Torch caffe binding
190 views
Skip to first unread message
srikanth sridhar
Jan 5, 2016, 1:58:06 PM1/5/16
to torch7
Hello all,
Recently to try the "Fully convolutional network for semantic segmentation" I have installed torch-caffe-binding. from here https://github.com/szagoruyko/torch-caffe-binding
Since this segmentation requires "caffe's future release version", i built the "future version" and the same future version is used to add torch-caffe-binding
Now to check the "torch-caffe-binding" i am trying the
Recently to try the "Fully convolutional network for semantic segmentation" I have installed torch-caffe-binding. from here https://github.com/szagoruyko/torch-caffe-binding
Since this segmentation requires "caffe's future release version", i built the "future version" and the same future version is used to add torch-caffe-binding
Now to check the "torch-caffe-binding" i am trying the
"bvlc_alexnet.caffemodel"
While building caffe's future release, i had some problems in 'make test',rest all passed
I continued with the compilation and when i tried require "caffe" th prompted a 'true'
Then I tried "bvlc_alexnet.caffemodel" as given in the torch-caffe-binding link,when the net:forward(input) is given,the itorch notebook is still running.The terminal displays error like,
th> output = net:forward(input)
F0106 00:23:46.470368 23604 caffe.cpp:43] Check failed: bottom->size[0]*bottom->size[1]*bottom->size[2]*bottom->size[3] == input_blobs[i]->count() (309174 vs. 1545870) MatCaffe input size does not match the input size of the network
*** Check failure stack trace: ***
I am not using MatCaffe and I cannot understand what is happening behind.
sorry for the long question.Any suggestions will be of great help
regards
srikanth
Sergey Zagoruyko
Jan 5, 2016, 2:57:16 PM1/5/16
to torch7 on behalf of srikanth sridhar
Looks like your input is not the same as caffe expects, try to make your input = torch.FloatTensor(10,3,227,227) instead of torch.FloatTensor(2,3,227,227).
Sergey.
--
You received this message because you are subscribed to the Google Groups "torch7" group.
To unsubscribe from this group and stop receiving emails from it, send an email to torch7+un...@googlegroups.com.
To post to this group, send email to tor...@googlegroups.com.
Visit this group at https://groups.google.com/group/torch7.
For more options, visit https://groups.google.com/d/optout.
srikanth sridhar
Jan 6, 2016, 12:03:06 AM1/6/16
to torch7
Hello sergey,
When I tried,
input = torch.FloatTensor(10,3,227,227) it shows the following output in terminal and the itorch notebook is still running
[I 10:26:06.317 NotebookApp] Adapting to protocol v4.0 for kernel c5bd5dd7-28ec-4824-adf3-be0e7f4cfec9
WARNING: comm_open not handled yet
WARNING: Logging before InitGoogleLogging() is written to STDERR
I0106 10:26:16.692391 3275 net.cpp:50] Initializing net from parameters:
name: "AlexNet"
input: "data"
state {
phase: TEST
}
input_shape {
dim: 10
dim: 3
dim: 227
dim: 227
}
layer {
name: "conv1"
type: "Convolution"
bottom: "data"
top: "conv1"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 96
kernel_size: 11
stride: 4
}
}
layer {
name: "relu1"
type: "ReLU"
bottom: "conv1"
top: "conv1"
}
layer {
name: "norm1"
type: "LRN"
bottom: "conv1"
top: "norm1"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "pool1"
type: "Pooling"
bottom: "norm1"
top: "pool1"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "conv2"
type: "Convolution"
bottom: "pool1"
top: "conv2"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 2
kernel_size: 5
group: 2
}
}
layer {
name: "relu2"
type: "ReLU"
bottom: "conv2"
top: "conv2"
}
layer {
name: "norm2"
type: "LRN"
bottom: "conv2"
top: "norm2"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "pool2"
type: "Pooling"
bottom: "norm2"
top: "pool2"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "conv3"
type: "Convolution"
bottom: "pool2"
top: "conv3"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
inner_product_param {
num_output: 1000
}
}
layer {
name: "prob"
type: "Softmax"
bottom: "fc8"
top: "prob"
}
I0106 10:26:16.693889 3275 net.cpp:436] Input 0 -> data
I0106 10:26:16.714329 3275 layer_factory.hpp:76] Creating layer conv1
I0106 10:26:16.714395 3275 net.cpp:111] Creating Layer conv1
