AttnGAN: Fine-Grained Text to Image Generation with Attentional Generative Adversarial Networks

• Published in: 2018 IEEE/CVF Conference on Computer Vision and Pattern Recognition pp. 1316 - 1324
• Main Authors: Xu, Tao, Zhang, Pengchuan, Huang, Qiuyuan, Zhang, Han, Gan, Zhe, Huang, Xiaolei, He, Xiaodong
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.06.2018
• Subjects: Computational modeling; Gallium nitride; Generative adversarial networks; Generators; Image generation; Semantics; Visualization
• DOI: 10.1109/CVPR.2018.00143

In this paper, we propose an Attentional Generative Adversarial Network (AttnGAN) that allows attention-driven, multi-stage refinement for fine-grained text-to-image generation. With a novel attentional generative network, the AttnGAN can synthesize fine-grained details at different sub-regions of the image by paying attentions to the relevant words in the natural language description. In addition, a deep attentional multimodal similarity model is proposed to compute a fine-grained image-text matching loss for training the generator. The proposed AttnGAN significantly outperforms the previous state of the art, boosting the best reported inception score by 14.14% on the CUB dataset and 170.25% on the more challenging COCO dataset. A detailed analysis is also performed by visualizing the attention layers of the AttnGAN. It for the first time shows that the layered attentional GAN is able to automatically select the condition at the word level for generating different parts of the image.