{"url":"/sota/image-text-matching-on-commercialadsdataset","task":{"name":"Image-text matching","url":"/task/image-text-matching","note":null},"dataset":{"name":"CommercialAdsDataset","url":"/dataset/commercialadsdataset"},"category":"Computer Vision","categories":["Computer Vision","Natural Language Processing"],"category_note":null,"description":"**Image-Text Matching** is a subtask within *Cross-Modal Retrieval (CMR)* that involves establishing associations between images and corresponding textual descriptions. The goal is to retrieve an image given a textual query or, conversely, retrieve a textual description given an image query. This task is challenging due to *the heterogeneity gap* between image and text data representations. Image-text matching is used in applications such as content-based image search, visual question answering, and multimodal summarization.\r\n\r\n<span class=\"description-source\">[Assessing Brittleness of Image-Text Retrieval Benchmarks from Vision-Language Models Perspective](https://arxiv.org/abs/2407.15239)</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":["ADD(S) AUC"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"ADD(S) AUC":"higher"}},"counts":{"rows":8,"rows_with_code":6,"rows_with_paper_page":8,"rows_dated":8,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"AlignCMSS","metrics":{"ADD(S) AUC":"91.73"},"uses_additional_data":false,"paper_date":"2023-09-28","paper":"/paper/align-before-search-aligning-ads-image-to","paper_url":"https://arxiv.org/abs/2309.16141v1","paper_title":"Align before Search: Aligning Ads Image to Text for Accurate Cross-Modal Sponsored Search","code":"https://github.com/pter61/aligncmss","n_code_links":1,"syntology":null},{"rank_in_archive_order":2,"model":"VinVL","metrics":{"ADD(S) AUC":"88.56"},"uses_additional_data":false,"paper_date":"2021-01-02","paper":"/paper/vinvl-making-visual-representations-matter-in","paper_url":"https://arxiv.org/abs/2101.00529v2","paper_title":"VinVL: Revisiting Visual Representations in Vision-Language Models","code":"https://github.com/microsoft/Oscar","n_code_links":7,"syntology":{"n_ran":2,"n_unverified":0,"n_samples":2,"n_pointer_only_licence":1}},{"rank_in_archive_order":3,"model":"AdsCVLR","metrics":{"ADD(S) AUC":"87.90"},"uses_additional_data":false,"paper_date":"2022-10-10","paper":"/paper/adscvlr-commercial-visual-linguistic","paper_url":"https://dl.acm.org/doi/abs/10.1145/3503161.3548226","paper_title":"AdsCVLR: Commercial Visual-Linguistic Representation Modeling in Sponsored Search","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":4,"model":"OSCAR","metrics":{"ADD(S) AUC":"87.45"},"uses_additional_data":false,"paper_date":"2020-04-13","paper":"/paper/oscar-object-semantics-aligned-pre-training","paper_url":"https://arxiv.org/abs/2004.06165v5","paper_title":"Oscar: Object-Semantics Aligned Pre-training for Vision-Language Tasks","code":"https://github.com/rmokady/clip_prefix_caption","n_code_links":4,"syntology":{"n_ran":13,"n_unverified":10,"n_samples":23,"n_pointer_only_licence":0}},{"rank_in_archive_order":5,"model":"VL-BERT","metrics":{"ADD(S) AUC":"86.27"},"uses_additional_data":false,"paper_date":"2019-08-22","paper":"/paper/vl-bert-pre-training-of-generic-visual","paper_url":"https://arxiv.org/abs/1908.08530v4","paper_title":"VL-BERT: Pre-training of Generic Visual-Linguistic Representations","code":"https://github.com/jackroos/VL-BERT","n_code_links":3,"syntology":{"n_ran":0,"n_unverified":1,"n_samples":1,"n_pointer_only_licence":0}},{"rank_in_archive_order":6,"model":"BLIP","metrics":{"ADD(S) AUC":"83.51"},"uses_additional_data":false,"paper_date":"2022-01-28","paper":"/paper/blip-bootstrapping-language-image-pre","paper_url":"https://arxiv.org/abs/2201.12086v2","paper_title":"BLIP: Bootstrapping Language-Image Pre-training for Unified Vision-Language Understanding and Generation","code":"https://github.com/huggingface/transformers","n_code_links":9,"syntology":null},{"rank_in_archive_order":7,"model":"Unicoder-VL","metrics":{"ADD(S) AUC":"83.16"},"uses_additional_data":false,"paper_date":"2019-08-16","paper":"/paper/unicoder-vl-a-universal-encoder-for-vision","paper_url":"https://arxiv.org/abs/1908.06066v3","paper_title":"Unicoder-VL: A Universal Encoder for Vision and Language by Cross-modal Pre-training","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":8,"model":"ALBEF","metrics":{"ADD(S) AUC":"82.74"},"uses_additional_data":false,"paper_date":"2021-07-16","paper":"/paper/align-before-fuse-vision-and-language","paper_url":"https://arxiv.org/abs/2107.07651v2","paper_title":"Align before Fuse: Vision and Language Representation Learning with Momentum Distillation","code":"https://github.com/salesforce/lavis","n_code_links":6,"syntology":{"n_ran":4,"n_unverified":1,"n_samples":5,"n_pointer_only_licence":3}}],"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 6,885 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":6885,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2737},"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":4,"rows_with_any_sample_ran":3,"distinct_papers_with_graph_line":4,"distinct_papers_with_any_sample_ran":3,"samples_over_distinct_papers":{"n_ran":19,"n_unverified":12,"n_samples":31,"n_pointer_only_licence":4,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":19,"n_unverified":12,"n_samples":31,"n_pointer_only_licence":4,"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"}}}