bingbangboom/war-gov-ufo-release-01-images

U.S. Department of War UFO/UAP Release 01 — Page Image Dataset

This dataset contains every page of the 120 PDF documents released by the U.S. Department of War on May 8, 2026 as part of Release 01 of the Presidential Unsealing and Reporting System for UAP Encounters (PURSUE). Each PDF page has been rendered as a JPEG image at 200 DPI and paired with metadata extracted from the official release manifest.

📋 Dataset Summary

Property	Value
Source	war.gov/UFO — U.S. Department of War PURSUE portal
Release Date	May 8, 2026
Total PDFs	120
Total Pages	4,134
Image Format	JPEG, 200 DPI
Split	train (complete corpus)

🗂️ Dataset Structure

Each row represents one page image from a single PDF document.

Column	Type	Description
image	Image	JPEG image of the PDF page
file_name	string	Original PDF filename
page_number	int32	Page number within the PDF (1-indexed)
title	string	Document title from the release manifest
agency	string	Originating agency (e.g., FBI, NASA, Department of War, Department of State)
release_date	string	Official release date (5/8/26)
incident_date	string	Date of the reported incident, if available
incident_location	string	Geographic location of the incident, if available

📊 Data Splits

Split	Rows	Description
train	4,134	Complete corpus of all page images

🏛️ Source Data

The source documents were downloaded from the official Department of War UFO/UAP release portal:

• Portal: https://www.war.gov/UFO/
• Program: Presidential Unsealing and Reporting System for UAP Encounters (PURSUE)
• Release: Release 01 — "Cleared for release - May 8, 2026"
• Agencies represented: FBI, NASA, Department of War, Department of State

Document Types

The 120 PDFs include:

• FBI investigative case files (62-HQ-83894 series)
• Department of War historical records (1946–1955)
• NASA mission transcripts and crew debriefings (Apollo, Gemini, Skylab)
• Department of State diplomatic cables
• Contemporary UAP mission reports (2020–2025)
• Range fouler debriefs and reporting forms
• Photographic evidence (FBI Photo B1–B24)

🔧 Processing Steps

1. Download: All 120 PDFs were downloaded from war.gov/medialink/ufo/release_1/ using the official manifest.
2. Verification: Each file was verified to start with the %PDF- magic bytes and have non-zero size.
3. Rendering: Each PDF page was rendered to an image using PyMuPDF at 200 DPI.
4. Encoding: Images were encoded as JPEG with quality determined by the renderer's default settings.
5. Metadata pairing: Each page was matched with metadata from the official release manifest (pdf_manifest.tsv).
6. Dataset creation: Images and metadata were assembled into a Hugging Face datasets Dataset and exported as Parquet.

⚖️ License

CC0 1.0 Universal (Public Domain Dedication)

These documents are works of the United States Government released under the Department of War's PURSUE program. Under 17 U.S.C. § 105, copyright protection is not available for works of the U.S. Government. The National Archives confirms that "copies of Federal records, as part of the public domain, are equally available to all."

This dataset is released as CC0-1.0 to explicitly dedicate it to the public domain worldwide, following the precedent established by other U.S. government document datasets on the Hugging Face Hub (e.g., EPA data releases).

Disclaimer: This dataset is a faithful reproduction of officially released government documents. It is provided as-is for research, journalism, and historical preservation purposes. The dataset curators make no claims regarding the accuracy, authenticity, or classification status of the content within the original documents.

🙏 Acknowledgements

• U.S. Department of War — for the release of these documents via the PURSUE portal.
• Denis Shiryaev (DenisSergeevitch) — for maintaining the UFO-USA GitHub archive, which provided the corrected PDF manifest used for metadata mapping and download verification.
• PyMuPDF — for the PDF rendering engine.
• Hugging Face Datasets — for the dataset library and Parquet export tools.

📖 Citation

If you use this dataset in your research, please cite both the original government release and this dataset:

@dataset{war_gov_ufo_release_01_images,
  title = {U.S. Department of War UFO/UAP Release 01 — Page Image Dataset},
  author = {{U.S. Department of War}},
  year = {2026},
  month = {May},
  publisher = {Hugging Face},
  howpublished = {\url{https://huggingface.co/datasets/bingbangboom/war-gov-ufo-release-01-images}},
  note = {Release 01 of the Presidential Unsealing and Reporting System for UAP Encounters (PURSUE). 120 PDFs, 4,134 pages.}
}

🔗 Related Resources

• Original release portal: https://www.war.gov/UFO/
• Markdown transcript archive: github.com/DenisSergeevitch/UFO-USA
• National Archives UAP records: https://www.archives.gov/research/topics/uaps
• AARO UAP Records: https://www.aaro.mil/UAP-Records/

🛠️ Usage

from datasets import load_dataset

# Load the full dataset
ds = load_dataset("bingbangboom/war-gov-ufo-release-01-images", split="train")

# Access a single page
sample = ds[0]
print(sample["file_name"])      # Original PDF filename
print(sample["page_number"])    # Page number
print(sample["agency"])         # Originating agency
sample["image"].show()          # Display the page image

Generated by ML Intern

This dataset repository was generated by ML Intern, an agent for machine learning research and development on the Hugging Face Hub.

• Try ML Intern: https://smolagents-ml-intern.hf.space
• Source code: https://github.com/huggingface/ml-intern