Hugging Face's logo

Datasets:
flickr-foundation
/
flickr-lifeboat-commons-1k-2025

Tasks:
Image Classification
Image-to-Text
Text Retrieval
Modalities:
Image
Text
Formats:
parquet
Languages:
English
Size:
1K - 10K
Tags:
flickr
flickr-commons
data-lifeboat
Libraries:
Datasets
Dask
License:
Dataset card Data Studio Files
xet
Community
1
flickr-lifeboat-commons-1k-2025 / README.md
davanstrien's picture
davanstrien HF Staff
Update dataset card with raw version cross-reference
baf421a verified
preview code
|
raw
history blame contribute delete
7.55 kB
metadata
license: other
license_name: various-open-licenses
license_link: https://www.flickr.com/commons/usage/
tags:
  - flickr
  - flickr-commons
  - data-lifeboat
size_categories:
  - 1K<n<10K
language:
  - en
task_categories:
  - image-classification
  - image-to-text
  - text-retrieval
  - visual-question-answering
pretty_name: Flickr Commons 1K Collection

Flickr Commons 1K Collection

Dataset Description

This dataset is a Flickr Data Lifeboat converted to a machine learning ready/ Hugging Face datasets compatible format. Data Lifeboats are digital preservation archives created by the Flickr Foundation to ensure long-term access to meaningful collections of Flickr photos and their rich community metadata.

What is a Data Lifeboat?

Data Lifeboats are self-contained archives designed to preserve not just images, but the social and cultural context that makes them meaningful:

  • Complete photo metadata and community interactions
  • User-generated tags, descriptions, and comments
  • Attribution and licensing information
  • No external dependencies for long-term preservation

Collection Details:

  • Original Collection ID: COMMONS_1K_2025
  • Created: 2025-01-16 17:03:19.752253
  • Curator: the Flickr Foundation

Purpose

The Flickr Foundation produced this special collection for release in January 2025. It showcases some of the most popular photography across the groundbreaking Flickr Commons program. We hope you enjoy it!

Related Datasets

Raw Data Lifeboat Version

This dataset also has a raw, self-contained archive version that preserves the original Data Lifeboat format:

🗂️ Raw Archive: davanstrien/flickr-lifeboat-commons-1k-2025-raw

The raw version includes:

  • Built-in web viewer (viewer/index.html)
  • Complete original file structure
  • Self-contained archive (no external dependencies)
  • Ideal for digital preservation and archival research

Choose the raw version if you need the complete preservation format or want to use the built-in viewer.

Dataset Statistics

Collection Overview

  • Total Photos: 1,000
  • Total Tags: 13,356
  • Total Comments: 24,831
  • Contributors: 20,281
  • Date Range: 2008-01-08 to 2025-01-04
  • Photos with Location: 216 (21.6%)

License Distribution

  • nkcr: 1000 photos

Format Distribution

  • jpg: 984 photos
  • png: 16 photos

Engagement Metrics

  • Average Views: 101,857 (max: 10,186,629)
  • Average Favorites: 247 (max: 29,270)
  • Average Comments: 0.0 (max: 0)

Top Contributing Institutions/Users

  • NASA on The Commons: 15 photos
  • LlGC ~ NLW: 14 photos
  • IWM Collections: 13 photos
  • Museum of Hartlepool: 13 photos
  • National Library of Australia Commons: 13 photos

Dataset Structure

Field Descriptions

Field Type Description
image Image Original resolution photo
thumbnail Image Thumbnail version
photo_id string Unique Flickr identifier
secret string Flickr URL construction parameter (enables direct access to public domain photos)
url string Original Flickr URL
title string Photo title
description string Photo description
uploader_username string Username of uploader
license_label string Human-readable license
date_taken string When photo was captured
date_uploaded timestamp When uploaded to Flickr
latitude/longitude float GPS coordinates (if available)
count_views/faves/comments int Community engagement metrics
tags list[string] User-generated tags
comments list[dict] Full comment threads with metadata

Note on secret field: This is a Flickr-specific parameter required for constructing direct image URLs. While normally sensitive, it's appropriate to include here because all photos are public domain and this enables the self-contained nature of Data Lifeboat archives.

Usage

Basic Loading 

from datasets import load_dataset

# Load the dataset
dataset = load_dataset("your-username/dataset-name")

# Access images and metadata
for example in dataset["train"]:
    image = example["image"]
    title = example["title"]
    tags = example["tags"]
    comments = example["comments"]

Research Applications

This dataset is suitable for:

  • Computer Vision: Image classification, scene understanding, visual content analysis
  • Vision-Language: Image-to-text generation, visual question answering
  • Digital Humanities: Cultural heritage analysis, community interaction studies
  • Information Retrieval: Text-based image search, tag prediction, metadata analysis
  • Social Media Research: Community engagement patterns, collaborative annotation

Working with Comments and Tags 

# Analyze community engagement
for example in dataset["train"]:
    photo_id = example["photo_id"]
    num_comments = len(example["comments"])
    tags = example["tags"]
    
    print(f"Photo {photo_id}: {num_comments} comments, {len(tags)} tags")

Source Data

Data Collection

This collection originates from Flickr's Commons program, which partners with cultural heritage institutions to share photographs with no known copyright restrictions. Photos are selected based on cultural significance, community engagement, and preservation value.

Community Annotations

All metadata reflects authentic community interaction:

  • Tags: User-generated keywords and descriptions
  • Comments: Community discussions and reactions
  • Engagement metrics: Real view counts, favorites, and interactions

Licensing and Ethics

License Information

Photos in this collection have open licenses enabling reuse:

  • Most common: "No known copyright restrictions"
  • Also includes: Creative Commons licenses (CC0, CC BY, etc.)
  • See license_label and license_url fields for specific licensing per image

Ethical Considerations

  • Historical Context: Some photos may depict people, places, or events from historical periods
  • Community Content: Comments and tags reflect community perspectives at time of creation
  • Attribution: Original photographers and institutions are preserved in metadata
  • Public Domain: All content was publicly accessible on Flickr under open licenses

Citation 

@misc{commons_1k_2025},
  title = {Flickr Commons 1K Collection},
  author = {the Flickr Foundation},
  year = {2025},
  publisher = {Hugging Face},
  url = {https://huggingface.co/datasets/[dataset-url]},
  note = {Flickr Data Lifeboat digital preservation format}
}

Additional Information

Future Considerations

The photos contained in this collection come to you with no known copyright restrictions. You are welcome to use them as you wish. Please consider giving credit to the cultural institutions who have shared these photos with us.

About Data Lifeboats

Learn more about the Data Lifeboat initiative:

Dataset Maintenance

This dataset was created using the Flickr Foundation's Data Lifeboat format and converted to Hugging Face using automated tools that preserve all original metadata and community context.