LAION Art is a subset of the LAION-5B dataset — a large-scale dataset consisting of five billion CLIP-filtered image-text pairs. This dataset was created for research purposes by having multiple lightweight models estimate how, on a scale from one to ten, people would rate each image, solely based on aesthetics. Researchers then created two datasets: LAION Aesthetics, comprising 120 million images with a score of seven or above, and LAION Art, comprising eight million images with a score of eight or above.
Read the LAION blog post to learn more.
Explore the LAION Art dataset with Labelbox Catalog by visualizing all data rows, filtering the dataset by various values, and viewing metadata for each image.