CommonForms: A Large, Diverse Dataset for Form Field Detection

Papers with Code Trending Papers
form-field-detection dataset object-detection document-understanding open-source

Summary

This paper introduces CommonForms, a large-scale, diverse dataset for form field detection, along with FFDNet models that achieve high precision on form field detection tasks. The dataset, models, and code are open-sourced.

This paper introduces CommonForms, a web-scale dataset for form field detection. It casts the problem of form field detection as object detection: given an image of a page, predict the location and type (Text Input, Choice Button, Signature) of form fields. The dataset is constructed by filtering Common Crawl to find PDFs that have fillable elements. Starting with 8 million documents, the filtering process is used to arrive at a final dataset of roughly 55k documents that have over 450k pages. Analysis shows that the dataset contains a diverse mixture of languages and domains; one third of the pages are non-English, and among the 14 classified domains, no domain makes up more than 25% of the dataset. In addition, this paper presents a family of form field detectors, FFDNet-Small and FFDNet-Large, which attain a very high average precision on the CommonForms test set. Each model cost less than $500 to train. Ablation results show that high-resolution inputs are crucial for high-quality form field detection, and that the cleaning process improves data efficiency over using all PDFs that have fillable fields in Common Crawl. A qualitative analysis shows that they outperform a popular, commercially available PDF reader that can prepare forms. Unlike the most popular commercially available solutions, FFDNet can predict checkboxes in addition to text and signature fields. This is, to our knowledge, the first large scale dataset released for form field detection, as well as the first open source models. The dataset, models, and code will be released at https://github.com/jbarrow/commonforms
Original Article
View Cached Full Text

Cached at: 07/12/26, 10:49 AM

Paper page - CommonForms: A Large, Diverse Dataset for Form Field Detection

Source: https://huggingface.co/papers/2509.16506 Published on Sep 20, 2025

Abstract

A web-scale dataset and models for form field detection are introduced, achieving high precision and supporting diverse languages and domains.

This paper introducesCommonForms, a web-scale dataset for form field detection. It casts the problem ofform field detectionasobject detection: given an image of a page, predict the location and type (Text Input, Choice Button, Signature) of form fields. The dataset is constructed by filtering Common Crawl to find PDFs that have fillable elements. Starting with 8 million documents, the filtering process is used to arrive at a final dataset of roughly 55k documents that have over 450k pages. Analysis shows that the dataset contains a diverse mixture of languages and domains; one third of the pages are non-English, and among the 14 classified domains, no domain makes up more than 25% of the dataset. In addition, this paper presents a family of form field detectors,FFDNet-SmallandFFDNet-Large, which attain a very high average precision on theCommonFormstest set. Each model cost less than $500 to train. Ablation results show thathigh-resolution inputsare crucial for high-quality form field detection, and that the cleaning process improvesdata efficiencyover using all PDFs that have fillable fields in Common Crawl. A qualitative analysis shows that they outperform a popular, commercially available PDF reader that can prepare forms. Unlike the most popular commercially available solutions, FFDNet can predictcheckboxesin addition to text and signature fields. This is, to our knowledge, the first large scale dataset released forform field detection, as well as the first open source models. The dataset, models, and code will be released at https://github.com/jbarrow/commonforms

View arXiv pageView PDFGitHub1.12kAdd to collection

Get this paper in your agent:

hf papers read 2509\.16506

Don’t have the latest CLI?curl \-LsSf https://hf\.co/cli/install\.sh \| bash

Models citing this paper5

#### jbarrow/FFDNet-L Object Detection• UpdatedOct 15, 2025 • 15 #### jbarrow/FFDNet-L-cpu Object Detection• UpdatedOct 15, 2025 • 4 #### jbarrow/FFDNet-S Object Detection• UpdatedOct 15, 2025 • 2 #### jbarrow/FFDNet-S-cpu Object Detection• UpdatedNov 26, 2025 Browse 5 models citing this paper## Datasets citing this paper5

#### Voxel51/commonforms_val_subset Viewer• UpdatedOct 23, 2025 • 1 • 2.57k • 1 #### jbarrow/CommonForms Viewer• UpdatedNov 16, 2025 • 487k • 1.72k • 57 #### WEwoCram/CommonForms Viewer• UpdatedApr 28 • 487k • 495 #### kurianmelvin/CommonForms Viewer• UpdatedJan 3 • 487k • 97 Browse 5 datasets citing this paper### Spaces citing this paper8

Collections including this paper2

Similar Articles

BaFCo: A Document Understanding Benchmark for Complex Bangla Form Comprehension

arXiv cs.CL

Introduces BaFCo, a benchmark dataset for Bangla form comprehension focusing on Document Layout Analysis (DLA) and Key Information Extraction (KIE). It includes 200 multi-page complex Bangladeshi government forms with fine-grained annotations across 26 entity types and evaluates multiple MLLMs, revealing limitations in understanding complex Bangla forms.

Robust, Scalable Detection of Text Containment in Large Web-Crawled Corpora

arXiv cs.CL

This paper presents FindMyText, an open-source Python package that efficiently detects whether a given text appears within a large web-crawled corpus, using a novel fingerprint chaining mechanism for near-verbatim copy detection. It demonstrates superior performance on benchmarks across ArXiv, Wikipedia, and generic web content.