Языковая генерация на основе визуальной информации

AbstractA major challenge in visually grounded language generation is to build robust benchmark datasets and models that can generalize well in real-world settings. To do this, it is critical to …...