NeurIPS 2019
Sun Dec 8th through Sat the 14th, 2019 at Vancouver Convention Center
Paper ID:479
Title:Learn, Imagine and Create: Text-to-Image Generation from Prior Knowledge

This paper proposes a new method for text to image generation. Pros • A new method is proposed. • State of the art results are obtained for the task. • The paper is generally clearly written. • The proposed method is reasonable and sound. Cons • There are several issues pointed out by the reviewers. Some of them have been addressed in the author response. • Comparison with some existing methods is missing. The authors have promised to add more. After the author response, the reviewers made a discussion and agreed that the paper can be a strong one if the issues are addressed in the next version.