[{"data":1,"prerenderedAt":-1},["ShallowReactive",2],{"doc-detail-84842-en":3,"doc-seo-84842-105":29,"detail-sidebar-cat-0-en-105":91},{"code":4,"msg":5,"data":6},0,"success",{"doc_id":7,"user_id":8,"nickname":9,"user_avatar":10,"doc_module":4,"category_id":11,"category_name":12,"doc_title":13,"doc_description":14,"doc_content":15,"file_id":16,"file_url":17,"file_type":18,"file_size":19,"view_count":20,"is_deleted":4,"is_public":20,"is_downloadable":20,"audit_status":20,"page_count":21,"language":22,"language_code":23,"site_id":24,"html_lang":23,"table_of_contents":25,"faqs":26,"seo_title":13,"seo_description":14,"update_tm":27,"read_time":28},84842,8796095462418,"Noah","https://ap-avatar.wpscdn.com/avatar/80000253c1241d02b47?x-image-process=image/resize,m_fixed,w_180,h_180&k=1778826106357471780",8,"Research & Report","BaFCo: A Document Understanding Benchmark for Complex Bangla Form Comprehension","Document comprehension is essential for real-world, human-centric applications, yet multimodal large language models struggle in low-resource settings like Bangla where high-quality annotated data is scarce. BaFCo introduces a benchmark for Bangla form comprehension, centered on Document Layout Analysis (DLA) and Key Information Extraction (KIE), built from 200 multi-page Bangladeshi government forms. The benchmark provides fine-grained entity annotations, labeled relationships, and an evaluation pipeline, revealing limitations in current MLLMs’ Bangla form localization accuracy.","arXiv :2607 .056 14v 1 [ cs .CL] 6 Jul 2026  \nBaFCo: A Document Understanding Benchmark for Complex Bangla Form Comprehension  \nAbu Tyeb Azad 1,⋆(B) , Ishita Sur Apan2,⋆, Fahim Ahmed2,⋆, Sumaiya Karim Katha2 , Ezharuddin Jubaer2 , Armun Alam2 , Pranjal Kumar Nandi3 , Amin Ahsan Ali2 , Aman Chadha4 ,‡, Md Mofijul Islam4 ,†,‡, and AKM Mahbubur  \nRahman2 ,†  \n1 Wichita State University, USA  \n2 Center for Computational & Data Sciences, Bangladesh  \n3 University of Dhaka, Bangladesh  \n4 Amazon GenAI, USA  \nAbstract. Document comprehension is a challenging yet impactful task for Multimodal Large Language Models, especially as these systems see growing adoption in real-world, human-centric applications. However, this adoption is limited for low-resource languages such as Bangla due to the scarcity of high-quality annotated data. To address this gap, we introduce BaFCo, a benchmark dataset for Bangla form comprehension with a focus on Document Layout Analysis (DLA) and Key Information Extraction (KIE) . BaFCo curates 200 multi-page complex Bangladeshi government forms, sourced from across diverse sectors including agriculture, education, banking, and land management. To accurately capture the structural and contextual complexity of these forms, we define a finegrained annotation schema comprising 26 types of form entities, along with a separate coarse form entity set consisting of 5 types. We evaluate the latest MLLMs from the ChatGPT, Gemini, Claude, Qwen, and Kimi series using zero-shot and chain-of-thought prompts under both low and high reasoning setups. Our results reveal limitations in current MLLMs’ability in comprehending Bangla forms, particularly in accurately localizing highly granular form entities. Our dataset and code is available at:  \n[https://huggingface.co/datasets/Mausul/bafco](https://huggingface.co/datasets/Mausul/bafco)  \nKeywords: Document Comprehension, Document Layout Analysis, Key Information Extraction, Low-resource NLP, Multimodal LLM  \n1 Introduction  \nInformation retrieval from documents such as forms underpins many real-world systems including banking, education, and public administration [28] . Document Layout Analysis (DLA) and Key Information Extraction (KIE)  \n⋆ Equal contribution. †Equal supervision. ‡Work done outside role at Amazon.  \n(B) Corresponding author: [mausulazad495@gmail.com](mausulazad495@gmail.com)  \nAccepted at the 19th European Conference on Computer Vision (ECCV), 2026 .  \n2 A. Azad et al.  \n[1, 8 , 25] are tasks at the core of Document Understanding. DLA identifies the structural elements of a page and the relationships between them. KIE involves locating and extracting values from form fields filled digitally or by hand. Both DLA and KIE are foundational precursors to downstream document tasks such as document question answering, entity linking, and summarization.  \nBangla remains a low-resource language in document understanding despite being the world’s 7th most spoken language, with 284 million speakers [24] . Government forms are also underrepresented in research, despite their semantic diversity and practical importance in public services. Benchmarks such as FUNSD [16] and XFUND [38] have advanced form understanding in English and other languages, but no comparable benchmark has been available for Bangla forms. Consequently, the lack of high-quality datasets and benchmarks continues to limit the development of Bangla document understanding systems.  \nTo address this gap, we introduce BaFCo, a curated benchmark for multipage Bangla form comprehension, focusing on government forms and the tasks of DLA and KIE. BaFCo provides fine-grained annotations spanning 26 entity types and labeled relationships between related fields. For DLA, the dataset contains 16,382 entities and 8,771 relationships across 200 forms (316 pages, with 1–5 pages per form) . For KIE, it further includes 1,926 key-value pairs spanning 156 forms (186 pages) . To maximize diversity, we prioritize complex and","cbCaid6eMJnawwOt","https://ap.wps.com/l/cbCaid6eMJnawwOt","pdf",24804927,1,30,"English","en",105,"# Introduction\n## BaFCo benchmark overview\n## DLA and KIE tasks and evaluation","[{\"question\":\"What problems does BaFCo aim to address for Bangla document understanding?