How to use VGG-S and how to extract output features by it?

Question

Tripoli Settou 2018-2-21

0
链接

此问题的直接链接

https://ww2.mathworks.cn/matlabcentral/answers/383946-how-to-use-vgg-s-and-how-to-extract-output-features-by-it

Hi everyone,

In deep learning there are many model of convolution neural network CNN. To try VGG-S model, I download "imagenet-vgg-s.mat" from here and I try it by this code to extract the output feature from 2nd fully connected layer:

    net = load('./Model/imagenet-vgg-s.mat');
    layer = net.layers{1,18}.name;
    outputFeatures = activations(net,Img,layer);

but this error shown:

    _ *Undefined function 'activations' for input arguments of type 'struct'.
      Error in Project (line 31)
            outputFeatures = activations(net,Img,layer);*_

So I download prototxt and caffe model of it "VGG_CNN_S.caffemodel" and "VGG_CNN_S_deploy.prototxt" from this link here . I try it by this code

protofile = './model/ VGG_CNN_S_deploy.prototxt ';
datafile = './model/ VGG_CNN_S.caffemodel ';
net = importCaffeNetwork(protofile, datafile);

but this error is shown:

* _Error using
nnet.internal.cnn.caffe.CaffeModelReader/checkIfPoolingLayerOutputSizeMatchesCaffe
(line 1081)
The pooling layer 'pool1' is not compatible with MATLAB. Caffe computes the output
size as [37 37 96] but MATLAB computes it as [36 36 96].
Error in nnet.internal.cnn.caffe.CaffeModelReader/makePoolingLayer (line 928)
            checkIfPoolingLayerOutputSizeMatchesCaffe(this, inputSize, internalLayer);
Error in nnet.internal.cnn.caffe.CaffeModelReader/importLayers (line 257)
                        [layers(layerIdx), internalLayer, outputSize] =
                        makePoolingLayer(this, origLayerIdxInFile, outputSize);
Error in nnet.internal.cnn.caffe.CaffeModelReader/importSeriesNetwork (line 310)
            layers = importLayers(this);
    Error in importCaffeNetwork (line 82)
        network = importSeriesNetwork(readerObj);_*