9}},{"rank_in_archive_order":10,"model":"ViSTA","metrics":{"Image-to-text R@1":"89.5","Image-to-text R@10":"99.6","Image-to-text R@5":"98.4","Text-to-image R@1":"75.8","Text-to-image R@10":"96.9","Text-to-image R@5":"94.2"},"uses_additional_data":true,"paper_date":"2022-03-31","paper":"/paper/vista-vision-and-scene-text-aggregation-for","paper_url":"https://arxiv.org/abs/2203.16778v1","paper_title":"ViSTA: Vision and Scene Text Aggregation for Cross-Modal Retrieval","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":11,"model":"IAIS","metrics":{"Image-to-text R@1":"88.3","Image-to-text R@10":"99.4","Image-to-text R@5":"98.4","Text-to-image R@1":"76.86","Text-to-image R@10":"95.72","Text-to-image R@5":"93.3"},"uses_additional_data":true,"paper_date":"2021-05-28","paper":"/paper/learning-relation-alignment-for-calibrated","paper_url":"https://arxiv.org/abs/2105.13868v2","paper_title":"Learning Relation Alignment for Calibrated Cross-modal Retrieval","code":"https://github.com/lancopku/IAIS","n_code_links":1,"syntology":{"n_ran":1,"n_unverified":1,"n_samples":2,"n_pointer_only_licence":0}},{"rank_in_archive_order":12,"model":"3SHNet","metrics":{"Image-to-text R@1":"87.1","Image-to-text R@10":"99.2","Image-to-text R@5":"98.2","Text-to-image R@1":"69.5","Text-to-image R@10":"94.7","Text-to-image R@5":"91.0"},"uses_additional_data":false,"paper_date":"2024-04-26","paper":"/paper/3shnet-boosting-image-sentence-retrieval-via","paper_url":"https://arxiv.org/abs/2404.17273v1","paper_title":"3SHNet: Boosting Image-Sentence Retrieval via Visual Semantic-Spatial Self-Highlighting","code":"https://github.com/xurige1995/3shnet","n_code_links":1,"syntology":{"n_ran":10,"n_unverified":0,"n_samples":10,"n_pointer_only_licence":0}},{"rank_in_archive_order":13,"model":"ViLT-B/32","metrics":{"Image-to-text R@1":"83.5","Image-to-text R@10":"98.6","Image-to-text R@5":"96.7","Text-to-image R@1":"64.4","Text-to-image R@10":"93.8","Text-to-image R@5":"88.7"},"uses_additional_data":true,"paper_date":"2021-02-05","paper":"/paper/vilt-vision-and-language-transformer-without","paper_url":"https://arxiv.org/abs/2102.03334v2","paper_title":"ViLT: Vision-and-Language Transformer Without Convolution or Region Supervision","code":"https://github.com/huggingface/transformers","n_code_links":6,"syntology":{"n_ran":1,"n_unverified":3,"n_samples":4,"n_pointer_only_licence":1}},{"rank_in_archive_order":14,"model":"DSMD","metrics":{"Image-to-text R@1":"82.5","Image-to-text R@10":"97.7","Image-to-text R@5":"95.5","Text-to-image R@1":"68.4","Text-to-image R@10":"94.4","Text-to-image R@5":"90.8"},"uses_additional_data":false,"paper_date":"2024-04-16","paper":"/paper/dynamic-self-adaptive-multiscale-distillation","paper_url":"https://arxiv.org/abs/2404.10838v1","paper_title":"Dynamic Self-adaptive Multiscale Distillation from Pre-trained Multimodal Large Model for Efficient Cross-modal Representation Learning","code":"https://github.com/chrisx599/dsmd","n_code_links":1,"syntology":null},{"rank_in_archive_order":15,"model":"RCAR","metrics":{"Image-to-text R@1":"82.3","Image-to-text R@10":"98.4","Image-to-text R@5":"96.0","Text-to-image R@1":"62.6","Text-to-image R@10":"91.1","Text-to-image R@5":"85.8"},"uses_additional_data":false,"paper_date":"2023-03-23","paper":"/paper/plug-and-play-regulators-for-image-text","paper_url":"https://arxiv.org/abs/2303.13371v1","paper_title":"Plug-and-Play Regulators for Image-Text Matching","code":"https://github.com/paranioar/rcar","n_code_links":1,"syntology":null},{"rank_in_archive_order":16,"model":"NAPReg","metrics":{"Image-to-text R@1":"79.6","Text-to-image