DU-VLG: Unifying Vision-and-Language Generation via Dual Sequence-to-Sequence Pre-training

32614

Knowledge Triples

Source Entity	Relationship	Target Entity
DU VLG Method	comparesWith	Image Captioning Task
DU VLG Method	comparesWith	Text to Image Generation Task
DU VLG Method	outperforms	State of the Art Metric
DU VLG Method	uses	Modality Translation Method
DU VLG Method	uses	Multi Modal Denoising Autoencoder Method