Download link in gen.py not working about synthtext

[ankush-me]

1. I'm able to download the file locally. It could be that the server is blocking your requests.
   I've uploaded it to Google Drive here:
   https://drive.google.com/open?id=0B6vslVxoQlvIZXZqYWFSc0htMVU
2. The demo is using a variant. The code is not available yet.

[ihph]

OK The link you posted in your reply works - I got the data...
Nice work by the way and thank you for open-sourcing !

( another off-topic question: Have you compared your approach to (Text Flow: A Unified Text Detection System in Natural Scene Images : Shangxuan Tian, Yifeng Pan, Chang Huang, Shijian Lu, Kai Yu, Chew Lim Tan
https://arxiv.org/abs/1604.06877 - I'm trying to choose which one to reproduce... )
Re-read the papers..you get better recall and they get better precision, but end up with f = 80.25 in icdar2013 - I'll try to reproduce your approach thx

[ankush-me]

Great. Thank you!

[JcmeLs]

@ankush-me Hello,I download data from Google Drive.But when i run gen.py,it have TypeError：

2017-03-07 15:37:11.706 Python[71212:46633970] 15:37:11.706 WARNING:  140: This application, or a library it uses, is using the deprecated Carbon Component Manager for hosting Audio Units. Support for this will be removed in a future release. Also, this makes the host incompatible with version 3 audio units. Please transition to the API's in AudioComponent.h.
0 of 4
Traceback (most recent call last):
  File "/Users/liba/Downloads/SynthText-master/synthgen.py", line 616, in render_text
    regions = self.filter_for_placement(xyz,seg,regions)
  File "/Users/liba/Downloads/SynthText-master/synthgen.py", line 392, in filter_for_placement
    res = get_text_placement_mask(xyz,seg==l,regions['coeff'][idx],pad=2)
  File "/Users/liba/Downloads/SynthText-master/synthgen.py", line 259, in get_text_placement_mask
    place_mask = 255*np.ones((np.ceil(COL)+pad,np.ceil(ROW)+pad),'uint8')
  File "/Library/Frameworks/Python.framework/Versions/2.7/lib/python2.7/site-packages/numpy/core/numeric.py", line 192, in ones
    a = empty(shape, dtype, order)
TypeError: 'numpy.float64' object cannot be interpreted as an index  

I don't know how to fix it.Can you help me ?
Thanke you

[ankush-me]

It seems that the update to NumPy breaks indexing by floats.
You just need to cast the floats (change line 259 in synthgen.py to the following):

place_mask = 255*np.ones((int(np.ceil(COL))+pad,int(np.ceil(ROW))+pad),'uint8')

[JcmeLs]

@ankush-me Thanks，it can work now.And if want to create my own data set,I must do that:

1. Use depth-from-single image ConvNets to create depth image
2. Use Holistic Edge Detection to create segmentation image
3. Use h5py to save to dset.h5

is it?

[ankush-me]

Yes, correct.