Is it possible for you to release the synthetic dataset generation code? Thank you so much!

GuillermoCarbajal / NonUniformBlurKernelEstimation

Code associated with arXiv paper Non-uniform Motion Blur Kernel Estimation via Adaptive Decomposition.

40 stars 6 forks source link

Is it possible for you to release the synthetic dataset generation code? Thank you so much! #1

Open zoezhou1999 opened 3 years ago

zoezhou1999 commented 3 years ago

Your work is very interesting! Thank you so much!

Kamalhsn commented 3 years ago

Hi, I am also working on a similar project. Can you provide any reference to creating a non-uniform motion blur image dataset?

liana07151018-yang commented 3 years ago

Hi York work is really interesting and I have tested the Lai dataset by running the image_deblurring.py script, but I did not get the same result as Figure 12 in the paper. I used the default parameter configuration and pre-trained network. Is there any problem with me ?

GuillermoCarbajal commented 2 years ago

Hi, we obtained those results with half-size images. Please, see the caption. The reason for subsampling is that we estimate kernels of size 33.

panyangh commented 1 year ago

Hi. I'm so sorry to bother you,but,the code of NonUniformBlurKernelEstimation Pretrained model link is not valid, can you provide a new link? because i do not connect authors