This repo contains instructions for how to prepare a comprehensive dataset of images with captions/descriptions for blind and low vision users. We prepare data from the following sources: SciCap dataset: here WiT dataset: here Twitter (scraped using tweepy and keywords)