[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-155408-105":59,"doc-detail-155408-en":123},{"code":4,"msg":5,"data":6},0,"success",[7,13,18,23,28,33,38,43,48,51,55],{"id":8,"doc_module":4,"doc_module_name":9,"category_name":10,"show_sort_weight":11,"slug":12},1,"Document","Story & Novel",90,"story-novel",{"id":14,"doc_module":4,"doc_module_name":9,"category_name":15,"show_sort_weight":16,"slug":17},2,"Literature",80,"literature",{"id":19,"doc_module":4,"doc_module_name":9,"category_name":20,"show_sort_weight":21,"slug":22},4,"Exam",70,"exam",{"id":24,"doc_module":4,"doc_module_name":9,"category_name":25,"show_sort_weight":26,"slug":27},5,"Comic",60,"comic",{"id":29,"doc_module":4,"doc_module_name":9,"category_name":30,"show_sort_weight":31,"slug":32},6,"Technology",50,"technology",{"id":34,"doc_module":4,"doc_module_name":9,"category_name":35,"show_sort_weight":36,"slug":37},7,"Healthcare",40,"healthcare",{"id":39,"doc_module":4,"doc_module_name":9,"category_name":40,"show_sort_weight":41,"slug":42},8,"Research & Report",30,"research-report",{"id":44,"doc_module":4,"doc_module_name":9,"category_name":45,"show_sort_weight":46,"slug":47},9,"Religion & Spirituality",20,"religion-spirituality",{"id":46,"doc_module":4,"doc_module_name":9,"category_name":49,"show_sort_weight":46,"slug":50},"World Cup","world-cup",{"id":52,"doc_module":4,"doc_module_name":9,"category_name":53,"show_sort_weight":52,"slug":54},10,"Lifestyle","lifestyle",{"id":56,"doc_module":4,"doc_module_name":9,"category_name":57,"show_sort_weight":24,"slug":58},19,"General","general",{"code":4,"msg":60,"data":61},"ok",{"site_id":62,"language":63,"slug":64,"title":65,"keywords":66,"description":67,"schema_data":68,"social_meta":116,"head_meta":118,"extra_data":120,"updated_unix":122},105,"en","m2c-towards-automatic-multimodal-manga-complement-research-paper","M2C - Towards Automatic Multimodal Manga Complement - research paper","","Multimodal manga analysis enhances manga understanding by jointly leveraging visual and textual cues, addressing practical issues in hand-drawn comics such as missing pages, text contamination, and aging that lead to incomplete comic text. The Multimodal Manga Complement (M2C) task is introduced to recover missing dialogue content via a shared semantic space for vision and language. The paper proposes M2C and a new bilingual benchmark dataset, an MCoT argumentation method powered by large language models, and the FVP-M2 baseline with fine-grained visual prompts. Experiments validate the complement effectiveness across missing-text scenarios.",{"@graph":69,"@context":115},[70,84,106],{"@type":71,"itemListElement":72},"BreadcrumbList",[73,77,79,82],{"item":74,"name":75,"@type":76,"position":8},"https://docshare.wps.com","Home","ListItem",{"item":78,"name":9,"@type":76,"position":14},"https://docshare.wps.com/document/",{"item":80,"name":40,"@type":76,"position":81},"https://docshare.wps.com/document/research-report/",3,{"item":83,"name":65,"@type":76,"position":19},"https://docshare.wps.com/document/m2c-towards-automatic-multimodal-manga-complement-research-paper/155408/",{"url":83,"name":65,"@type":85,"image":86,"author":91,"headline":65,"publisher":94,"fileFormat":97,"inLanguage":63,"description":67,"dateModified":98,"datePublished":99,"encodingFormat":97,"isAccessibleForFree":100,"interactionStatistic":101},"DigitalDocument",{"url":87,"@type":88,"width":89,"height":90},"https://docshare.wps.com/thumbnails/m2c-towards-automatic-multimodal-manga-complement-research-paper/155408.png","ImageObject",300,407,{"name":92,"@type":93},"Asher","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-23","2026-08-28",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",11,{"@type":107,"mainEntity":108},"FAQPage",[109],{"name":110,"@type":111,"acceptedAnswer":112},"What is the role of the FVP-M2 baseline?","Question",{"text":113,"@type":114},"FVP-M2 uses fine-grained visual prompts: it encodes visual tokens, generates prompts by aggregating global and local visual information, and then performs dialogue complement with vision-guided language token generation to predict the missing content.","Answer","https://schema.org",{"og:url":83,"og:type":117,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":119,"canonical":83},"index,follow",{"doc_id":121,"site_id":62},155408,1787911544,{"code":4,"msg":5,"data":124},{"doc_id":121,"user_id":125,"nickname":92,"user_avatar":126,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":127,"file_id":128,"file_url":129,"file_type":130,"file_size":131,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":34,"language":132,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":133,"faqs":134,"seo_title":135,"seo_description":67,"update_tm":122,"read_time":136,"preview_page_count":4},687197207639,"https://ap-avatar.wpscdn.com/davatar_a8503ba1806abce46bf441b54a3ca4cd","M2C: