mlsanityData to 14 Oct 2026

AI's Blind Spots: Geographic Knowledge and Diversity Deficit in Generated Urban Scenario

arXiv 2506.16898 ↗First posted 20 Jun 2025cs.AI cs.CV cs.CYcited by 0

On the map: Diffusion models → 3D, point cloud and scene generation

Abstract (arXiv)

Diffusion-based text-to-image models are increasingly used for urban analysis and scenario generation, but their geographic knowledge and representational biases remain poorly understood. We evaluate FLUX 1-schnell and Stable Diffusion 3.5-Large in the United States by generating 150 street-view images for each state, each state capital, and a generic "USA" prompt. Images are embedded with DINO-v2 ViT-S/14 and compared with Fr'echet Inception Distance (FID). Pairwise FID clustering shows that geographically proximate states and capitals often group together, indicating implicit geographic structure. However, the generic ``USA'' prompt collapses this diversity into a metropolitan stereotype: frontier, desert, tropical, rural, and small-city environments are underrepresented or distant in FID space. These results show that diffusion models can encode fine-grained geography while still reproducing narrow national-scale visual stereotypes.

Key facts. No citing papers in the map yet (data through 14 Oct 2026). Placed in 3D, point cloud and scene generation, one of 14 directions in Diffusion models. Builds on 15 of its 22 references that are in the map. First posted on arXiv 20 Jun 2025.

Builds on

15 of 22 references.

References · 15

  1. 2020Denoising Diffusion Probabilistic Modelscited by 21,076
  2. 2023DINOv2: Learning Robust Visual Features without Supervisioncited by 7,974
  3. 2021GLIDE: Towards Photorealistic Image Generation and Editing with Text-Guided Diffusion Modelscited by 3,614
  4. 2021SDEdit: Guided Image Synthesis and Editing with Stochastic Differential Equationscited by 1,937
  5. 2022Make-A-Video: Text-to-Video Generation without Text-Video Datacited by 1,742
  6. 2020StereoSet: Measuring stereotypical bias in pretrained language modelscited by 924
  7. 2021BBQ: A Hand-Built Bias Benchmark for Question Answeringcited by 701
  8. 2024Survey of Bias In Text-to-Image Generation: Definition, Evaluation, and Mitigationcited by 66
  9. 2024ViSAGe: A Global-Scale Analysis of Visual Stereotypes in Text-to-Image Generationcited by 27
  10. 2024Large Language Models are Zero-Shot Next Location Predictorscited by 22
  11. 2024Towards Geographic Inclusion in the Evaluation of Text-to-Image Modelscited by 16
  12. 2023DIG In: Evaluating Disparities in Image Generations with Indicators for Geographic Diversitycited by 9
3 more
  1. 2024Uncovering the human motion pattern: Pattern Memory-based Diffusion Model for Trajectory Predictioncited by 4
  2. 2025Multimodal Large Language Models Predict Urban Safety Perception but Encode Non-Neutral Demographic Priorscited by 3
  3. 2025The LLM Wears Prada: Analysing Gender Bias and Stereotypes through Online Shopping Datacited by 1

Built on by

Not cited yet.

Similar, not cited

Close in wording, no citation either way

  1. 2026Designing streetscapes from street-view imagery using diffusion models
  2. 2026GeoDiv: Framework For Measuring Geographical Diversity In Text-To-Image Models
  3. 2025Generative AI for Urban Planning: Synthesizing Satellite Imagery via Diffusion Models
  4. 2025UrbanFeel: A Comprehensive Benchmark for Temporal and Perceptual Understanding of City Scenes through Human Perspective
  5. 2024Decomposed evaluations of geographic disparities in text-to-image models
  6. 2024Improving Geo-diversity of Generated Images with Contextualized Vendi Score Guidance
  7. 2025Urban In-Context Learning: Bridging Pretraining and Inference through Masked Diffusion for Urban Profiling
  8. 2023Inspecting the Geographical Representativeness of Images from Text-to-Image Models