Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

论文阅读与实现--DDR - 阮少辉的博客 | Slade Blog #1

Open
eragonruan opened this issue Nov 13, 2017 · 8 comments
Open

论文阅读与实现--DDR - 阮少辉的博客 | Slade Blog #1

eragonruan opened this issue Nov 13, 2017 · 8 comments

Comments

@eragonruan
Copy link
Owner

http://slade-ruan.me/2017/10/24/text-detection-ddr/

@ilovin
Copy link

ilovin commented Nov 13, 2017

mark

@argman
Copy link

argman commented Nov 14, 2017

博主复现了他们的结果吗?看到博主做了一些改进,问下会有论文或者arxiv么

@eragonruan
Copy link
Owner Author

@argman 模型基本能复现,但是文章有一部分没讲清楚或者说是我没理解吧(主要是真值的标定那部分),除开这部分,测试部分也没有按照文章里提的,用sliding windows的策略,当时直接整图测试的,其他完全按照文章的话离文章里的结果还是有差距的。F-measure可能差了10个点吧。论文暂时没有

@xiaomaxiao
Copy link

直接回归的方法 ,对于长文本行,是乏力的吧,博主有没有测试过?

@eragonruan
Copy link
Owner Author

对的,对尺度比较敏感

@ihollywhy
Copy link

EAST和这篇论文思路简直就是一模一样的。。。

@ZhuanDJ
Copy link

ZhuanDJ commented Jun 28, 2018

我最近在复现这篇论文,想请教一下博主对真值标定的具体形式,特别是回归的标注这部分的理解。另外博主有没有已复现的程序放在github上,想学习一下

@eragonruan
Copy link
Owner Author

@ZhuanDJ 回归这部分我的理解是每个像素点都预测8个距离,分别是距离左上,右上,右下,左下文本角点的水平位移和竖直位移。就是直接回归。

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Projects
None yet
Development

No branches or pull requests

6 participants