golbstein / edvr_keras

Keras implementation of EDVR: Video Restoration with Enhanced Deformable Convolutional Networks

EDVR Video-Enhancement and Super-Resolution in Keras

EDVR: Video Restoration with Enhanced Deformable Convolutional Networks

Keras implementation of "Video Restoration with Enhanced Deformable Convolutional Networks" CVPR, 2019: EDVR Based on author's Pytorch implementation

Model Architecture

TODO:

Help needed in translation of authors' DFA module to Keras\Tensorflow: DCN_V2

• PreDeblur Module
• Alignment with Pyramid, Cascading and Deformable Convolution (PCD) Module
• Temporal Spatial Attention (TSA) Fusion Module
• Reconstruction Module
• Pixel shuffle layer (Subpixel)
• Support changing frame shape (currently supports only fixed shape)

Results

• Dependencies

  • Python 3.6
  • Keras>2.2.x

edvr_keras's Issues

Simple Test

Hi,
I would like to test your work on a simple medical video I have.
How can I do it?
Regards,

some advice

hi,there's a typo in the RB function as the second conv2d layers' input should be "out".And also you can refer to this implementation which implemented by Keras.It may be useful because i can output a model summary on my pc but i have not taken any test by using this implementation.if you have enough resources to run a model,think you can try this.

May have found an implementation deformable conv

Hi,

Thank you for your initiative to implement EDVR with Keras. I noticed that you have not yet implemented the deformable convolution. I saw one implemented with tf, the link is https://github.com/DHZS/tf-deformable-conv-layer/blob/master/nets/deformable_conv_layer.py
I cannot wait to read what do you think about it.

Cheers !

Deformable Convolution

Hello!
I don't understand one thing in your code. Why have you commented out all the deformable layers parts? Can you please shed some light on this? If that's not the case, then can you please point out the Keras implementation of DeformableConvolution() in the code?
Thank You!