-
HouChang X-AI.
-
Thanks for the author's (@argman), help!
-
Thanks for the author's (@zxytim) help!
-
Thanks for the author's (@SakuraRiven)help!
-
Please cite zxytim's paper if you find this useful.
-
Using dice loss instead of class-balanced cross-entropy loss. Some codes refer to argman/EAST and songdejia/EAST
这里一共有两个版本的实现,一个tf1版本的,一个pytorch版本的