search

ankush-me / SynthText

Code for generating synthetic text images as described in "Synthetic Data for Text Localisation in Natural Images", Ankush Gupta, Andrea Vedaldi, Andrew Zisserman, CVPR 2016.
http://www.robots.ox.ac.uk/~vgg/data/scenetext/
Apache License 2.0
2.01k stars 621 forks source link

Text visibility on image #253

Open RGBmatrix3 opened 2 years ago

RGBmatrix3 commented 2 years ago

Hi @ankush-me ,

when I load other photos from Synthtext dataset, on some images color of text is similar to background (see on top of the image). What I need to change? Figure_1

And one more question: Can script generate text shape something like this (snake shape text):

Snake shape text

ankush-me commented 2 years ago

For text contrast check here and here.

There's support for a single semicircle here, but not for a 'snake' shape as shown.

92xianshen commented 2 years ago

邮件已收到，谢谢。。。I have received your email. Thank you.

RGBmatrix3 commented 2 years ago

@ankush-me
IM_TOP and IM_BACK , is that background image and text instance?? In poisson_reconstruct.py

There says: combine images using poission editing. IM_TOP and IM_BACK should be of the same size.

But what images script combines?

RGBmatrix3 commented 2 years ago

@ankush-me

Also in the poisson_reconstruct.py

What is these gxs,gys,gxd,gyd? This is X and Y coordinates of something? 

[gxs,gys] = get_grads(ims)
[gxd,gyd] = get_grads(imd)

Is the alpha blend done here? 

 elif mode=='blend': # from recursive call:
            # just do an alpha blend
            gx = gxs+gxd
            gy = gys+gyd

When I replace 'blend' instead 'max' in function blit_images(), nothing happens?

ankush-me commented 2 years ago

Poisson blending is based on the code here: http://web.media.mit.edu/~raskar/photo/code.pdf May be check there for comments to understand it better.