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

Due to the limitations of the model structure and pre-training objectives, existing visionand-language generation models cannot utilize pair-wise images and text through bidirectional generation. In this paper, we propose DU-VLG, a framework which unifies vision-and-language generation as sequence generation problems. DU-VLG is trained with novel dual pre-training tasks: multi-modal denoising autoencoder tasks and modality translation tasks. To bridge the gap between image understanding and generation, we further design a novel commitment loss. We compare pre-training objectives on image captioning and text-to-image generation datasets. Results show that DU-VLG yields better performance than variants trained with uni-directional generation objectives or the variant without the commitment loss. On the image captioning task, our model reaches better performance than other pre-trained systems. On text-toimage generation datasets, our model achieves better or comparable results than previous state-of-the-art models. In addition, human judges further confirm that our model generates real and relevant images as well as faithful and informative captions. Image Captioning Ground Truth: Rows of unripe bananas on a display shelf. DU-VLG: ! Several bunches of green bananas are on a shelf. w/o dual pre-training: " A bunch of green bananas sitting on a table. Ground Truth DU-VLG! w/o dual pre-training" Input: Rows of unripe bananas on a display shelf.