Наборы данных для описания изображений

AbstractHumans learn language by listening, speaking, writing, reading, and also, via interaction with the multimodal real world. Existing language pre-training frameworks show the effectiveness of text-only self-supervision while we explore …...

AbstractBy supporting multi-modal retrieval training and evaluation, image captioning datasets have spurred remarkable progress on representation learning. Unfortunately, datasets have limited cross-modal associations: images are not paired with other images, …...