softwaregitgub

but,when i use python2 to make mdb still have this wrong?

and i use other's correct mdb still have this question,so it should not be the problem of data.

> I have the same question.Have you solved it? No, i don't know how to solve it? Do you have any ideas?

i change the dataset.py=======> return self.nSamples + 1 to self.nSamples, and it works.

stay: getting data.. -> done Storing the output in: results/SynthText.h5 0 of 4 ** instance # : 0 Traceback (most recent call last): File "/home/stay/Desktop/SynthTextPython3/synthgen.py", line 668, in render_text regions['homography_inv'][ireg])...

> 我用tensorflow已经完成fots的编写，单独的检测和识别的分支训练效果都没问题，但是合并在一起训练效果就比较差，期待pytorch的实验结论～ 老师可以借鉴下您的代码吗，邮箱：[email protected] 非常感谢！