DOCCI: Descriptions of Connected and Contrasting Images

arXiv:2404.19753v1 Announce Type: cross
Abstract: Vision-language datasets are vital for both text-to-image (T2I) and image-to-text (I2T) research. However, current datasets lack descriptions with fine-grained detail that would allow for richer associations to be learned by models. To fill the gap, we introduce Descriptions of Connected and Contrasting Images (DOCCI), a dataset with long, human-annotated English descriptions for 15k images that were taken, curated and donated by a single researcher intent on capturing key challenges such as spatial relations, counting, text …

abstract arxiv cs.ai cs.cl cs.cv cs.lg current dataset datasets fine-grained gap however image images image-to-text language research text text-to-image type vision vision-language vital