Oct 172022 Image captioning datasets can be limited for languages other than English So, today we are releasing the Crossmodal 3,600 dataset, which provides 261,375 reference captions in 36 languages for a geographically diverse set of 3,600 images.
