Tian Zhi

@Jayhello Chinese recognition remains an open problem and we currently have no idea as well.

@Jayhello Personally speaking, I think the method you mention can be used to recognize Chinese. However, you need to train it with Chinese data. Thank you:-).

@Jayhello I am sorry that I didn't use it. I think you may need to retrain it with your own training data. Thank you:-).

@lxx1884896 The max size we used over ICDAR2015 is 2200. Hope it is useful to you. Thank you.

@rayush7 Thank you for your interest. 1. Only the bounding boxes ground truth values and the input images are needed. 2. I don't know because I'm not familiar with it....

@melody-rain 1. Not yet. 2. Yes. 3. You can refer to the issue #15 to find out the difference. Thank you very much:-).

@AndyVer2 Of course, you can realize it using the Caffe. Thank you.

@raaka1 We're so sorry for that. The demo website got unstable a few days ago because of some unknown reasons. We're trying to fix it. Thank you very much:-).

@raaka1 hi, we have fixed the problem of the online demo. It's now up and running. Thank you:-).

@JacobianTang in test phase, the text proposals are firstly connected into text line, and then the side-refinement is applied to the text proposals on both ends.