datasets
Training and evaluation data, with the modality, task and licence stated up front. Listed live from the Hugging Face Hub.
FormStruct-Bench
FormStruct-Bench
Dataset Description
FormStruct-Bench is a multilingual benchmark for extracting the semantic and
spatial structure of forms from document images. The repository combines a
7,000-page main benchmark, a controlled visual-degradation set, and
template-level layout annotations. It supports evaluation of vision-language
models and document AI systems on hierarchical key-value extraction, document
structure recovery, region localization, table and… See the full description on the dataset page: https://huggingface.co/datasets/D2I-CUHK-Shenzhen/FormStruct-Bench.ANC-formsindian_dance_formsThis dataset is taken from https://www.kaggle.com/datasets/aditya48/indian-dance-form-classification but is originally from the Hackerearth deep learning contest of identifying Indian dance forms. All the credits of dataset goes to them.
Content
The dataset consists of 599 images belonging to 8 categories, namely manipuri, bharatanatyam, odissi, kathakali, kathak, sattriya, kuchipudi, and mohiniyattam. The original dataset was quite unstructured and all the images were put together.… See the full description on the dataset page: https://huggingface.co/datasets/tanmaykm/indian_dance_forms.synthetic-us-forms-preview
SymageDocs — Synthetic US Forms Preview
A small, CC-BY-4.0, fully synthetic document-AI training set: 525
labeled page images across six families of US business and government forms,
each page shipping FUNSD ground truth plus a LayoutLM-ready token/bbox/tag view.
This is a preview subset. It exists so you can load real output from the
SymageDocs generator, inspect the label quality, and
decide whether generating your own corpus is worth your time — without an
account, an email… See the full description on the dataset page: https://huggingface.co/datasets/Symage/synthetic-us-forms-preview.handwriting_forms
Dataset Card for Dataset Name
Dataset Summary
This dataset card aims to be a base template for new datasets. It has been generated using this raw template.
Supported Tasks and Leaderboards
[More Information Needed]
Languages
[More Information Needed]
Dataset Structure
Data Instances
[More Information Needed]
Data Fields
[More Information Needed]
Data Splits
[More Information Needed]
Dataset Creation… See the full description on the dataset page: https://huggingface.co/datasets/ift/handwriting_forms.US_tax_forms_donut
NIST-SD2 (US Tax Forms) Donut Dataset
Processed version of NIST Special Database 2 for document understanding tasks, formatted for use with the Donut architecture. Contains 5,590 annotated document images (5,031 train, 559 test) across 20 tax form classes.
Description
Curated by: National Institute of Standards and Technology (NIST)
License: MIT
Total Size: 947.33 MB
Annotations:
Class labels (20 tax form types)
Full text ground truth
UFLA-FORMS
UFLA-FORMS: an Academic Forms Dataset for Information Extraction in the Portuguese Language
About
UFLA-FORMS is a manually labeled dataset of document forms in Brazilian Portuguese extracted from the domains of the Federal University of Lavras (UFLA). The dataset emphasizes the hierarchical structure between the entities of a document through their relationships, in addition to the extraction of key-value pairs. Samples were labeled using ToolRI.
Overview… See the full description on the dataset page: https://huggingface.co/datasets/Victorgonl/UFLA-FORMS.handwriting_forms_cleaned
handwriting_forms_cleaned
The handwriting_forms__x family of the ElliotVL supervised-fine-tuning pool, after VLM cleaning.
images
1,360
QA turns
5,198
answers rewritten by the cleaning pass
166
QA created by the cleaning pass (new_qa)
3,860 (74.3%)
shards
1
How this was cleaned
A vision-language model read each image together with its QA and judged the item. The pass is
not a filter that only removes rows — it rewrites answers it finds… See the full description on the dataset page: https://huggingface.co/datasets/Elliot-Data/handwriting_forms_cleaned.OMR-forms
Dataset Card for "OMR-forms"
More Information needed
medical-forms-datasethandwriting_forms_cleaned
handwriting_forms_cleaned
The handwriting_forms__x family of the ElliotVL supervised-fine-tuning pool, after VLM cleaning.
images
1,360
QA turns
5,198
answers rewritten by the cleaning pass
166
QA created by the cleaning pass (new_qa)
3,860 (74.3%)
shards
1
How this was cleaned
A vision-language model read each image together with its QA and judged the item. The pass is
not a filter that only removes rows — it rewrites answers it finds… See the full description on the dataset page: https://huggingface.co/datasets/elliot-mllm/handwriting_forms_cleaned.coherent-forms-1040-cms1500-i9
SymageDocs — Coherent US Tax / Health / Employment Forms (FUNSD)
A fully synthetic document-AI training set: three US forms — IRS Form 1040,
CMS-1500, and USCIS Form I-9 — filled from the same synthetic
identity, so name / SSN / address / employer flow consistently across all
three renderings. Each page ships with FUNSD ground truth (word boxes,
entity labels, key–value linking) plus a LayoutLMv3-ready token/bbox/tag view.
3,000 page-level image + annotation rows (train 2,400 /… See the full description on the dataset page: https://huggingface.co/datasets/Symage/coherent-forms-1040-cms1500-i9.indian_dance_formsThis dataset is taken from https://www.kaggle.com/datasets/aditya48/indian-dance-form-classification but is originally from the Hackerearth deep learning contest of identifying Indian dance forms. All the credits of dataset goes to them.
Content
The dataset consists of 599 images belonging to 8 categories, namely manipuri, bharatanatyam, odissi, kathakali, kathak, sattriya, kuchipudi, and mohiniyattam. The original dataset was quite unstructured and all the images were put together.… See the full description on the dataset page: https://huggingface.co/datasets/SiddharthStarr/indian_dance_forms.forms-from-rvl-cdip