I0106 10:26:16.714411 3275 net.cpp:478] conv1 <- data
I0106 10:26:16.714431 3275 net.cpp:434] conv1 -> conv1
I0106 10:26:16.720396 3275 net.cpp:156] Setting up conv1
I0106 10:26:16.720448 3275 net.cpp:164] Top shape: 10 96 55 55 (2904000)
I0106 10:26:16.720484 3275 layer_factory.hpp:76] Creating layer relu1
I0106 10:26:16.720499 3275 net.cpp:111] Creating Layer relu1
I0106 10:26:16.720507 3275 net.cpp:478] relu1 <- conv1
I0106 10:26:16.720517 3275 net.cpp:420] relu1 -> conv1 (in-place)
I0106 10:26:16.723407 3275 net.cpp:156] Setting up relu1
I0106 10:26:16.723448 3275 net.cpp:164] Top shape: 10 96 55 55 (2904000)
I0106 10:26:16.723462 3275 layer_factory.hpp:76] Creating layer norm1
I0106 10:26:16.723484 3275 net.cpp:111] Creating Layer norm1
I0106 10:26:16.723495 3275 net.cpp:478] norm1 <- conv1
I0106 10:26:16.723508 3275 net.cpp:434] norm1 -> norm1
I0106 10:26:16.723534 3275 net.cpp:156] Setting up norm1
I0106 10:26:16.723551 3275 net.cpp:164] Top shape: 10 96 55 55 (2904000)
I0106 10:26:16.723561 3275 layer_factory.hpp:76] Creating layer pool1
I0106 10:26:16.723572 3275 net.cpp:111] Creating Layer pool1
I0106 10:26:16.723587 3275 net.cpp:478] pool1 <- norm1
I0106 10:26:16.723597 3275 net.cpp:434] pool1 -> pool1
I0106 10:26:16.723628 3275 net.cpp:156] Setting up pool1
I0106 10:26:16.723639 3275 net.cpp:164] Top shape: 10 96 27 27 (699840)
I0106 10:26:16.723649 3275 layer_factory.hpp:76] Creating layer conv2
I0106 10:26:16.723662 3275 net.cpp:111] Creating Layer conv2
I0106 10:26:16.723671 3275 net.cpp:478] conv2 <- pool1
I0106 10:26:16.723681 3275 net.cpp:434] conv2 -> conv2
I0106 10:26:16.724416 3275 net.cpp:156] Setting up conv2
I0106 10:26:16.724467 3275 net.cpp:164] Top shape: 10 256 27 27 (1866240)
I0106 10:26:16.724491 3275 layer_factory.hpp:76] Creating layer relu2
I0106 10:26:16.724509 3275 net.cpp:111] Creating Layer relu2
I0106 10:26:16.724529 3275 net.cpp:478] relu2 <- conv2
I0106 10:26:16.724545 3275 net.cpp:420] relu2 -> conv2 (in-place)
I0106 10:26:16.724565 3275 net.cpp:156] Setting up relu2
I0106 10:26:16.724581 3275 net.cpp:164] Top shape: 10 256 27 27 (1866240)
I0106 10:26:16.724596 3275 layer_factory.hpp:76] Creating layer norm2
I0106 10:26:16.724609 3275 net.cpp:111] Creating Layer norm2
I0106 10:26:16.724617 3275 net.cpp:478] norm2 <- conv2
I0106 10:26:16.724627 3275 net.cpp:434] norm2 -> norm2
I0106 10:26:16.724642 3275 net.cpp:156] Setting up norm2
I0106 10:26:16.724652 3275 net.cpp:164] Top shape: 10 256 27 27 (1866240)
I0106 10:26:16.724659 3275 layer_factory.hpp:76] Creating layer pool2
I0106 10:26:16.724669 3275 net.cpp:111] Creating Layer pool2
I0106 10:26:16.724678 3275 net.cpp:478] pool2 <- norm2
I0106 10:26:16.724686 3275 net.cpp:434] pool2 -> pool2
I0106 10:26:16.724697 3275 net.cpp:156] Setting up pool2
I0106 10:26:16.724706 3275 net.cpp:164] Top shape: 10 256 13 13 (432640)
I0106 10:26:16.724714 3275 layer_factory.hpp:76] Creating layer conv3
I0106 10:26:16.724725 3275 net.cpp:111] Creating Layer conv3
I0106 10:26:16.724732 3275 net.cpp:478] conv3 <- pool2
I0106 10:26:16.724741 3275 net.cpp:434] conv3 -> conv3
I0106 10:26:16.726446 3275 net.cpp:156] Setting up conv3
I0106 10:26:16.726500 3275 net.cpp:164] Top shape: 10 384 13 13 (648960)
I0106 10:26:16.726517 3275 layer_factory.hpp:76] Creating layer relu3
I0106 10:26:16.726533 3275 net.cpp:111] Creating Layer relu3
I0106 10:26:16.726543 3275 net.cpp:478] relu3 <- conv3
I0106 10:26:16.726554 3275 net.cpp:420] relu3 -> conv3 (in-place)
I0106 10:26:16.726567 3275 net.cpp:156] Setting up relu3
I0106 10:26:16.726575 3275 net.cpp:164] Top shape: 10 384 13 13 (648960)
I0106 10:26:16.726584 3275 layer_factory.hpp:76] Creating layer conv4
I0106 10:26:16.726595 3275 net.cpp:111] Creating Layer conv4
I0106 10:26:16.726603 3275 net.cpp:478] conv4 <- conv3
I0106 10:26:16.726611 3275 net.cpp:434] conv4 -> conv4
I0106 10:26:16.727447 3275 net.cpp:156] Setting up conv4
I0106 10:26:16.727493 3275 net.cpp:164] Top shape: 10 384 13 13 (648960)
I0106 10:26:16.727506 3275 layer_factory.hpp:76] Creating layer relu4
I0106 10:26:16.727520 3275 net.cpp:111] Creating Layer relu4
I0106 10:26:16.727531 3275 net.cpp:478] relu4 <- conv4
I0106 10:26:16.727540 3275 net.cpp:420] relu4 -> conv4 (in-place)
