Bridging the Digital Divide: Performance Variation across SocioEconomic Factors in Vision-Language Models

This repo contains code (in Jupyter Notebooks) for the analysis featured in our EMNLP 2023 paper:

Bridging the Digital Divide: Performance Variation of Vision Language Models across SocioEconomic Factors

Joan Nwatu$, [Oana Ignat](https://oanaignat.github.io/)$, and Rada Mihalcea

($ equal contribution)

Data

The data we used is from the GapMinder project called Dollar Street.

View the images of various households across the world through Dollar Street: photos as data to kill country stereotypes

The dataset is available for download at ML Commons Dollar Street Dataset

More information about paper insights coming soon!