Authors: Filippo Merlo*, Ece Takmaz, Wenkai Chen, Albert Gatt
*Corresponding author:
f.merlo.research@gmail.com
Affiliations: Utrecht University, University of Trento
Status: Submitted to TACL (under review)
Code & Data:
https://github.com/cs-nlp-uu/sceneregCOOCO is a large-scale dataset designed to investigate how Vision-Language Models (VLMs) leverage scene context during referring expression… See the full description on the dataset page:
https://huggingface.co/datasets/fmerlo/COOCO.