I0106 10:26:16.727552 3275 net.cpp:156] Setting up relu4
I0106 10:26:16.727560 3275 net.cpp:164] Top shape: 10 384 13 13 (648960)
I0106 10:26:16.727568 3275 layer_factory.hpp:76] Creating layer conv5
I0106 10:26:16.727579 3275 net.cpp:111] Creating Layer conv5
I0106 10:26:16.727586 3275 net.cpp:478] conv5 <- conv4
I0106 10:26:16.727604 3275 net.cpp:434] conv5 -> conv5
I0106 10:26:16.728345 3275 net.cpp:156] Setting up conv5
I0106 10:26:16.728363 3275 net.cpp:164] Top shape: 10 256 13 13 (432640)
I0106 10:26:16.728376 3275 layer_factory.hpp:76] Creating layer relu5
I0106 10:26:16.728387 3275 net.cpp:111] Creating Layer relu5
I0106 10:26:16.728394 3275 net.cpp:478] relu5 <- conv5
I0106 10:26:16.728409 3275 net.cpp:420] relu5 -> conv5 (in-place)
I0106 10:26:16.728420 3275 net.cpp:156] Setting up relu5
I0106 10:26:16.728427 3275 net.cpp:164] Top shape: 10 256 13 13 (432640)
I0106 10:26:16.728435 3275 layer_factory.hpp:76] Creating layer pool5
I0106 10:26:16.728443 3275 net.cpp:111] Creating Layer pool5
I0106 10:26:16.728451 3275 net.cpp:478] pool5 <- conv5
I0106 10:26:16.728458 3275 net.cpp:434] pool5 -> pool5
I0106 10:26:16.728469 3275 net.cpp:156] Setting up pool5
I0106 10:26:16.728477 3275 net.cpp:164] Top shape: 10 256 6 6 (92160)
I0106 10:26:16.728484 3275 layer_factory.hpp:76] Creating layer fc6
I0106 10:26:16.728497 3275 net.cpp:111] Creating Layer fc6
I0106 10:26:16.728503 3275 net.cpp:478] fc6 <- pool5
I0106 10:26:16.728512 3275 net.cpp:434] fc6 -> fc6
I0106 10:26:16.775821 3275 net.cpp:156] Setting up fc6
I0106 10:26:16.775907 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.775940 3275 layer_factory.hpp:76] Creating layer relu6
I0106 10:26:16.776005 3275 net.cpp:111] Creating Layer relu6
I0106 10:26:16.776067 3275 net.cpp:478] relu6 <- fc6
I0106 10:26:16.776123 3275 net.cpp:420] relu6 -> fc6 (in-place)
I0106 10:26:16.776195 3275 net.cpp:156] Setting up relu6
I0106 10:26:16.776240 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.776301 3275 layer_factory.hpp:76] Creating layer drop6
I0106 10:26:16.776422 3275 net.cpp:111] Creating Layer drop6
I0106 10:26:16.776455 3275 net.cpp:478] drop6 <- fc6
I0106 10:26:16.776480 3275 net.cpp:420] drop6 -> fc6 (in-place)
I0106 10:26:16.776542 3275 net.cpp:156] Setting up drop6
I0106 10:26:16.776573 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.776593 3275 layer_factory.hpp:76] Creating layer fc7
I0106 10:26:16.776619 3275 net.cpp:111] Creating Layer fc7
I0106 10:26:16.776638 3275 net.cpp:478] fc7 <- fc6
I0106 10:26:16.776662 3275 net.cpp:434] fc7 -> fc7
I0106 10:26:16.798665 3275 net.cpp:156] Setting up fc7
I0106 10:26:16.798724 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.798741 3275 layer_factory.hpp:76] Creating layer relu7
I0106 10:26:16.798758 3275 net.cpp:111] Creating Layer relu7
I0106 10:26:16.798765 3275 net.cpp:478] relu7 <- fc7
I0106 10:26:16.798774 3275 net.cpp:420] relu7 -> fc7 (in-place)
I0106 10:26:16.798789 3275 net.cpp:156] Setting up relu7
I0106 10:26:16.798796 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.798804 3275 layer_factory.hpp:76] Creating layer drop7
I0106 10:26:16.798817 3275 net.cpp:111] Creating Layer drop7
I0106 10:26:16.798825 3275 net.cpp:478] drop7 <- fc7
I0106 10:26:16.798833 3275 net.cpp:420] drop7 -> fc7 (in-place)
I0106 10:26:16.798845 3275 net.cpp:156] Setting up drop7
I0106 10:26:16.798857 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.798879 3275 layer_factory.hpp:76] Creating layer fc8
I0106 10:26:16.798894 3275 net.cpp:111] Creating Layer fc8
I0106 10:26:16.798907 3275 net.cpp:478] fc8 <- fc7
I0106 10:26:16.798920 3275 net.cpp:434] fc8 -> fc8
I0106 10:26:16.804666 3275 net.cpp:156] Setting up fc8
I0106 10:26:16.804715 3275 net.cpp:164] Top shape: 10 1000 (10000)
I0106 10:26:16.804729 3275 layer_factory.hpp:76] Creating layer prob
I0106 10:26:16.804746 3275 net.cpp:111] Creating Layer prob
I0106 10:26:16.804757 3275 net.cpp:478] prob <- fc8
I0106 10:26:16.804769 3275 net.cpp:434] prob -> prob
I0106 10:26:16.804788 3275 net.cpp:156] Setting up prob
I0106 10:26:16.804801 3275 net.cpp:164] Top shape: 10 1000 (10000)
I0106 10:26:16.804810 3275 net.cpp:241] prob does not need backward computation.
