{"about":{"site":"https://codewithpapers.app","non_affiliation":"Code with Papers and Syntology are not affiliated with, endorsed by, or sponsored by Papers with Code, Meta, or the pwc-archive mirror.","licence":"CC BY-SA 4.0","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","attribution":"https://codewithpapers.app/attribution","modified":"archive material modified by Syntology; see the attribution page"},"url":"/paper/what-you-see-is-what-you-read-improving-text-1","title":"What You See is What You Read? Improving Text-Image Alignment Evaluation","arxiv_id":"2305.10400","date":"2023-05-17","proceeding":"NeurIPS 2023 11","authors":["Michal Yarom","Yonatan Bitton","Soravit Changpinyo","Roee Aharoni","Jonathan Herzig","Oran Lang","Eran Ofek","Idan Szpektor"],"abstract":"Automatically determining whether a text and a corresponding image are semantically aligned is a significant challenge for vision-language models, with applications in generative text-to-image and image-to-text tasks. In this work, we study methods for automatic text-image alignment evaluation. We first introduce SeeTRUE: a comprehensive evaluation set, spanning multiple datasets from both text-to-image and image-to-text generation tasks, with human judgements for whether a given text-image pair is semantically aligned. We then describe two automatic methods to determine alignment: the first involving a pipeline based on question generation and visual question answering models, and the second employing an end-to-end classification approach by finetuning multimodal pretrained models. Both methods surpass prior approaches in various text-image alignment tasks, with significant improvements in challenging cases that involve complex composition or unnatural images. Finally, we demonstrate how our approaches can localize specific misalignments between an image and a given text, and how they can be used to automatically re-rank candidates in text-to-image generation.","url_abs":"https://arxiv.org/abs/2305.10400v4","url_pdf":"https://arxiv.org/pdf/2305.10400v4.pdf","source":{"archive":"pwc-archive (Hugging Face), CC BY-SA 4.0","snapshot":"2025-07-28","licence_url":"https://creativecommons.org/licenses/by-sa/4.0/legalcode","row_kind":"abstracts"},"code_links":[{"paper_slug":"what-you-see-is-what-you-read-improving-text-1","repo_url":"https://github.com/yonatanbitton/wysiwyr","is_official":1,"mentioned_in_paper":0,"mentioned_in_github":0,"framework":"pytorch","reach":{"status":"ok","spdx":"Apache-2.0"}}],"tasks":[{"task_slug":"image-generation","task_name":"Image Generation"},{"task_slug":"image-to-text","task_name":"Image to text"},{"task_slug":"question-answering","task_name":"Question Answering"},{"task_slug":"question-generation","task_name":"Question Generation"},{"task_slug":"question-generation","task_name":"Question-Generation"},{"task_slug":"text-generation","task_name":"Text Generation"},{"task_slug":"text-to-image-generation-1","task_name":"Text to Image Generation"},{"task_slug":"text-to-image-generation","task_name":"Text-to-Image Generation"},{"task_slug":"visual-question-answering-1","task_name":"Visual Question Answering"},{"task_slug":"visual-reasoning","task_name":"Visual Reasoning"}],"methods":[],"datasets_introduced":[],"methods_introduced":[],"results":[{"leaderboard":"/sota/visual-reasoning-on-winoground","task":"Visual Reasoning","dataset":"Winoground","model":"VQ2","rank_in_archive_order":12,"of":114,"metrics":{"Group Score":"30.5","Image Score":"42.2","Text Score":"47"},"uses_additional_data":false},{"leaderboard":"/sota/visual-reasoning-on-winoground","task":"Visual Reasoning","dataset":"Winoground","model":"PaLI (ft SNLI-VE + Synthetic Data)","rank_in_archive_order":15,"of":114,"metrics":{"Group Score":"28.75","Image Score":"38","Text Score":"46.5"},"uses_additional_data":false},{"leaderboard":"/sota/visual-reasoning-on-winoground","task":"Visual Reasoning","dataset":"Winoground","model":"PaLI (ft SNLI-VE)","rank_in_archive_order":18,"of":114,"metrics":{"Group Score":"28.70","Image Score":"41.50","Text Score":"45.00"},"uses_additional_data":false},{"leaderboard":"/sota/visual-reasoning-on-winoground","task":"Visual Reasoning","dataset":"Winoground","model":"BLIP2 (ft COCO)","rank_in_archive_order":22,"of":114,"metrics":{"Group Score":"23.50","Image Score":"26.00","Text Score":"44.00"},"uses_additional_data":false},{"leaderboard":"/sota/visual-reasoning-on-winoground","task":"Visual Reasoning","dataset":"Winoground","model":"COCA ViT-L14 (f.t on COCO)","rank_in_archive_order":78,"of":114,"metrics":{"Group Score":"8.25","Image Score":"11.50","Text Score":"28.25"},"uses_additional_data":false},{"leaderboard":"/sota/visual-reasoning-on-winoground","task":"Visual Reasoning","dataset":"Winoground","model":"OFA large (ft SNLI-VE)","rank_in_archive_order":81,"of":114,"metrics":{"Group Score":"9.00","Image Score":"14.30","Text Score":"27.70"},"uses_additional_data":false},{"leaderboard":"/sota/visual-reasoning-on-winoground","task":"Visual Reasoning","dataset":"Winoground","model":"CLIP RN50x64","rank_in_archive_order":83,"of":114,"metrics":{"Group Score":"10.25","Image Score":"13.75","Text Score":"26.50"},"uses_additional_data":false},{"leaderboard":"/sota/visual-reasoning-on-winoground","task":"Visual Reasoning","dataset":"Winoground","model":"TIFA","rank_in_archive_order":103,"of":114,"metrics":{"Group Score":"11.30","Image Score":"12.50","Text Score":"19.00"},"uses_additional_data":false}],"syntology":{"syntology_url":null,"atlas_url":"https://app.syntology.ai/?focus=2305.10400","mcp":null,"developers":"https://syntology.ai/developers"},"arxiv_metadata":null,"syntology_extracted_results":null}