search

acaglayan / exploitCNN-RNN

Exploiting Multi-Layer Features Using a CNN-RNN Approach for RGB-D Object Recognition (ECCV 2018 workshops)
12 stars 2 forks source link

missing folder #2

Open GianniMoreandy opened 4 years ago

GianniMoreandy commented 4 years ago

hi, I tried this code but I continue receive this error: Extracting CNN features for surface normals of depth images... Warning: Name is nonexistent or not a directory: /home/matt/Desktop/exploitCNN-RNN-master/matconvnet-1.0-beta25/matlab/mex

In path (line 109) In addpath (line 86) In vl_setupnn (line 13) In run (line 91) In extractCNNFeatures (line 2) In main (line 90) Warning: MatConvNet is not compiled. Consider running vl_compilenn. In vl_setupnn (line 24) In run (line 91) In extractCNNFeatures (line 2) In main (line 90)

SurfNorm CNN features are passing through RNNs for Pool-1 (5) layer ... RNN Forward Prop Train... Error using Incorrect dimensions for matrix multiplication. Check that the number of columns in the first matrix matches the number of rows in the second matrix. To perform elementwise multiplication, use '.'.

Error in forwardRNN>forward (line 46) child = W * reshape(tree(:,row:row+RFS(1)-1,col:col+RFS(2)-1,:),[],numImgs);

Error in forwardRNN (line 10) train = forward(train, rnn, params);

Error in main (line 103) [surfNormPool1RNNTrain, surfNormPool1RNNTest] = forwardRNN(surfNormNetTrain.pool1, surfNormNetTest.pool1, params);

What I'm supposed to do?

acaglayan commented 4 years ago

Hi, first you need to install MatConvNet toolbox, which requires a proper C/C++ compiler, please check the below link for installation about MatConvNet: https://www.vlfeat.org/matconvnet/quick/ If you are able to extact CNN features, then, there should not be any problem on RNN part. Please debug the code step by step to see the problems.