I0106 10:26:16.804817 3275 net.cpp:241] fc8 does not need backward computation.
I0106 10:26:16.804824 3275 net.cpp:241] drop7 does not need backward computation.
I0106 10:26:16.804831 3275 net.cpp:241] relu7 does not need backward computation.
I0106 10:26:16.804838 3275 net.cpp:241] fc7 does not need backward computation.
I0106 10:26:16.804846 3275 net.cpp:241] drop6 does not need backward computation.
I0106 10:26:16.804853 3275 net.cpp:241] relu6 does not need backward computation.
I0106 10:26:16.804860 3275 net.cpp:241] fc6 does not need backward computation.
I0106 10:26:16.804867 3275 net.cpp:241] pool5 does not need backward computation.
I0106 10:26:16.804874 3275 net.cpp:241] relu5 does not need backward computation.
I0106 10:26:16.804882 3275 net.cpp:241] conv5 does not need backward computation.
I0106 10:26:16.804889 3275 net.cpp:241] relu4 does not need backward computation.
I0106 10:26:16.804896 3275 net.cpp:241] conv4 does not need backward computation.
I0106 10:26:16.804903 3275 net.cpp:241] relu3 does not need backward computation.
I0106 10:26:16.804910 3275 net.cpp:241] conv3 does not need backward computation.
I0106 10:26:16.804919 3275 net.cpp:241] pool2 does not need backward computation.
I0106 10:26:16.804927 3275 net.cpp:241] norm2 does not need backward computation.
I0106 10:26:16.804936 3275 net.cpp:241] relu2 does not need backward computation.
I0106 10:26:16.804944 3275 net.cpp:241] conv2 does not need backward computation.
I0106 10:26:16.804952 3275 net.cpp:241] pool1 does not need backward computation.
I0106 10:26:16.804960 3275 net.cpp:241] norm1 does not need backward computation.
I0106 10:26:16.804968 3275 net.cpp:241] relu1 does not need backward computation.
I0106 10:26:16.804976 3275 net.cpp:241] conv1 does not need backward computation.
I0106 10:26:16.804985 3275 net.cpp:284] This network produces output prob
I0106 10:26:16.805032 3275 net.cpp:298] Network initialization done.
I0106 10:26:16.805055 3275 net.cpp:299] Memory required for data: 77048960
I0106 10:26:19.392630 3275 upgrade_proto.cpp:611] Attempting to upgrade input file specified using deprecated transformation parameters: bvlc_alexnet.caffemodel
I0106 10:26:19.392684 3275 upgrade_proto.cpp:614] Successfully upgraded file specified using deprecated data transformation parameters.
W0106 10:26:19.392699 3275 upgrade_proto.cpp:616] Note that future Caffe releases will only support transform_param messages for transformation fields.
I0106 10:26:19.392709 3275 upgrade_proto.cpp:620] Attempting to upgrade input file specified using deprecated V1LayerParameter: bvlc_alexnet.caffemodel
I0106 10:26:19.571703 3275 upgrade_proto.cpp:628] Successfully upgraded file specified using deprecated V1LayerParameter
F0106 10:27:12.355908 3275 caffe.cpp:56] Unknown Caffe mode.