\",\"answer\":\"BaFCo addresses the lack of high-quality annotated data and comparable benchmarks for Bangla form understanding, focusing on DLA and KIE tasks needed for downstream document applications.\"},{\"question\":\"How is the BaFCo dataset constructed and what does it contain?\",\"answer\":\"BaFCo curates 200 multi-page complex Bangladeshi government forms across diverse sectors, with fine-grained entity annotations and labeled relationships, plus key-value pairs for KIE.\"},{\"question\":\"How are multimodal LLMs evaluated on BaFCo and what key limitation is found?\",\"answer\":\"The work evaluates ChatGPT, Gemini, Claude, Qwen, and Kimi models with zero-shot and chain-of-thought prompts under low/high reasoning setups. Results show current MLLMs have difficulty comprehending Bangla forms, especially in accurately localizing highly granular entities.\"}]",1784198738,76,{"code":4,"msg":30,"data":31},"ok",{"site_id":24,"language":23,"slug":32,"title":13,"keywords":33,"description":14,"schema_data":34,"social_meta":86,"head_meta":88,"extra_data":90,"updated_unix":27},"bafco-a-document-understanding-benchmark-for-complex-bangla-form-comprehension","",{"@graph":35,"@context":85},[36,53,68],{"@type":37,"itemListElement":38},"BreadcrumbList",[39,43,47,50],{"item":40,"name":41,"@type":42,"position":20},"https://docshare.wps.com","Home","ListItem",{"item":44,"name":45,"@type":42,"position":46},"https://docshare.wps.com/document/","Document",2,{"item":48,"name":12,"@type":42,"position":49},"https://docshare.wps.com/document/research-report/",3,{"item":51,"name":13,"@type":42,"position":52},"https://docshare.wps.com/document/bafco-a-document-understanding-benchmark-for-complex-bangla-form-comprehension/84842/",4,{"url":51,"name":13,"@type":54,"author":55,"headline":13,"publisher":57,"fileFormat":60,"inLanguage":23,"description":14,"dateModified":61,"datePublished":62,"encodingFormat":60,"isAccessibleForFree":63,"interactionStatistic":64},"DigitalDocument",{"name":9,"@type":56},"Person",{"url":40,"name":58,"@type":59},"DocShare","Organization","application/pdf","2026-07-17","2026-07-16",true,{"@type":65,"interactionType":66,"userInteractionCount":20},"InteractionCounter",{"@type":67},"ViewAction",{"@type":69,"mainEntity":70},"FAQPage",[71,77,81],{"name":72,"@type":73,"acceptedAnswer":74},"What problems does BaFCo aim to address for Bangla document understanding?","Question",{"text":75,"@type":76},"BaFCo addresses the lack of high-quality annotated data and comparable benchmarks for Bangla form understanding, focusing on DLA and KIE tasks needed for downstream document applications.","Answer",{"name":78,"@type":73,"acceptedAnswer":79},"How is the BaFCo dataset constructed and what does it contain?",{"text":80,"@type":76},"BaFCo curates 200 multi-page complex Bangladeshi government forms across diverse sectors, with fine-grained entity annotations and labeled relationships, plus key-value pairs for KIE.",{"name":82,"@type":73,"acceptedAnswer":83},"How are multimodal LLMs evaluated on BaFCo and what key limitation is found?",{"text":84,"@type":76},"The work evaluates ChatGPT, Gemini, Claude, Qwen, and Kimi models with zero-shot and chain-of-thought prompts under low/high reasoning setups. Results show current MLLMs have difficulty comprehending Bangla forms, especially in accurately localizing highly granular entities.","https://schema.org",{"og:url":51,"og:type":87,"og:title":13,"og:site_name":58,"og:description":14},"article",{"robots":89,"canonical":51},"index,follow",{"doc_id":7,"site_id":24},{"code":4,"msg":5,"data":92},[93,97,101,105,110,115,120,122,127,130,134],{"id":20,"doc_module":4,"doc_module_name":45,"category_name":94,"show_sort_weight":95,"slug":96},"Story & Novel",90,"story-novel",{"id":46,"doc_module":4,"doc_module_name":45,"category_name":98,"show_sort_weight":99,"slug":100},"Literature",80,"literature",{"id":52,"doc_module":4,"doc_module_name":45,"category_name":102,"show_sort_weight":103,"slug":104},"Exam",70,"exam",{"id":106,"doc_module":4,"doc_module_name":45,"category_name":107,"show_sort_weight":108,"slug":109},5,"Comic",60,"comic",{"id":111,"doc_module":4,"doc_module_name":45,"category_name":112,"show_sort_weight":113,"slug":114},6,"Technology",50,"technology",{"id":116,"doc_module":4,"doc_module_name":45,"category_name":117,"show_sort_weight":118,"slug":119},7,"Healthcare",40,"healthcare",{"id":11,"doc_module":4,"doc_module_name":45,"category_name":12,"show_sort_weight":21,"slug":121},"research-report",{"id":123,"doc_module":4,"doc_module_name":45,"category_name":124,"show_sort_weight":125,"slug":126},9,"Religion & Spirituality",20,"religion-spirituality",{"id":125,"doc_module":4,"doc_module_name":45,"category_name":128,"show_sort_weight":125,"slug":129},"World Cup","world-cup",{"id":131,"doc_module":4,"doc_module_name":45,"category_name":132,"show_sort_weight":131,"slug":133},10,"Lifestyle","lifestyle",{"id":135,"doc_module":4,"doc_module_name":45,"category_name":136,"show_sort_weight":106,"slug":137},19,"General","general"]