The VASR dataset is a large-scale visual analogy dataset for situation recognition, where the task is to select an image candidate B’ that completes the analogy (A to A’ is like B to what?).
BibTex:
Before browse our site, please accept our cookies policy