When I tried,
input = torch.FloatTensor(10,3,227,227) it shows the following output in terminal and the itorch notebook is still running
[I 10:26:06.317 NotebookApp] Adapting to protocol v4.0 for kernel c5bd5dd7-28ec-4824-adf3-be0e7f4cfec9
WARNING: comm_open not handled yet
WARNING: Logging before InitGoogleLogging() is written to STDERR
I0106 10:26:16.692391 3275 net.cpp:50] Initializing net from parameters:
name: "AlexNet"
input: "data"
state {
phase: TEST
}
input_shape {
dim: 10
dim: 3
dim: 227
dim: 227
}
layer {
name: "conv1"
type: "Convolution"
bottom: "data"
top: "conv1"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 96
kernel_size: 11
stride: 4
}
}
layer {
name: "relu1"
type: "ReLU"
bottom: "conv1"
top: "conv1"
}
layer {
name: "norm1"
type: "LRN"
bottom: "conv1"
top: "norm1"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "pool1"
type: "Pooling"
bottom: "norm1"
top: "pool1"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "conv2"
type: "Convolution"
bottom: "pool1"
top: "conv2"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 2
kernel_size: 5
group: 2
}
}
layer {
name: "relu2"
type: "ReLU"
bottom: "conv2"
top: "conv2"
}
layer {
name: "norm2"
type: "LRN"
bottom: "conv2"
top: "norm2"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "pool2"
type: "Pooling"
bottom: "norm2"
top: "pool2"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "conv3"
type: "Convolution"
bottom: "pool2"
top: "conv3"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
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"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
inner_product_param {
num_output: 1000
}
}
layer {
name: "prob"
type: "Softmax"
bottom: "fc8"
top: "prob"
}
I0106 10:26:16.693889 3275 net.cpp:436] Input 0 -> data
I0106 10:26:16.714329 3275 layer_factory.hpp:76] Creating layer conv1
I0106 10:26:16.714395 3275 net.cpp:111] Creating Layer conv1
I0106 10:26:16.714411 3275 net.cpp:478] conv1 <- data
I0106 10:26:16.714431 3275 net.cpp:434] conv1 -> conv1
I0106 10:26:16.720396 3275 net.cpp:156] Setting up conv1
I0106 10:26:16.720448 3275 net.cpp:164] Top shape: 10 96 55 55 (2904000)
I0106 10:26:16.720484 3275 layer_factory.hpp:76] Creating layer relu1
I0106 10:26:16.720499 3275 net.cpp:111] Creating Layer relu1
I0106 10:26:16.720507 3275 net.cpp:478] relu1 <- conv1
I0106 10:26:16.720517 3275 net.cpp:420] relu1 -> conv1 (in-place)
I0106 10:26:16.723407 3275 net.cpp:156] Setting up relu1
I0106 10:26:16.723448 3275 net.cpp:164] Top shape: 10 96 55 55 (2904000)
I0106 10:26:16.723462 3275 layer_factory.hpp:76] Creating layer norm1
I0106 10:26:16.723484 3275 net.cpp:111] Creating Layer norm1
I0106 10:26:16.723495 3275 net.cpp:478] norm1 <- conv1
I0106 10:26:16.723508 3275 net.cpp:434] norm1 -> norm1
I0106 10:26:16.723534 3275 net.cpp:156] Setting up norm1
I0106 10:26:16.723551 3275 net.cpp:164] Top shape: 10 96 55 55 (2904000)
I0106 10:26:16.723561 3275 layer_factory.hpp:76] Creating layer pool1
I0106 10:26:16.723572 3275 net.cpp:111] Creating Layer pool1
I0106 10:26:16.723587 3275 net.cpp:478] pool1 <- norm1
I0106 10:26:16.723597 3275 net.cpp:434] pool1 -> pool1
I0106 10:26:16.723628 3275 net.cpp:156] Setting up pool1
I0106 10:26:16.723639 3275 net.cpp:164] Top shape: 10 96 27 27 (699840)
I0106 10:26:16.723649 3275 layer_factory.hpp:76] Creating layer conv2
I0106 10:26:16.723662 3275 net.cpp:111] Creating Layer conv2
I0106 10:26:16.723671 3275 net.cpp:478] conv2 <- pool1
I0106 10:26:16.723681 3275 net.cpp:434] conv2 -> conv2
I0106 10:26:16.724416 3275 net.cpp:156] Setting up conv2
I0106 10:26:16.724467 3275 net.cpp:164] Top shape: 10 256 27 27 (1866240)
I0106 10:26:16.724491 3275 layer_factory.hpp:76] Creating layer relu2
I0106 10:26:16.724509 3275 net.cpp:111] Creating Layer relu2
I0106 10:26:16.724529 3275 net.cpp:478] relu2 <- conv2
I0106 10:26:16.724545 3275 net.cpp:420] relu2 -> conv2 (in-place)
I0106 10:26:16.724565 3275 net.cpp:156] Setting up relu2
I0106 10:26:16.724581 3275 net.cpp:164] Top shape: 10 256 27 27 (1866240)