R@1":"60.0"},"uses_additional_data":false,"paper_date":"2023-01-07","paper":"/paper/napreg-nouns-as-proxies-regularization-for","paper_url":"https://openaccess.thecvf.com/content/WACV2023/html/Jawade_NAPReg_Nouns_As_Proxies_Regularization_for_Semantically_Aware_Cross-Modal_Embeddings_WACV_2023_paper.html","paper_title":"NAPReg: Nouns As Proxies Regularization for Semantically Aware Cross-Modal Embeddings","code":"https://github.com/bhavinjawade/NAPReq","n_code_links":1,"syntology":null},{"rank_in_archive_order":17,"model":"SGRAF","metrics":{"Image-to-text R@1":"77.8","Image-to-text R@10":"97.4","Image-to-text R@5":"94.1","Text-to-image R@1":"58.5","Text-to-image R@10":"88.8","Text-to-image R@5":"83.0"},"uses_additional_data":false,"paper_date":"2021-01-05","paper":"/paper/similarity-reasoning-and-filtration-for-image","paper_url":"https://arxiv.org/abs/2101.01368v1","paper_title":"Similarity Reasoning and Filtration for Image-Text Matching","code":"https://github.com/Paranioar/SGRAF","n_code_links":1,"syntology":{"n_ran":10,"n_unverified":2,"n_samples":12,"n_pointer_only_licence":12}},{"rank_in_archive_order":18,"model":"GSMN","metrics":{"Image-to-text R@1":"76.4","Image-to-text R@10":"97.3","Image-to-text R@5":"94.3","Text-to-image R@1":"57.4","Text-to-image R@10":"89.0","Text-to-image R@5":"82.3"},"uses_additional_data":false,"paper_date":"2020-04-01","paper":"/paper/graph-structured-network-for-image-text","paper_url":"https://arxiv.org/abs/2004.00277v1","paper_title":"Graph Structured Network for Image-Text Matching","code":"https://github.com/CrossmodalGroup/GSMN","n_code_links":1,"syntology":null},{"rank_in_archive_order":19,"model":"Pearl","metrics":{"Image-to-text R@1":"75.3","Image-to-text R@10":"97.3","Image-to-text R@5":"93.4","Text-to-image R@1":"54.98","Text-to-image R@10":"88.26","Text-to-image R@5":"81.3"},"uses_additional_data":false,"paper_date":null,"paper":null,"paper_url":null,"paper_title":"","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":20,"model":"IMRAM","metrics":{"Image-to-text R@1":"74.1","Image-to-text R@10":"96.6","Image-to-text R@5":"93.0","Text-to-image R@1":"53.9","Text-to-image R@10":"87.2","Text-to-image R@5":"79.4"},"uses_additional_data":false,"paper_date":"2020-03-08","paper":"/paper/imram-iterative-matching-with-recurrent","paper_url":"https://arxiv.org/abs/2003.03772v1","paper_title":"IMRAM: Iterative Matching with Recurrent Attention Memory for Cross-Modal Image-Text Retrieval","code":"https://github.com/HuiChen24/IMRAM","n_code_links":1,"syntology":{"n_ran":2,"n_unverified":1,"n_samples":3,"n_pointer_only_licence":3}},{"rank_in_archive_order":21,"model":"SCAN","metrics":{"Image-to-text R@1":"67.4","Image-to-text R@10":"95.8","Image-to-text R@5":"90.3","Text-to-image R@1":"48.6","Text-to-image R@10":"85.2","Text-to-image R@5":"77.7"},"uses_additional_data":false,"paper_date":"2018-03-21","paper":"/paper/stacked-cross-attention-for-image-text","paper_url":"http://arxiv.org/abs/1803.08024v2","paper_title":"Stacked Cross Attention for Image-Text Matching","code":"https://github.com/kuanghuei/SCAN","n_code_links":6,"syntology":{"n_ran":7,"n_unverified":9,"n_samples":16,"n_pointer_only_licence":1}},{"rank_in_archive_order":22,"model":"Dual-Path\n  (ResNet)","metrics":{"Image-to-text R@1":"55.6","Image-to-text R@5":"81.9"},"uses_additional_data":false,"paper_date":"2017-11-15","paper":"/paper/dual-path-convolutional-image-text-embedding","paper_url":"https://arxiv.org/abs/1711.05535v4","paper_title":"Dual-Path Convolutional Image-Text Embeddings with Instance Loss","code":"https://github.com/layumi/Image-Text-Embedding","n_code_links":2,"syntology":null},{"rank_in_archive_order":23,"model":"SCO\n  (ResNet)","metrics":{"Image-to-text R@1":"55.5","Image-to-text R@10":"89.3","Image-to-text R@5":"82.0","Text-to-image R@1":"41.1","Text-to-image R@10":"80.1","Text-to-image R@5":"70.5"},"uses_additional_data":false,"paper_date":"2017-12-06","paper":"/paper/learning-semantic-concepts-and-order-for","paper_url":"http://arxiv.org/abs/1712.02036v1","paper_title":"Learning Semantic Concepts and Order for Image and Sentence Matching","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":24,"model":"VSE++\n  (ResNet)","metrics":{"Image-to-text R@1":"52.9","Image-to-text R@10":"87.2","Image-to-text R@5":"80.5","Text-to-image R@1":"39.6","Text-to-image R@10":"79.5","Text-to-image R@5":"70.1"},"uses_additional_data":false,"paper_date":"2017-07-18","paper":"/paper/vse-improving-visual-semantic-embeddings-with","paper_url":"http://arxiv.org/abs/1707.05612v4","paper_title":"VSE++: Improving Visual-Semantic Embeddings with Hard Negatives","code":"https://github.com/fartashf/vsepp","n_code_links":10,"syntology":{"n_ran":2,"n_unverified":3,"n_samples":5,"n_pointer_only_licence":4}},{"rank_in_archive_order":25,"model":"CMPL\n  (ResNet)","metrics":{"Image-to-text R@1":"49.6","Image-to-text R@10":"86.1","Image-to-text R@5":"76.8","Text-to-image R@1":"37.3","Text-to-image R@10":"75.5","Text-to-image R@5":"65.7"},"uses_additional_data":false,"paper_date":"2018-09-01","paper":"/paper/deep-cross-modal-projection-learning-for","paper_url":"http://openaccess.thecvf.com/content_ECCV_2018/html/Ying_Zhang_Deep_Cross-Modal_Projection_ECCV_2018_paper.html","paper_title":"Deep Cross-Modal Projection Learning for Image-Text Matching","code":"https://github.com/YingZhangDUT/Cross-Modal-Projection-Learning","n_code_links":1,"syntology":null},{"rank_in_archive_order":26,"model":"Dual-Path (ResNet)","metrics":{"Image-to-text R@10":"89.5","Text-to-image R@1":"39.1","Text-to-image R@10":"80.9","Text-to-image R@5":"69.2"},"uses_additional_data":false,"paper_date":"2017-11-15","paper":"/paper/dual-path-convolutional-image-text-embedding","paper_url":"https://arxiv.org/abs/1711.05535v4","paper_title":"Dual-Path Convolutional Image-Text Embeddings with Instance Loss","code":"https://github.com/layumi/Image-Text-Embedding","n_code_links":2,"syntology":null},{"rank_in_archive_order":27,"model":"VAST","metrics":{"Text-to-image R@1":"91.0","Text-to-image R@10":"99.5","Text-to-image R@5":"98.5"},"uses_additional_data":true,"paper_date":"2023-05-29","paper":"/paper/vast-a-vision-audio-subtitle-text-omni-1","paper_url":"https://arxiv.org/abs/2305.18500v2","paper_title":"VAST: A Vision-Audio-Subtitle-Text Omni-Modality Foundation Model and Dataset","code":"https://github.com/TXH-mercury/VALOR","n_code_links":2,"syntology":{"n_ran":15,"n_unverified":27,"n_samples":42,"n_pointer_only_licence":0}}],"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 6,264 of the 9,581 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":9581,"papers_checked":6264,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":3316},"order":"newest first by month (arXiv date, else the arXiv-id month), then arXiv id descending","columns":[],"entries":[]},"syntology":{"read_at":"2026-09-24T18:15:14+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":12,"rows_with_any_sample_ran":12,"distinct_papers_with_graph_line":11,"distinct_papers_with_any_sample_ran":11,"samples_over_distinct_papers":{"n_ran":59,"n_unverified":52,"n_samples":111,"n_pointer_only_licence":36,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":61,"n_unverified":56,"n_samples":117,"n_pointer_only_licence":42,"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"}}}