Skip to main content

What EXIF data photographs still carry: the dataset

One row per fetched file across 3000 JPEG files drawn at random from Wikimedia Commons, with every metadata field the reader found recorded field by field, and the platform’s own metadata field recorded beside it for comparison.

What EXIF Data Do Photos Still Carry? reads the measurement. This page is the dataset itself.

What the sample is

Wikimedia Commons, file namespace, JPEG files only, drawn with the platform's own random generator. 3000 files were fetched; 2984 are scored and 16 are excluded, and every exclusion is flagged in a column of its own rather than removed. The files were uploaded to Commons between 2004 and 2026.

The table carries no image content, no coordinate values and no serial numbers, only whether a field was present. The uploader identifier is a hash, so a row cannot be traced back to an account.

What a row carries

Which file it is and how it was fetched. row_id, upload_year, http_status, versions, uploader_hash, excluded, parse_error and whole_file_read. The uploader identifier is a hash, and no account name is published.

What the file itself carries. has_exif, make, model, body_serial_exif, lens_serial_exif, serial_xmp, gps_block, gps_all_zero, datetime_original and software.

What the platform reports. platform_make, platform_gps and platform_serial, read from the platform’s own metadata field rather than from the file.

What the second reader found. The exiftool columns, which report the same serial-number question asked of the standard fields, of XMP, and of the maker notes separately.

What is here

  • DATASET.csv The measurements: one row per fetched file, one column per field read. It carries no image content, no coordinate values and no serial numbers, only whether a field was there.
  • DATASET.json The run: where the sample came from, how the reading program was checked, the second reader it was compared against, and the counts behind every group.
  • PREREGISTRATION.md The question, the predicted shares and the counting rules, all fixed before the first file was fetched, with every later change recorded beside it.
  • LICENSE.txt CC0 1.0 public domain dedication, so the data can be copied, changed and reused without asking.

The program that produced the table is published beside it:

  • analyze.py Turns the published table into every number the article reports, with nothing typed in by hand.

How to cite

Viallo, What EXIF Data Do Photos Still Carry?, measured 10 September 2026. Viallo. https://www.viallo.app/studies/what-exif-data-photos-still-carry

Licence

CC0 1.0 Universal, a public domain dedication. You may copy, modify, distribute and perform the work, even for commercial purposes, all without asking permission.