I0106 10:26:16.724596 3275 layer_factory.hpp:76] Creating layer norm2
I0106 10:26:16.724609 3275 net.cpp:111] Creating Layer norm2
I0106 10:26:16.724617 3275 net.cpp:478] norm2 <- conv2
I0106 10:26:16.724627 3275 net.cpp:434] norm2 -> norm2
I0106 10:26:16.724642 3275 net.cpp:156] Setting up norm2
I0106 10:26:16.724652 3275 net.cpp:164] Top shape: 10 256 27 27 (1866240)
I0106 10:26:16.724659 3275 layer_factory.hpp:76] Creating layer pool2
I0106 10:26:16.724669 3275 net.cpp:111] Creating Layer pool2
I0106 10:26:16.724678 3275 net.cpp:478] pool2 <- norm2
I0106 10:26:16.724686 3275 net.cpp:434] pool2 -> pool2
I0106 10:26:16.724697 3275 net.cpp:156] Setting up pool2
I0106 10:26:16.724706 3275 net.cpp:164] Top shape: 10 256 13 13 (432640)
I0106 10:26:16.724714 3275 layer_factory.hpp:76] Creating layer conv3
I0106 10:26:16.724725 3275 net.cpp:111] Creating Layer conv3
I0106 10:26:16.724732 3275 net.cpp:478] conv3 <- pool2
I0106 10:26:16.724741 3275 net.cpp:434] conv3 -> conv3
I0106 10:26:16.726446 3275 net.cpp:156] Setting up conv3
I0106 10:26:16.726500 3275 net.cpp:164] Top shape: 10 384 13 13 (648960)
I0106 10:26:16.726517 3275 layer_factory.hpp:76] Creating layer relu3
I0106 10:26:16.726533 3275 net.cpp:111] Creating Layer relu3
I0106 10:26:16.726543 3275 net.cpp:478] relu3 <- conv3
I0106 10:26:16.726554 3275 net.cpp:420] relu3 -> conv3 (in-place)
I0106 10:26:16.726567 3275 net.cpp:156] Setting up relu3
I0106 10:26:16.726575 3275 net.cpp:164] Top shape: 10 384 13 13 (648960)
I0106 10:26:16.726584 3275 layer_factory.hpp:76] Creating layer conv4
I0106 10:26:16.726595 3275 net.cpp:111] Creating Layer conv4
I0106 10:26:16.726603 3275 net.cpp:478] conv4 <- conv3
I0106 10:26:16.726611 3275 net.cpp:434] conv4 -> conv4
I0106 10:26:16.727447 3275 net.cpp:156] Setting up conv4
I0106 10:26:16.727493 3275 net.cpp:164] Top shape: 10 384 13 13 (648960)
I0106 10:26:16.727506 3275 layer_factory.hpp:76] Creating layer relu4
I0106 10:26:16.727520 3275 net.cpp:111] Creating Layer relu4
I0106 10:26:16.727531 3275 net.cpp:478] relu4 <- conv4
I0106 10:26:16.727540 3275 net.cpp:420] relu4 -> conv4 (in-place)
I0106 10:26:16.727552 3275 net.cpp:156] Setting up relu4
I0106 10:26:16.727560 3275 net.cpp:164] Top shape: 10 384 13 13 (648960)
I0106 10:26:16.727568 3275 layer_factory.hpp:76] Creating layer conv5
I0106 10:26:16.727579 3275 net.cpp:111] Creating Layer conv5
I0106 10:26:16.727586 3275 net.cpp:478] conv5 <- conv4
I0106 10:26:16.727604 3275 net.cpp:434] conv5 -> conv5
I0106 10:26:16.728345 3275 net.cpp:156] Setting up conv5
I0106 10:26:16.728363 3275 net.cpp:164] Top shape: 10 256 13 13 (432640)
I0106 10:26:16.728376 3275 layer_factory.hpp:76] Creating layer relu5
I0106 10:26:16.728387 3275 net.cpp:111] Creating Layer relu5
I0106 10:26:16.728394 3275 net.cpp:478] relu5 <- conv5
I0106 10:26:16.728409 3275 net.cpp:420] relu5 -> conv5 (in-place)
I0106 10:26:16.728420 3275 net.cpp:156] Setting up relu5
I0106 10:26:16.728427 3275 net.cpp:164] Top shape: 10 256 13 13 (432640)
I0106 10:26:16.728435 3275 layer_factory.hpp:76] Creating layer pool5
I0106 10:26:16.728443 3275 net.cpp:111] Creating Layer pool5
I0106 10:26:16.728451 3275 net.cpp:478] pool5 <- conv5
I0106 10:26:16.728458 3275 net.cpp:434] pool5 -> pool5
I0106 10:26:16.728469 3275 net.cpp:156] Setting up pool5
I0106 10:26:16.728477 3275 net.cpp:164] Top shape: 10 256 6 6 (92160)
I0106 10:26:16.728484 3275 layer_factory.hpp:76] Creating layer fc6
I0106 10:26:16.728497 3275 net.cpp:111] Creating Layer fc6
I0106 10:26:16.728503 3275 net.cpp:478] fc6 <- pool5
I0106 10:26:16.728512 3275 net.cpp:434] fc6 -> fc6
I0106 10:26:16.775821 3275 net.cpp:156] Setting up fc6
I0106 10:26:16.775907 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.775940 3275 layer_factory.hpp:76] Creating layer relu6
I0106 10:26:16.776005 3275 net.cpp:111] Creating Layer relu6
I0106 10:26:16.776067 3275 net.cpp:478] relu6 <- fc6
I0106 10:26:16.776123 3275 net.cpp:420] relu6 -> fc6 (in-place)
I0106 10:26:16.776195 3275 net.cpp:156] Setting up relu6
