{"url":"/sota/image-captioning-on-nocaps-val-in-domain","task":{"name":"Image Captioning","url":"/task/image-captioning","note":null},"dataset":{"name":"nocaps-val-in-domain","url":"/dataset/nocaps"},"category":"Computer Vision","categories":["Computer Vision","Natural Language Processing"],"category_note":null,"description":"**Image Captioning** is the task of describing the content of an image in words. This task lies at the intersection of computer vision and natural language processing. Most image captioning systems use an encoder-decoder framework, where an input image is encoded into an intermediate representation of the information in the image, and then decoded into a descriptive text sequence. The most popular benchmarks are nocaps and COCO, and models are typically evaluated according to a BLEU or CIDER metric.\r\n\r\n( Image credit: [Reflective Decoding Network for Image Captioning, ICCV'19](https://openaccess.thecvf.com/content_ICCV_2019/papers/Ke_Reflective_Decoding_Network_for_Image_Captioning_ICCV_2019_paper.pdf))","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":["CIDEr","SPICE","Pre-train (#images)"],"metric_direction":{"note":"inferred from the metric name only (the archive records no direction); null = not inferred, chart draws points only","by_metric":{"CIDEr":null,"SPICE":null,"Pre-train (#images)":null}},"counts":{"rows":11,"rows_with_code":8,"rows_with_paper_page":11,"rows_dated":11,"rows_using_additional_data":0},"rows":[{"rank_in_archive_order":1,"model":"BLIP-2 ViT-G FlanT5 XL (zero-shot)","metrics":{"CIDEr":"123.7","Pre-train (#images)":"1.1B","SPICE":"16.3"},"uses_additional_data":false,"paper_date":"2023-01-30","paper":"/paper/blip-2-bootstrapping-language-image-pre","paper_url":"https://arxiv.org/abs/2301.12597v3","paper_title":"BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models","code":"https://github.com/huggingface/transformers","n_code_links":17,"syntology":{"n_ran":4,"n_unverified":4,"n_samples":8,"n_pointer_only_licence":1}},{"rank_in_archive_order":2,"model":"BLIP-2 ViT-G OPT 6.7B (zero-shot)","metrics":{"CIDEr":"123.7","Pre-train (#images)":"1.1B","SPICE":"15.8"},"uses_additional_data":false,"paper_date":"2023-01-30","paper":"/paper/blip-2-bootstrapping-language-image-pre","paper_url":"https://arxiv.org/abs/2301.12597v3","paper_title":"BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models","code":"https://github.com/huggingface/transformers","n_code_links":17,"syntology":{"n_ran":4,"n_unverified":4,"n_samples":8,"n_pointer_only_licence":1}},{"rank_in_archive_order":3,"model":"BLIP-2 ViT-G OPT 2.7B (zero-shot)","metrics":{"CIDEr":"123","Pre-train (#images)":"1.1B","SPICE":"15.8"},"uses_additional_data":false,"paper_date":"2023-01-30","paper":"/paper/blip-2-bootstrapping-language-image-pre","paper_url":"https://arxiv.org/abs/2301.12597v3","paper_title":"BLIP-2: Bootstrapping Language-Image Pre-training with Frozen Image Encoders and Large Language Models","code":"https://github.com/huggingface/transformers","n_code_links":17,"syntology":{"n_ran":4,"n_unverified":4,"n_samples":8,"n_pointer_only_licence":1}},{"rank_in_archive_order":4,"model":"LEMON_large","metrics":{"CIDEr":"116.9","Pre-train (#images)":" 200M","SPICE":"15.8"},"uses_additional_data":false,"paper_date":"2021-11-24","paper":"/paper/scaling-up-vision-language-pre-training-for","paper_url":"https://arxiv.org/abs/2111.12233v2","paper_title":"Scaling Up Vision-Language Pre-training for Image Captioning","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":5,"model":"BLIP_ViT-L","metrics":{"CIDEr":"114.9","Pre-train (#images)":"129M","SPICE":"15.2"},"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":6,"model":"SimVLM","metrics":{"CIDEr":"113.7","Pre-train (#images)":"1.8B","SPICE":"-"},"uses_additional_data":false,"paper_date":"2021-08-24","paper":"/paper/simvlm-simple-visual-language-model","paper_url":"https://arxiv.org/abs/2108.10904v3","paper_title":"SimVLM: Simple Visual Language Model Pretraining with Weak Supervision","code":"https://github.com/yulong-XJTU/SimVLM","n_code_links":2,"syntology":{"n_ran":18,"n_unverified":19,"n_samples":37,"n_pointer_only_licence":28}},{"rank_in_archive_order":7,"model":"BLIP_CapFilt-L","metrics":{"CIDEr":"111.8","Pre-train (#images)":"129M","SPICE":"14.9"},"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":8,"model":"LEMON_base","metrics":{"CIDEr":"107.7","Pre-train (#images)":"200M","SPICE":"14.7"},"uses_additional_data":false,"paper_date":"2021-11-24","paper":"/paper/scaling-up-vision-language-pre-training-for","paper_url":"https://arxiv.org/abs/2111.12233v2","paper_title":"Scaling Up Vision-Language Pre-training for Image Captioning","code":null,"n_code_links":0,"syntology":null},{"rank_in_archive_order":9,"model":"OmniVL","metrics":{"CIDEr":"104.6","Pre-train (#images)":"14M","SPICE":"15"},"uses_additional_data":false,"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":10,"model":"VinVL","metrics":{"CIDEr":"103.1","Pre-train (#images)":"5.7M","SPICE":" 14.2"},"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":11,"model":"Enc-Dec","metrics":{"CIDEr":"92.6","Pre-train (#images)":"15M","SPICE":" 12.5"},"uses_additional_data":false,"paper_date":"2021-02-17","paper":"/paper/conceptual-12m-pushing-web-scale-image-text","paper_url":"https://arxiv.org/abs/2102.08981v2","paper_title":"Conceptual 12M: Pushing Web-Scale Image-Text Pre-Training To Recognize Long-Tail Visual Concepts","code":"https://github.com/google-research-datasets/conceptual-12m","n_code_links":3,"syntology":null}],"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 6,795 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":6795,"papers_extracted_not_yet_verified":0,"boards_without_verdict":2,"papers_not_yet_extracted":2785},"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":5,"rows_with_any_sample_ran":5,"distinct_papers_with_graph_line":3,"distinct_papers_with_any_sample_ran":3,"samples_over_distinct_papers":{"n_ran":24,"n_unverified":23,"n_samples":47,"n_pointer_only_licence":30,"note":"each paper (arXiv id) counted once, however many rows it is behind; this is the page-level figure"},"samples_row_weighted":{"n_ran":32,"n_unverified":31,"n_samples":63,"n_pointer_only_licence":32,"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"}}}