CODEWITHSUNDEEP
machine-learningneural-networkdeep-learningcaffeconv-neural-network
user4911648
user4911648

Reputation:

ND Blob support caffe: legacy accessors fail for blobs > 4 axes

I am getting an error when executing following .prototxt and I have absolutely no idea why I get an error there:

layer {
    name: "conv"
    type: "Convolution"
    bottom: "image"
    top: "conv"
    convolution_param {
        num_output: 2
        kernel_size: 5
        pad: 2
        stride: 1
        weight_filler {
            type: "xavier"
        }
        bias_filler {
            type: "constant"
            value: 0
        }
    }
}

This is the error output. As I have seen in the latest caffe-master-branch it should be possible to use 5D-Blobs.

I1202 14:54:58.617269  2393 hdf5_data_layer.cpp:93] Number of HDF5 files: 9
I1202 14:54:58.631134  2393 hdf5.cpp:35] Datatype class: H5T_INTEGER
I1202 14:54:59.159739  2393 net.cpp:150] Setting up train-data
I1202 14:54:59.159760  2393 net.cpp:157] Top shape: 1 1 1 128 128 (16384)
I1202 14:54:59.159765  2393 net.cpp:157] Top shape: 1 1 8 128 128 (131072)
I1202 14:54:59.159766  2393 net.cpp:165] Memory required for data: 589824
I1202 14:54:59.159773  2393 layer_factory.hpp:77] Creating layer down_level_0_conv
I1202 14:54:59.159790  2393 net.cpp:100] Creating Layer down_level_0_conv
I1202 14:54:59.159795  2393 net.cpp:434] down_level_0_conv <- image
I1202 14:54:59.159804  2393 net.cpp:408] down_level_0_conv -> down_level_0_conv
F1202 14:54:59.159915  2393 blob.hpp:140] Check failed: num_axes() <= 4 (5 vs. 4) Cannot use legacy accessors on Blobs with > 4 axes.

Do I need to go to a certain branch? I just pulled from caffe-master-branch again to make sure it is the newest version. I then made a make clean make all command and it still does not work.

Upvotes: 3

Views: 676

Answers (3)

Dale
Dale

Reputation: 1608

As a complement of Shai's answer, in order to be compatible with the ND Convolution and InnerProduct layer, the "Xavier" filler code

virtual void Fill(Blob<Dtype>* blob) {
...
int fan_in = blob->count() / blob->num();
int fan_out = blob->count() / blob->channels();
Dtype n = fan_in;  // default to fan_in
...
Dtype scale = sqrt(Dtype(3) / n);
caffe_rng_uniform<Dtype>(blob->count(), -scale, scale,
        blob->mutable_cpu_data());
CHECK_EQ(this->filler_param_.sparse(), -1)
         << "Sparsity not supported by this Filler.";
}

in caffe, should be more like this:

...
int fan_in = blob->count() / blob->shape(0);
int fan_out = blob->num_axis() == 2 ? blob->shape(0) : blob->count() / blob->shape(1);
...//original stuff

This little change should also make your prototxt work.

Upvotes: 1

Shai
Shai

Reputation: 114866

AFAIK, this error comes from the "Xavier" filler: this filler computes the ratio between the input and output channels. If you replace it with a different filler you should be Okay with ND blob.

Upvotes: 2

user4911648
user4911648

Reputation:

When I remove the check in line 140 blob.hpp it does work. This is one way to solve it, not the best one though.

(But this cannot be the proper solution. Is there anything else?)

Upvotes: 0

Related Questions