Towards Automatic Multimodal Manga Complement  \nHongcheng Guo1 , Boyang Wang* 1 , Jiaqi Bai 1 , Jiaheng Liu 1 , Jian Yang 1 , Zhoujun LiB 1  \n1Beihang University  \n{hongchengguo,wangboyang,bjq,liujiaheng,jiaya, [lizj}@buaa.edu.cn](lizj}@buaa.edu.cn)  \nAbstract  \nMultimodal manga analysis focuses on enhancing manga understanding with visual and textual features, which has attracted considerable attention from both natural language processing and computer vision communities. Currently, most comics are hand-drawn and prone to problems such as missing pages, text contamination, and aging, resulting in missing comic text content and seriously hindering human comprehension. In other words, the Multimodal Manga Complement (M2C) task has not been investigated, which aims to handle the aforementioned issues by providing a shared semantic space for vision and language understanding. To this end, we first propose the Multimodal Manga Complement task by establishing a new M2C benchmark dataset covering two languages. First, we design a manga argumentation method called MCoT to mine event knowledge in comics with large language models. Then, an effective baseline FVP-M2 using fine-grained visual prompts is proposed to sup  \nport manga complement. Extensive experimental results show the effectiveness of FVP-M2 method for Multimodal Mange Complement .  \n1 Introduction  \nComics are enjoyed globally, often featuring a combination of text and illustrations. However, due to their hand-drawn nature, they are susceptible to damage during circulation, such as missing pages, text contamination, and aging. To address these issues, we propose a task called Multimodal Manga Complement (M2C), illustrated in Fig. 1. This task extends conventional text-based content complement by incorporating corresponding images as additional inputs to mitigate data sparsity and ambiguity (Ive et al., 2019a) . Similar to other multimodal manga tasks (e.g., Multimodal Manga  \n First two authors contributed equally. B Corresponding author. [https://github.com/HC-Guo/M2C](https://github.com/HC-Guo/M2C)  \nDialogue3: Huh?  \nDialogue4: Could it be because everyone hates you?  \nFigure 1: Motivation of our M2C task. The left part describes both visual and language modalities as the input (vision part depicts a sequence of images from a page of comics, and the language part is the corresponding text dialogues) . The right part displays a page of comics and the text that we want to complete.  \nTranslation (Hinami et al., 2021) and Comic Image Inpainting (Ono et al., 2021)), M2C aims to exploit the effectiveness of visual features for manga complement. Furthermore, M2C can serve as a foundation for other comic-related tasks.  \nHowever, as we can see in Fig. 1, compared to image caption (Vinyals et al., 2015) or other multimodal tasks, the comic data we use has its own characteristics. Here we conclude the differences and corresponding challenges: I) . Comics have shorter conversational text lengths than other natural language forms, and contain many inflections or auxiliaries. This leads to a greater randomness in content generation that requires complement. II) . The text and images of comics are complementary, often in pairs. This requires us to design better ways to exploit image features. III) . Comic images and text contain timelines and implicit logic information, which shows us a challenge to leverage the logical knowledge to help improve the stability of generating complementary content. Therefore, in our work, we first propose the Multimodal Manga Complement (M2C) task to achieve the complement for with both vision and language information.  \n9876  \nFindings of the Association for Computational Linguistics: EMNLP 2023 , pages 9876–9882 December 6-10, 2023 ©2023 Association for Computational Linguistics  \nTo eliminate the above limitations, inspired by recent CoTs (Wei et al., 2022 ; Wang et al., 2022), we first design the three-hop Manga Chain-of-Thought (MCoT) argum","cbCailJS8rv9IxTL","https://ap.wps.com/l/cbCailJS8rv9IxTL","pdf",1319504,"English","# Abstract\n# Introduction\n# Contributions\n# Related Works","[{\"question\":\"What is the role of the FVP-M2 baseline?\",\"answer\":\"FVP-M2 uses fine-grained visual prompts: it encodes visual tokens, generates prompts by aggregating global and local visual information, and then performs dialogue complement with vision-guided language token generation to predict the missing content.\"}]","M2C - Towards Automatic Multimodal Manga Complement - research paper | PDF",18]