AVA: Large-Scale Database for Aesthetic Visual Analysis

255,530 images labeled with aesthetic ratings on a scale of 1 to 10 provided by thousands of photographers. The dataset includes semantic tags for 66 categories and 14 photographic style labels for each image, enabling multi-faceted visual analysis.

Use Cases

• Train aesthetic quality assessment models using the distribution of 1-10 ratings
• Develop style classification algorithms using the 14 photographic style labels
• Perform semantic image retrieval based on the 66 provided category tags

Strengths

• 255,530 total images sourced from dpchallenge.com
• Aesthetic ratings on a scale of 1 to 10 for every image
• 66 semantic tags and 14 photographic style labels