[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"detail-sidebar-cat-0-en-105":3,"doc-seo-136928-105":59,"doc-detail-136928-en":131},{"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":124,"head_meta":126,"extra_data":128,"updated_unix":130},105,"en","comix-a-comprehensive-benchmark-for-multi-task-comic-understanding","CoMix - A Comprehensive Benchmark for Multi-Task Comic Understanding","","CoMix presents a comprehensive, multi-task benchmark for comic analysis, addressing the gap where existing evaluation datasets and metrics remain focused on small-scale or single-style tasks. The benchmark extends three prior datasets with richer annotations and adds a carefully selected American comic-style dataset to reduce manga-style dominance. It evaluates zero-shot and limited fine-tuning transfer across comic styles while covering detection, speaker identification, character re-identification, reading order, and multimodal reasoning such as character naming and dialogue generation.",{"@graph":69,"@context":123},[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/comix-a-comprehensive-benchmark-for-multi-task-comic-understanding/136928/",{"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/comix-a-comprehensive-benchmark-for-multi-task-comic-understanding/136928.png","ImageObject",300,407,{"name":92,"@type":93},"Skyler","Person",{"url":74,"name":95,"@type":96},"DocShare","Organization","application/pdf","2026-10-17","2026-08-22",true,{"@type":102,"interactionType":103,"userInteractionCount":105},"InteractionCounter",{"@type":104},"ViewAction",11,{"@type":107,"mainEntity":108},"FAQPage",[109,115,119],{"name":110,"@type":111,"acceptedAnswer":112},"What is the main goal of the CoMix benchmark?","Question",{"text":113,"@type":114},"CoMix is designed to evaluate multi-task capabilities for comic analysis, covering a broad set of vision and multimodal tasks beyond isolated benchmarks.","Answer",{"name":116,"@type":111,"acceptedAnswer":117},"How does CoMix improve style diversity compared with manga-centric datasets?",{"text":118,"@type":114},"CoMix incorporates a new American comic-style dataset to mitigate the over-representation of manga-style data and to support cross-style transfer evaluation.",{"name":120,"@type":111,"acceptedAnswer":121},"Which tasks does CoMix include for character-related understanding and multimodal reasoning?",{"text":122,"@type":114},"CoMix covers speaker identification, character re-identification, reading order, and multimodal reasoning tasks including character naming and dialogue generation.","https://schema.org",{"og:url":83,"og:type":125,"og:title":65,"og:site_name":95,"og:description":67},"article",{"robots":127,"canonical":83},"index,follow",{"doc_id":129,"site_id":62},136928,1787395498,{"code":4,"msg":5,"data":132},{"doc_id":129,"user_id":133,"nickname":92,"user_avatar":134,"doc_module":4,"category_id":39,"category_name":40,"doc_title":65,"doc_description":67,"doc_content":135,"file_id":136,"file_url":137,"file_type":138,"file_size":139,"view_count":105,"is_deleted":4,"is_public":8,"is_downloadable":8,"audit_status":8,"page_count":140,"language":141,"language_code":63,"site_id":62,"html_lang":63,"table_of_contents":142,"faqs":143,"seo_title":144,"seo_description":67,"update_tm":130,"read_time":145},2336464648746,"https://ap-avatar.wpscdn.com/davatar_276721f389ce27ea32af1340a28f341c","arXiv :2407 .03550v2 [ cs .CV] 31 Oct 2024  \nCoMix: A Comprehensive Benchmark for Multi-Task  \nComic Understanding  \nEmanuele Vivoli 1 ,2 Marco Bertini2 Dimosthenis Karatzas 1  \n1 Computer Vision Center, UAB, Spain 2MICC, University of Florence, Italy  \n{evivoli,[dimos}@cvc.uab.cat](dimos}@cvc.uab.cat)[ ](dimos}@cvc.uab.cat){emanuele.vivoli, [marco.bertini}@unifi.it](marco.bertini}@unifi.it)  \nFigure 1: Composition of the CoMix benchmark. The top part of the figure provides a qualitative representation of the datasets included in CoMix. The accompanying bar charts depict the differences between the original annotations and those extended in CoMix. The left chart shows the increased number of annotations per dataset, whereas the right chart details the increase per task.  \nAbstract  \nThe comic domain is rapidly advancing with the development of single-page analysis and synthesis models. However, evaluation metrics and datasets lag behind, often limited to small-scale or single-style test sets. We introduce a novel benchmark, CoMix, designed to evaluate the multi-task capabilities of models in comic analysis. Unlike existing benchmarks that focus on isolated tasks such as object detection or text recognition, CoMix addresses a broader range of tasks including object detection, speaker identification, character re-identification, reading order, and multi-modal reasoning tasks like character naming and dialogue generation.  \nOur benchmark comprises three existing datasets with expanded annotations to support multi-task evaluation. To mitigate the over-representation of manga-style data, we have incorporated a new dataset of carefully selected American comicstyle books, thereby enriching the diversity of comic styles. CoMix is designed to assess pre-trained models in zero-shot and limited fine-tuning settings, probing their transfer capabilities across different comic styles and tasks. The validation split of the benchmark is publicly available for research purposes, and an evaluation server for the held-out test split is also provided. Comparative results between human performance and state-of-the-art models reveal a significant performance gap, highlighting substantial opportunities for advancements in comic understanding. The dataset, baseline models, and code are accessible at the repository link.  \nThis initiative sets a new standard for comprehensive comic analysis, providing the community with a common benchmark for evaluation on a large and varied set.  \n38th Conference on Neural Information Processing Systems (NeurIPS 2024) Track on Datasets and Benchmarks.  \n1 Introduction  \nComics, a distinct medium that seamlessly blends textual and visual components, has emerged as a globally celebrated form of cultural expression. While the accessibility of comics allows even young readers to comprehend and appreciate them with ease, the intricacy of comic page layouts poses significant challenges for computational understanding. The classical elements of comics, including panels, speech balloons, characters, text, and onomatopoeia, are heavily shaped by the creator’s imaginative vision and artistic flair, rendering the undertaking of comic image analysis a complex and multifaceted endeavor.  \nIn recent works[34], authors approached comics research to more complex tasks moving from classification [33], detection[36] and captioning[35] to diarization[19, 28], where low-level tasks-such as detecting objects, defining reading order and speaker identification and character re-identificationare used as a medium to generate an ordered transcription of who said what on single page. Despite the innovative application of this complex bottom-up approach to comics understanding tasks, thereis no benchmark on character naming or diarization, nor metrics to assess their correctness [34] . A number of available datasets support some of these tasks. For example, eBDtheque [10] and DCM772 [23] provide detection annotations for panels, characters, tex","cbCail2Bn0MU31bq","https://ap.wps.com/l/cbCail2Bn0MU31bq","pdf",22826160,23,"English","# Abstract\n## Introduction\n## CoMix Benchmark Overview\n## Datasets and Annotation Statistics\n## Evaluation Protocol and Baselines","[{\"question\":\"What is the main goal of the CoMix benchmark?\",\"answer\":\"CoMix is designed to evaluate multi-task capabilities for comic analysis, covering a broad set of vision and multimodal tasks beyond isolated benchmarks.\"},{\"question\":\"How does CoMix improve style diversity compared with manga-centric datasets?\",\"answer\":\"CoMix incorporates a new American comic-style dataset to mitigate the over-representation of manga-style data and to support cross-style transfer evaluation.\"},{\"question\":\"Which tasks does CoMix include for character-related understanding and multimodal reasoning?\",\"answer\":\"CoMix covers speaker identification, character re-identification, reading order, and multimodal reasoning tasks including character naming and dialogue generation.\"}]","CoMix - A Comprehensive Benchmark for Multi-Task Comic Understanding | PDF",58]