jannemus / deepgyro Goto Github PK

Hello, I have read your another paper , which is similar to DeepGyro, named "Fast Motion Deblurring for Feature Detection and
Matching Using Inertial Measurements". Could you share the Paper code? I want to learn how you implement it in code.

Hi, @jannemus, great work!

Could you provide code for generation of blur fields?

I follow the readme to test the effect in paper,but i got this.


could you please tell me how to transform this image to the natural image,Thanks!

Hi! Thanks for your sharing。I want to know the method of using blur fields to generate the blurred image in your paper. Could you share your code about it?

Can you provide the link for the dataset(100k images) used for training the Deepgyro architecture?

Interesting method, but DeepBlind is not usually performs better than DeblurGAN in your proj samples. So, this may be not robust in reality?

the paper mentioned "The data generation tool will be made publicly available upon the publication of the paper ", can you please share the data generation tool for training of deepgyro? Thanks a lot.

I need train the model on my dataset, can you share the train code?Thank you very much!

hi

I could not find inference speed info in the paper, does someone have some measurement to share?
Also, I see the weight file is pretty large (335Mb), so I assume the memory usage is large too, but does anyone have measurement to share on this too?

thanks