WEwoCram/CommonForms

1---2license: apache-2.03task_categories:4- object-detection5tags:6- document-intelligence7- form-field-detection8- pdf9- multilingual10configs:11- config_name: default12  data_files:13  - split: train14    path: data/train-*15  - split: test16    path: data/test-*17  - split: valid18    path: data/valid-*19dataset_info:20  features:21  - name: id22    dtype: int6423  - name: image_id24    dtype: int6425  - name: image26    dtype: image27  - name: file_name28    dtype: string29  - name: width30    dtype: int6431  - name: height32    dtype: int6433  - name: objects34    sequence:35    - name: id36      dtype: int6437    - name: area38      dtype: float3239    - name: bbox40      sequence: float3241    - name: category42      dtype:43        class_label:44          names:45            '0': Text46            '1': CheckBox47            '2': Signature48    - name: category_id49      dtype: int6450    - name: iscrowd51      dtype: int6452    - name: segmentation53      sequence: int6454  splits:55  - name: train56    num_bytes: 160974075696.3957    num_examples: 43569858  - name: test59    num_bytes: 11950697193.87760    num_examples: 3306161  - name: valid62    num_bytes: 5942589583.7963    num_examples: 1819564  download_size: 15333640697265  dataset_size: 178867362474.0570466---67 68# CommonForms: A Large, Diverse Dataset for Form Field Detection69 70This repository hosts the **CommonForms** dataset, a web-scale dataset for form field detection, introduced in the paper [CommonForms: A Large, Diverse Dataset for Form Field Detection](https://huggingface.co/papers/2509.16506).71 72CommonForms 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.73 74**Key Features:**75*   **Scale:** Roughly 55,000 documents comprising over 450,000 pages.76*   **Source:** Constructed by filtering Common Crawl to find PDFs with fillable elements.77*   **Diversity:** Contains a diverse mixture of languages (one third non-English) and domains, with no single domain making up more than 25% of the dataset.78*   **Purpose:** The first large-scale dataset released for form field detection, aimed at fostering the development of robust form field detectors.79 80**Paper:** [CommonForms: A Large, Diverse Dataset for Form Field Detection](https://huggingface.co/papers/2509.16506)81**Code:** [https://github.com/jbarrow/commonforms](https://github.com/jbarrow/commonforms)