AI Paper Atlas

Research map / Image generation / Text-to-image diffusion

GeoDiv: Framework for Measuring Geographical Diversity in Text-to-Image Models

Abhipsa Basu, Mohana Singh, Shashank Agnihotri, Margret Keuper, Venkatesh Babu Radhakrishnan

ICLR 2026 · Image generation › Text-to-image diffusion · approach: T2I evaluation and controllable adapters

Read the paperCheck your idea against 137,217 papers

AI-extracted. The measured effects below were read by AI from the paper's full text (no summary yet). Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.

Measured effects reported in the paper

What was changedWhereMeasuredEffect
use Gemini-2.5-flash instead of weaker VLMs as the VQA judge (ablation)GeoDiv VDI · VQA validation, overallaccuracy: 0.68 (LLaVA) -> 0.86 (Gemini-2.5-flash)Improved
compare VLM judges on SEVI Maintenance alignment (Gemini vs gpt-4o) (ablation)GeoDiv SEVI · SEVI validation, MaintenanceSpearman correlation: Gemini 0.69 vs gpt-4o 0.76Mixed
newer diffusion model versions (SD2.1 to FLUX.1) versus olderT2I models · GeoDiv overall scoreGeoDiv score: 0.4251 (SD2.1) -> 0.3153 (FLUX.1)Hurt
FLUX.1 polished generation versus SD modelsFLUX.1-dev · Mean Maintenance ratingrating: Maintenance 3.23 (SD2.1) -> 4.73 (FLUX.1); diversity 0.15Mixed
real-world GeoDE images versus synthetic imagesimage dataset · Entity-Appearance DiversityDiversity-Score: 0.44 -> 0.60Improved
real-world GeoDE images versus synthetic imagesimage dataset · Background-Appearance DiversityDiversity-Score: 0.31 -> 0.42Improved
prompt for wealthy countries (Japan, UAE, UK) versus India, Nigeria, ColombiaT2I models · Mean Affluence ratingrating: 2.31 -> 3.53Improved
use Vendi-Score instead of GeoDiv for non-entity axesVendi-Score · Correlation with GeoDiv MaintenancePearson correlation: 0.56 (entity) vs 0.06 (maintenance)Hurt
use open-source Qwen2.5-VL in place of Gemini for GeoDiv (ablation)GeoDiv · agreement across four diversity axesaverage correlation: rho = 0.83No clear effect
country wealth (GDP) versus GeoDiv scoreT2I models · GeoDiv score by countrycorrelation: rho = -0.27 (GDP), -0.28 (per capita)Hurt

Closest papers by meaning

Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.

Open scime draft

Get the next scime Dispatch and a monthly field update

Free. We send a confirmation link first; every email has a one-click unsubscribe. Privacy