DIG-In
Evaluate the quality, diversity, and consistency of image generation models across different geographical regions.
PremiumNewProductImageImage GenerationGeographic Diversity
DIG-In is a library for evaluating the quality, diversity, and consistency differences of text-to-image generation models across geographical regions. It utilizes GeoDE and DollarStreet as reference datasets and computes metrics such as accuracy, coverage, and diversity of generated images. It also employs the CLIPScore metric to measure model consistency. This library empowers researchers and developers to conduct geographical diversity audits of their image generation models, ensuring fairness and inclusivity on a global scale.
DIG-In Visit Over Time
Monthly Visits
488643166
Bounce Rate
37.28%
Page per Visit
5.7
Visit Duration
00:06:37