I0106 10:26:16.776240 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.776301 3275 layer_factory.hpp:76] Creating layer drop6
I0106 10:26:16.776422 3275 net.cpp:111] Creating Layer drop6
I0106 10:26:16.776455 3275 net.cpp:478] drop6 <- fc6
I0106 10:26:16.776480 3275 net.cpp:420] drop6 -> fc6 (in-place)
I0106 10:26:16.776542 3275 net.cpp:156] Setting up drop6
I0106 10:26:16.776573 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.776593 3275 layer_factory.hpp:76] Creating layer fc7
I0106 10:26:16.776619 3275 net.cpp:111] Creating Layer fc7
I0106 10:26:16.776638 3275 net.cpp:478] fc7 <- fc6
I0106 10:26:16.776662 3275 net.cpp:434] fc7 -> fc7
I0106 10:26:16.798665 3275 net.cpp:156] Setting up fc7
I0106 10:26:16.798724 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.798741 3275 layer_factory.hpp:76] Creating layer relu7
I0106 10:26:16.798758 3275 net.cpp:111] Creating Layer relu7
I0106 10:26:16.798765 3275 net.cpp:478] relu7 <- fc7
I0106 10:26:16.798774 3275 net.cpp:420] relu7 -> fc7 (in-place)
I0106 10:26:16.798789 3275 net.cpp:156] Setting up relu7
I0106 10:26:16.798796 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.798804 3275 layer_factory.hpp:76] Creating layer drop7
I0106 10:26:16.798817 3275 net.cpp:111] Creating Layer drop7
I0106 10:26:16.798825 3275 net.cpp:478] drop7 <- fc7
I0106 10:26:16.798833 3275 net.cpp:420] drop7 -> fc7 (in-place)
I0106 10:26:16.798845 3275 net.cpp:156] Setting up drop7
I0106 10:26:16.798857 3275 net.cpp:164] Top shape: 10 4096 (40960)
I0106 10:26:16.798879 3275 layer_factory.hpp:76] Creating layer fc8
I0106 10:26:16.798894 3275 net.cpp:111] Creating Layer fc8
I0106 10:26:16.798907 3275 net.cpp:478] fc8 <- fc7
I0106 10:26:16.798920 3275 net.cpp:434] fc8 -> fc8
I0106 10:26:16.804666 3275 net.cpp:156] Setting up fc8
I0106 10:26:16.804715 3275 net.cpp:164] Top shape: 10 1000 (10000)
I0106 10:26:16.804729 3275 layer_factory.hpp:76] Creating layer prob
I0106 10:26:16.804746 3275 net.cpp:111] Creating Layer prob
I0106 10:26:16.804757 3275 net.cpp:478] prob <- fc8
I0106 10:26:16.804769 3275 net.cpp:434] prob -> prob
I0106 10:26:16.804788 3275 net.cpp:156] Setting up prob
I0106 10:26:16.804801 3275 net.cpp:164] Top shape: 10 1000 (10000)
I0106 10:26:16.804810 3275 net.cpp:241] prob does not need backward computation.
I0106 10:26:16.804817 3275 net.cpp:241] fc8 does not need backward computation.
I0106 10:26:16.804824 3275 net.cpp:241] drop7 does not need backward computation.
I0106 10:26:16.804831 3275 net.cpp:241] relu7 does not need backward computation.
I0106 10:26:16.804838 3275 net.cpp:241] fc7 does not need backward computation.
I0106 10:26:16.804846 3275 net.cpp:241] drop6 does not need backward computation.
I0106 10:26:16.804853 3275 net.cpp:241] relu6 does not need backward computation.
I0106 10:26:16.804860 3275 net.cpp:241] fc6 does not need backward computation.
I0106 10:26:16.804867 3275 net.cpp:241] pool5 does not need backward computation.
I0106 10:26:16.804874 3275 net.cpp:241] relu5 does not need backward computation.
I0106 10:26:16.804882 3275 net.cpp:241] conv5 does not need backward computation.
I0106 10:26:16.804889 3275 net.cpp:241] relu4 does not need backward computation.
I0106 10:26:16.804896 3275 net.cpp:241] conv4 does not need backward computation.
I0106 10:26:16.804903 3275 net.cpp:241] relu3 does not need backward computation.
I0106 10:26:16.804910 3275 net.cpp:241] conv3 does not need backward computation.
I0106 10:26:16.804919 3275 net.cpp:241] pool2 does not need backward computation.
I0106 10:26:16.804927 3275 net.cpp:241] norm2 does not need backward computation.
I0106 10:26:16.804936 3275 net.cpp:241] relu2 does not need backward computation.
I0106 10:26:16.804944 3275 net.cpp:241] conv2 does not need backward computation.
I0106 10:26:16.804952 3275 net.cpp:241] pool1 does not need backward computation.
I0106 10:26:16.804960 3275 net.cpp:241] norm1 does not need backward computation.
I0106 10:26:16.804968 3275 net.cpp:241] relu1 does not need backward computation.
I0106 10:26:16.804976 3275 net.cpp:241] conv1 does not need backward computation.
