1

Image retrieval on real-life images with pre-trained vision-and-language models
We extend the task of composed image retrieval by introducing the first open-domain dataset along with a transformer based method.
Zheyuan (David) Liu, Cristian Rodriguez-Opazo, Damien Teney, Stephen Gould
PDF Cite Dataset Project Preprint (incl. supp. mat.)
Image retrieval on real-life images with pre-trained vision-and-language models