Abstract | ||
---|---|---|
In this work, we propose a new task called Story Visualization. Given a multi-sentence paragraph, the story is visualized by generating a sequence of images, one for each sentence. In contrast to video generation, story visualization focuses less on the continuity in generated images (frames), but more on the global consistency across dynamic scenes and characters - a challenge that has not been addressed by any single-image or video generation methods. Therefore, we propose a new story-to-image-sequence generation model, StoryGAN, based on the sequential conditional GAN framework. Our model is unique in that it consists of a deep Context Encoder that dynamically tracks the story flow, and two discriminators at the story and image levels, to enhance the image quality and the consistency of the generated sequences. To evaluate the model, we modified existing datasets to create the CLEVR-SV and Pororo-SV datasets. Empirically, StoryGAN outperformed stateof-the-art models in image quality, contextual consistency metrics, and human evaluation. |
Year | DOI | Venue |
---|---|---|
2018 | 10.1109/CVPR.2019.00649 | 2019 IEEE/CVF CONFERENCE ON COMPUTER VISION AND PATTERN RECOGNITION (CVPR 2019) |
Field | DocType | Volume |
Pattern recognition,Visualization,Computer science,Image quality,Paragraph,Encoder,Natural language processing,Artificial intelligence,Global consistency,Sentence | Journal | abs/1812.02784 |
ISSN | Citations | PageRank |
1063-6919 | 7 | 0.40 |
References | Authors | |
0 | 9 |
Name | Order | Citations | PageRank |
---|---|---|---|
Yitong Li | 1 | 44 | 7.98 |
Zhe Gan | 2 | 319 | 32.58 |
Yelong Shen | 3 | 709 | 35.97 |
Jingjing Liu | 4 | 88 | 13.20 |
Yu Cheng | 5 | 615 | 55.76 |
Yuexin Wu | 6 | 99 | 5.78 |
L. Carin | 7 | 4603 | 339.36 |
David E. Carlson | 8 | 182 | 15.35 |
Jianfeng Gao | 9 | 5729 | 296.43 |