I am trying to do 3d image deconvolution using convolution neural network. But I cannot find many famous 3d convnets. Can any one point out some for me?

Background: I am using PyTorch, but any language is OK. What I want to know most is the network structure. I can't find papers on this topic.

Links to research papers would be especially appreciated.

  • 1
    $\begingroup$ Programming language? $\endgroup$ – FreezePhoenix Apr 10 '18 at 13:41
  • 1
    $\begingroup$ I am using PyTorch, But PyTorch or TensorFlow or Theano all OK. What I want to know most is the network structure. I can't find papers on this topic. Any link to papers would be appreciated. $\endgroup$ – Rickyim Apr 10 '18 at 14:19
  • 1
    $\begingroup$ Is there any chance you could use JS? $\endgroup$ – FreezePhoenix Apr 10 '18 at 14:23
  • $\begingroup$ Any language is OK. $\endgroup$ – Rickyim Apr 11 '18 at 7:59
  • $\begingroup$ I think we can borrow some nets from 3D scenario completion like VOXELNET. $\endgroup$ – Rickyim Apr 11 '18 at 8:00

There are many approaches for training CNN on 3d data , but the decision to use a particular architechture is heavily dependant upon the format of you dataset. if you are using 3d point cloud data , i would suggest you to go through pointnet , pointcnn literature https://arxiv.org/abs/1612.00593 https://github.com/yangyanli/PointCNN but training CNN on 3d point clouds is very tough. there is also a way to train convnets by posing the 3d structure from different viewpoints (multiview cnn https://arxiv.org/abs/1505.00880?context=cs) but remember that training cnn on 3d data is really a tough task , if you plan to use a voxelized input data format i suggest to go through (voxelnet .) since you are mentioning about deconvolution the most relavant paper i can come across is https://arxiv.org/abs/1606.06650.

But deconvolution in its own right is a very expensive operation which acting on 3d data makes it very hard, so i would suggest you to check for alternate methods.

  • $\begingroup$ Thanks! And for the last sentence, I understand the 3d convolution is a memory consuming operation. Is it because of that that you say it is hard. And also what is the alternate methods except optimization methods like RLTV and etc. ? Thanks again $\endgroup$ – Rickyim May 30 '18 at 12:59
  • $\begingroup$ yes , deconvolution is more consuming than convolution due to upsampling involved in it . regarding optimization methods i don't know much. PS: can you expand RLTV so that i can check? $\endgroup$ – riemann77 May 30 '18 at 17:02
  • $\begingroup$ why is it difficult to learn on from 3D data? $\endgroup$ – Pinocchio Aug 14 '19 at 19:47

Your Answer

By clicking “Post Your Answer”, you agree to our terms of service, privacy policy and cookie policy

Not the answer you're looking for? Browse other questions tagged or ask your own question.