Skip to content

A good example of deformable convolutional network for mnist classification

License

Notifications You must be signed in to change notification settings

xingshulicc/deformable-convolution-Neural-Network

Folders and files

NameName
Last commit message
Last commit date

Latest commit

 

History

8 Commits
 
 
 
 
 
 
 
 
 
 
 
 
 
 

Repository files navigation

deformable-convolution-Neural-Network

A good example of deformable convolutional network for mnist classification

Paper Reference: Deformabel convolutional netwokrs -- Jifeng Dai, Haozhi Qi, Yuwen Xiong, Yi Li, et al. 5 Jun 2017

Code Reference: https://github.com/felixlaumon/deform-conv

The code is written by keras with tensorflow backend. (Theano backend does not work)

Deformable convolutional networks can reduce the effect of geometric transformation on image classification accuracy. In which, the additional convolutional layers are used to learn unknown affine transformations. On the other hand, this structure increases training cost -- time consumption and computational source.

A review of bilinear interpolation is also presented in this repository and I also explian how to realize this algorithm with tensorflow code library. Please refer to tf_map_coords.pptx file.

In cnn.py I defined a deform_cnn function. You can easily construct your own network structures by using this function.

Any discussion about CNN model is welcome and please send email to me: [email protected].

About

A good example of deformable convolutional network for mnist classification

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published

Languages