Unsplash, an online image provider, has released an open image dataset. The new Library currently contains over 2 million high-quality images, with 16GB of accompanying data. The accompanying data includes keyword-image conversions (in search results), community and AI-generated keywords, EXIF, location and landmarks, image categories/subcategories, user-generated collection, image views, and download statistics.
Unsplash separates itself from other open-source image libraries based on the quality of the pictures and the accompanying data. The dataset is free to use for anyone conducting research, and the company thinks it will be especially helpful for researchers in Artificial Intelligence and Machine Learning. A subset of the library is available for commercial use.
Over 200,000 photographers have contributed to the creation of the new dataset. This represents billions of searches from thousands of applications in various use cases and contexts. To get started, users need a developer account. From there, the Docs include how-to instructions including setting up an application, searching by keyword, new photos, and random photo selection.
The Docs are also available through GitHub. The Unsplash team welcomes feedback on the project and encourages developers to share this new tool. Unsplash is pitching it as “the most complete high-quality open image set ever released” and hopes it will be as successful as the Unsplash APITrack this API released in 2016.