I0106 10:26:16.804985 3275 net.cpp:284] This network produces output prob
I0106 10:26:16.805032 3275 net.cpp:298] Network initialization done.
I0106 10:26:16.805055 3275 net.cpp:299] Memory required for data: 77048960
I0106 10:26:19.392630 3275 upgrade_proto.cpp:611] Attempting to upgrade input file specified using deprecated transformation parameters: bvlc_alexnet.caffemodel
I0106 10:26:19.392684 3275 upgrade_proto.cpp:614] Successfully upgraded file specified using deprecated data transformation parameters.
W0106 10:26:19.392699 3275 upgrade_proto.cpp:616] Note that future Caffe releases will only support transform_param messages for transformation fields.
I0106 10:26:19.392709 3275 upgrade_proto.cpp:620] Attempting to upgrade input file specified using deprecated V1LayerParameter: bvlc_alexnet.caffemodel
I0106 10:26:19.571703 3275 upgrade_proto.cpp:628] Successfully upgraded file specified using deprecated V1LayerParameter
F0106 10:27:12.355908 3275 caffe.cpp:56] Unknown Caffe mode.
*** Check failure stack trace: ***
[I 10:28:05.956 NotebookApp] Saving file at /Untitled1.ipynb
Is it something to do with the way built caffe future version?
regards
srikanth
Is it something to do with the way built caffe future version?
regards
srikanth
On Wednesday, 6 January 2016 01:27:16 UTC+5:30, Sergey Zagoruyko wrote:
Looks like your input is not the same as caffe expects, try to make your input = torch.FloatTensor(10,3,227,227) instead of torch.FloatTensor(2,3,227,227).Sergey.
srikanth sridhar
Jan 6, 2016, 12:20:57 AM1/6/16
to torch7
Hello sergey,
After seeing the caffe.CPP file, I did net:setModeCPU()
Now its giving a 10x1000x1x1 output probabilities
Is it possible to use the GPU .I can run GPU in torch 7,but dint build caffe with gpu.I run GPU in torch 7 using cutorch,cunn not cudnn.Have some issues in installing cudnn.
So is there a way to run this using GPU without loadcaffe.Bcz i am having issues using loadcaffe for the "Fully convolutional semantic segmentation" which is what I have to experiment with.
Regards
After seeing the caffe.CPP file, I did net:setModeCPU()
Now its giving a 10x1000x1x1 output probabilities
Is it possible to use the GPU .I can run GPU in torch 7,but dint build caffe with gpu.I run GPU in torch 7 using cutorch,cunn not cudnn.Have some issues in installing cudnn.
So is there a way to run this using GPU without loadcaffe.Bcz i am having issues using loadcaffe for the "Fully convolutional semantic segmentation" which is what I have to experiment with.
Regards
Francisco Vitor Suzano Massa
Jan 6, 2016, 1:41:05 AM1/6/16
to torch7
you need to compile caffe with GPU support (not necessarily with cudnn support though) if you want to run it in GPU inside torch.
when it's done, simply do
net:setModeGPU()
In any case, in this caffe-binding, you always pass a FloatTensor as input to the network, even if it's on the GPU (the conversion will be done inside the network)
when it's done, simply do
net:setModeGPU()
In any case, in this caffe-binding, you always pass a FloatTensor as input to the network, even if it's on the GPU (the conversion will be done inside the network)
srikanth sridhar
Jan 6, 2016, 2:49:32 AM1/6/16
to torch7
Hello Francisco,
Thank you for replying
I built caffe by uncommenting the CPU_ONLY option in the "Makefile.config"
While running that the last but one line,TEST_GPUID := 0 is uncommented.Does this mean I have actually built caffe with gpu as well or should I comment CPU_ONLY and uncomment USE_cuDNN
regards
srikanth
Thank you for replying
I built caffe by uncommenting the CPU_ONLY option in the "Makefile.config"
While running that the last but one line,TEST_GPUID := 0 is uncommented.Does this mean I have actually built caffe with gpu as well or should I comment CPU_ONLY and uncomment USE_cuDNN
regards
srikanth
Francisco Vitor Suzano Massa
Jan 6, 2016, 5:55:22 AM1/6/16
to torch7
you need to comment the line containing
CPU_ONLY
https://github.com/BVLC/caffe/blob/master/Makefile.config.example#L7-L8
if you need to use cudnn, uncomment the line
USE_cuDNN
https://github.com/BVLC/caffe/blob/master/Makefile.config.example#L4-L5
CPU_ONLY
https://github.com/BVLC/caffe/blob/master/Makefile.config.example#L7-L8
if you need to use cudnn, uncomment the line
USE_cuDNN
https://github.com/BVLC/caffe/blob/master/Makefile.config.example#L4-L5
srikanth sridhar
Jan 6, 2016, 11:17:50 AM1/6/16
to torch7
Thank you Francisco.I built caffe in GPU by commenting CPU_ONLY.It works fine when i tried alexnet in gpu mode.When running 'make runtest' it passed and printed that two tests are disabled.
Great and thank you very much
Great and thank you very much
Reply all
Reply to author
Forward
0 new messages