Netron:caffemodel weights Tensor data is empty

Question

0.Question can not get conv1 weights data in the bvlc_reference_caffenet_2.caffemodel.

1.info bvle/caffe:https://github.com/BVLC/caffe nvidia-caffe-version:0.17.3

2.convert train caffe model to test model:http://dl.caffe.berkeleyvision.org/bvlc_reference_caffenet.caffemodel prototxt:https://github.com/BVLC/caffe/blob/master/models/bvlc_reference_caffenet/deploy.prototxt code:

import caffe
net = caffe.Net('bvlc_reference_caffenet/deploy_2.prototxt', 'bvlc_reference_caffenet/bvlc_reference_caffenet.caffemodel', caffe.TEST)
new_net.save('bvlc_reference_caffenet/bvlc_reference_caffenet_2.caffemodel')

3.caffemodel test in the caffe container. everything is ok

armnn @container:~/caffe$ ls -lh models/bvlc_reference_caffenet/ total 698M -rw-r--r-- 1 armnn pfcgroup 233M Aug 8 10:47 bvlc_reference_caffenet.caffemodel -rw-r--r-- 1 armnn pfcgroup 233M Aug 8 11:41 bvlc_reference_caffenet_1.caffemodel -rw-r--r-- 1 armnn pfcgroup 233M Aug 10 07:23 bvlc_reference_caffenet_2.caffemodel -rw-r--r-- 1 armnn pfcgroup 2.9K Aug 8 09:10 deploy.prototxt -rw-r--r-- 1 armnn pfcgroup 2.8K Aug 15 08:07 deploy_1.prototxt -rw-r--r-- 1 armnn pfcgroup 2.9K Aug 15 08:07 deploy_2.prototxt -rw-r--r-- 1 armnn pfcgroup 1.3K Aug 8 09:10 readme.md -rw-r--r-- 1 armnn pfcgroup 315 Aug 8 09:10 solver.prototxt -rw-r--r-- 1 armnn pfcgroup 5.6K Aug 8 09:10 train_val.prototxt.

4. show the caffemodel on the Netron

Netron: caffemodel weights Tensor data is empty.

5.deploy on armnn21.02 when using armnn parser caffemodel, cat not find out conv1 weights data in caffemodel. log details:

08-02 21:48:39.866 31795 31795 D armnn: Fatal: Armnn Error: Data blob at index 0 in layer conv1 has an unexpected size. Expected 34848 elements but got 0 elements.

enter image description here

7. deploy_2.prototxt

layer {
  name: "data"
  type: "Input"
  top: "data"
  input_param { shape: { dim: 1 dim: 3 dim: 227 dim: 227 } }
}
layer {
  name: "conv1"
  type: "Convolution"
  bottom: "data"
  top: "conv1"
  convolution_param {
    num_output: 96
    kernel_size: 11
    stride: 4
  }
}
layer {
  name: "relu1"
  type: "ReLU"
  bottom: "conv1"
  top: "conv1"
}
layer {
  name: "pool1"
  type: "Pooling"
  bottom: "conv1"
  top: "pool1"
  pooling_param {
    pool: MAX
    kernel_size: 3
    stride: 2
  }
}
layer {
  name: "norm1"
  type: "LRN"
  bottom: "pool1"
  top: "norm1"
  lrn_param {
    local_size: 5
    alpha: 0.0001
    beta: 0.75
  }
}
layer {
  name: "conv2"
  type: "Convolution"
  bottom: "norm1"
  top: "conv2"
  convolution_param {
    num_output: 256
    pad: 2
    kernel_size: 5
    group: 2
  }
}
layer {
  name: "relu2"
  type: "ReLU"
  bottom: "conv2"
  top: "conv2"
}
layer {
  name: "pool2"
  type: "Pooling"
  bottom: "conv2"
  top: "pool2"
  pooling_param {
    pool: MAX
    kernel_size: 3
    stride: 2
  }
}
layer {
  name: "norm2"
  type: "LRN"
  bottom: "pool2"
  top: "norm2"
  lrn_param {
    local_size: 5
    alpha: 0.0001
    beta: 0.75
  }
}
layer {
  name: "conv3"
  type: "Convolution"
  bottom: "norm2"
  top: "conv3"
  convolution_param {
    num_output: 384
    pad: 1
    kernel_size: 3
  }
}
layer {
  name: "relu3"
  type: "ReLU"
  bottom: "conv3"
  top: "conv3"
}
layer {
  name: "conv4"
  type: "Convolution"
  bottom: "conv3"
  top: "conv4"
  convolution_param {
    num_output: 384
    pad: 1
    kernel_size: 3
    group: 2
  }
}
layer {
  name: "relu4"
  type: "ReLU"
  bottom: "conv4"
  top: "conv4"
}
layer {
  name: "conv5"
  type: "Convolution"
  bottom: "conv4"
  top: "conv5"
  convolution_param {
    num_output: 256
    pad: 1
    kernel_size: 3
    group: 2
  }
}
layer {
  name: "relu5"
  type: "ReLU"
  bottom: "conv5"
  top: "conv5"
}
layer {
  name: "pool5"
  type: "Pooling"
  bottom: "conv5"
  top: "pool5"
  pooling_param {
    pool: MAX
    kernel_size: 3
    stride: 2
  }
}
layer {
  name: "fc6"
  type: "InnerProduct"
  bottom: "pool5"
  top: "fc6"
  inner_product_param {
    num_output: 4096
  }
}
layer {
  name: "relu6"
  type: "ReLU"
  bottom: "fc6"
  top: "fc6"
}
layer {
  name: "drop6"
  type: "Dropout"
  bottom: "fc6"
  top: "fc6"
  dropout_param {
    dropout_ratio: 0.5
  }
}
layer {
  name: "fc7"
  type: "InnerProduct"
  bottom: "fc6"
  top: "fc7"
  inner_product_param {
    num_output: 4096
  }
}
layer {
  name: "relu7"
  type: "ReLU"
  bottom: "fc7"
  top: "fc7"
}
layer {
  name: "drop7"
  type: "Dropout"
  bottom: "fc7"
  top: "fc7"
  dropout_param {
    dropout_ratio: 0.5
  }
}
layer {
  name: "fc8"
  type: "InnerProduct"
  bottom: "fc7"
  top: "fc8"
  inner_product_param {
    num_output: 1000
  }
}
layer {
  name: "prob"
  type: "Softmax"
  bottom: "fc8"
  top: "prob"

}

Answer 1

bvlc_refernece_caffenet.caffemodel is caffe1.0, but using caffev2 to convert train model to deploy model according to deploy.proto.

And of course, Netorn can support show the caffemodel only from caffev2.