Towards Multimodal Video Paragraph Captioning Models Robust to Missing Modality

Open in new window