- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
name: "j"
layer {
name: "data"
type: "Input"
top: "data"
input_param {
shape {
dim: 1
dim: 3
dim: 224
dim: 224
}
}
}
layer {
name: "conv1"
type: "Convolution"
bottom: "data"
top: "conv1"
convolution_param {
num_output: 64
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 4
pad_w: 4
kernel_h: 7
kernel_w: 7
stride_h: 2
stride_w: 2
}
}
layer {
name: "relu2"
type: "ReLU"
bottom: "conv1"
top: "conv1"
}
layer {
name: "pool3"
type: "Pooling"
bottom: "conv1"
top: "pool3"
pooling_param {
pool: MAX
kernel_h: 3
kernel_w: 3
stride_h: 3
stride_w: 3
pad_h: 0
pad_w: 0
}
}
layer {
name: "conv4"
type: "Convolution"
bottom: "pool3"
top: "conv4"
convolution_param {
num_output: 128
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 1
pad_w: 1
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu5"
type: "ReLU"
bottom: "conv4"
top: "conv4"
}
layer {
name: "conv6"
type: "Convolution"
bottom: "conv4"
top: "conv6"
convolution_param {
num_output: 128
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 0
pad_w: 0
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu7"
type: "ReLU"
bottom: "conv6"
top: "conv6"
}
layer {
name: "conv8"
type: "Convolution"
bottom: "conv6"
top: "conv8"
convolution_param {
num_output: 128
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 0
pad_w: 0
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu9"
type: "ReLU"
bottom: "conv8"
top: "conv8"
}
layer {
name: "conv10"
type: "Convolution"
bottom: "conv8"
top: "conv10"
convolution_param {
num_output: 128
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 0
pad_w: 0
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu11"
type: "ReLU"
bottom: "conv10"
top: "conv10"
}
layer {
name: "pool12"
type: "Pooling"
bottom: "conv10"
top: "pool12"
pooling_param {
pool: MAX
kernel_h: 2
kernel_w: 2
stride_h: 2
stride_w: 2
pad_h: 0
pad_w: 0
}
}
layer {
name: "conv13"
type: "Convolution"
bottom: "pool12"
top: "conv13"
convolution_param {
num_output: 256
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 1
pad_w: 1
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu14"
type: "ReLU"
bottom: "conv13"
top: "conv13"
}
layer {
name: "conv15"
type: "Convolution"
bottom: "conv13"
top: "conv15"
convolution_param {
num_output: 256
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 1
pad_w: 1
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu16"
type: "ReLU"
bottom: "conv15"
top: "conv15"
}
layer {
name: "conv17"
type: "Convolution"
bottom: "conv15"
top: "conv17"
convolution_param {
num_output: 256
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 0
pad_w: 0
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu18"
type: "ReLU"
bottom: "conv17"
top: "conv17"
}
layer {
name: "conv19"
type: "Convolution"
bottom: "conv17"
top: "conv19"
convolution_param {
num_output: 256
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 0
pad_w: 0
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu20"
type: "ReLU"
bottom: "conv19"
top: "conv19"
}
layer {
name: "pool21"
type: "Pooling"
bottom: "conv19"
top: "pool21"
pooling_param {
pool: MAX
kernel_h: 3
kernel_w: 3
stride_h: 3
stride_w: 3
pad_h: 0
pad_w: 0
}
}
layer {
name: "conv22"
type: "Convolution"
bottom: "pool21"
top: "conv22"
convolution_param {
num_output: 2304
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 0
pad_w: 0
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu23"
type: "ReLU"
bottom: "conv22"
top: "conv22"
}
layer {
name: "conv24"
type: "Convolution"
bottom: "conv22"
top: "conv24"
convolution_param {
num_output: 256
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
pad_h: 1
pad_w: 1
kernel_h: 2
kernel_w: 2
stride_h: 1
stride_w: 1
}
}
layer {
name: "relu25"
type: "ReLU"
bottom: "conv24"
top: "conv24"
}
layer {
name: "fc26"
type: "InnerProduct"
bottom: "conv24"
top: "fc26"
inner_product_param {
num_output: 4096
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
name: "relu27"
type: "ReLU"
bottom: "fc26"
top: "fc26"
}
layer {
name: "fc28"
type: "InnerProduct"
bottom: "fc26"
top: "fc28"
inner_product_param {
num_output: 4096
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
name: "relu29"
type: "ReLU"
bottom: "fc28"
top: "fc28"
}
layer {
name: "output"
type: "InnerProduct"
bottom: "fc28"
top: "output"
inner_product_param {
num_output: 1000
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
Link Copied
4 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@csarron We are still investigating the root cause but I have a work around for you in the mean time. Here are the steps to create a conf file that should be placed in the same directory as your prototxt file.
1) Create a new text file named "j.conf"
2) Add the following lines to the "j.conf" text file:
conv15
im2col_v2
conv24
im2col_v2
3) Save the file and place it in the same directory as your "j.prototxt". You should be able to use this network as long as the conf file accompanies it.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@Tome_at_Intel Thank you for the reply. I tried the conf workaround as you said, still gave the same error, I also got the following log:
/usr/local/bin/ncsdk/Controllers/FileIO.py:52: UserWarning: [93mYou are using a large type. Consider reducing your data sizes for best performance[0m
"Consider reducing your data sizes for best performance\033[0m")
[91m[Error 25] Myriad Error: "Major or Minor Slices of MatMul are zero".[0m
[1mmvNCProfile v02.00, Copyright @ Movidius Ltd 2016[0m
0 0x80000000
Layer conv1 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer pool3 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer conv4 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer conv6 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer conv8 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer conv10 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer pool12 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer conv13 use the generic optimisations which is: 0x80000000
Spec opt found opt_conv_im2col_v2 1<< 2
Layer (a) conv15 use the optimisation mask which is: 0x4
0 0x80000000
Layer conv17 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer conv19 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer pool21 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer conv22 use the generic optimisations which is: 0x80000000
Spec opt found opt_conv_im2col_v2 1<< 2
4 0x80000004
Layer conv24 use the generic optimisations which is: 0x80000004
0 0x80000000
Layer fc26 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer fc28 use the generic optimisations which is: 0x80000000
0 0x80000000
Layer output use the generic optimisations which is: 0x80000000
USB: Transferring Data...
[Error 25] Myriad Error: "Major or Minor Slices of MatMul are zero".
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@Tome_at_Intel , given the optimization configuration you mentioned, can you share more about how to configure the optimization mask if possible?
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
@csarron Make sure you are using the latest version of the SDK and please try this again. Thanks.
Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page