qikizh / ee-gan

Explore and then Exploit: A Fine-grained Hierarchical Network for Text-to-Image Synthesis

EE-GAN

The project is an implement of our paper "Explore and then Exploit: A Fine-grained Hierarchical Network for Text-to-Image Synthesis", which has been accepted in China Multimedia 2022 (GuiYang).

Moreover, it also provides some convenient and clear implements of T2I functions, such as metric measuring and visualization. In other word, These functions are able to be plug and play in your code.

Hen tao Sheng, Qike Zhao, Junchen Zhu, Lianli Gao, Jingkuan Song.

UESTC, School of Computer Science and Engineering, Center for Future Media.

该项目是我们的论文《发掘和利用：细粒度层次化网络的文本到图像生成》的实现，该论文已经被2022年**多媒体大会（贵阳）接受。

同时，该项目提供了一系列便捷和简洁的文本生成图像方向的函数实现，比如评价指标和图像可视化。换而言之，这些函数可以在你的项目中即插即用。