Unknown caffe.NetParameter type IMAGE_SEG_DATA ?
1,926 views
Skip to first unread message
Ruud
Sep 25, 2015, 8:13:22 AM9/25/15
to Caffe Users
Hi all,
I am trying to run the example of DeepLab on top of Caffe. However, it appears the field type in my prototext is not recognized by Caffe!
[libprotobuf ERROR google/protobuf/text_format.cc:274] Error parsing text-format caffe.NetParameter: 24:3: Unknown enumeration value of "IMAGE_SEG_DATA" for field "type".
Any ideas how come?
Best,
Ruud
I am trying to run the example of DeepLab on top of Caffe. However, it appears the field type in my prototext is not recognized by Caffe!
[libprotobuf ERROR google/protobuf/text_format.cc:274] Error parsing text-format caffe.NetParameter: 24:3: Unknown enumeration value of "IMAGE_SEG_DATA" for field "type".
Any ideas how come?
Best,
Ruud
Ali Mousavi
Sep 26, 2015, 3:51:13 PM9/26/15
to Caffe Users
"You need to save the png annotations without the colormap; otherwise OpenCV will read the wrong values. See the script at matlab/my_script/SavePngAsRawPng.m for reference."
Ruud
Sep 27, 2015, 10:36:21 AM9/27/15
to Caffe Users
Hi Ali,
Thank you for your response. I also read that FAQ, but I didn't not link my problem to that in the first place; the error is a bit cryptic.
I will try to convert the annotations and come back here to report.
Best,
Ruud
Evan Shelhamer
Sep 27, 2015, 2:34:06 PM9/27/15
to Ruud, Caffe Users
You have to compile and run the DeepLab authors' fork of Caffe -- IMAGE_SEG_DATA is not a layer type in BVLC/caffe
--
You received this message because you are subscribed to the Google Groups "Caffe Users" group.
To unsubscribe from this group and stop receiving emails from it, send an email to caffe-users...@googlegroups.com.
To post to this group, send email to caffe...@googlegroups.com.
To view this discussion on the web visit https://groups.google.com/d/msgid/caffe-users/389b899f-4e55-4f2b-a23a-2401a7f918b8%40googlegroups.com.
For more options, visit https://groups.google.com/d/optout.
Ruud
Sep 28, 2015, 2:11:17 AM9/28/15
to Caffe Users, rud...@gmail.com
Hi Evan,
That makes sense! It was not apparent from the instructions, so I assumed it was built on top of the regular Caffe release.
Best,
Ruud
That makes sense! It was not apparent from the instructions, so I assumed it was built on top of the regular Caffe release.
Best,
Ruud
Message has been deleted
Dan Shulman
Jul 26, 2016, 8:47:26 AM7/26/16
to Caffe Users, rud...@gmail.com
Where can I find the DeepLab authors' fork of Caffe? I'm using https://bitbucket.org/aquariusjay/deeplab-public-ver2
Thanks,
Dan
Ruud
Jul 26, 2016, 8:51:03 AM7/26/16
to Caffe Users, rud...@gmail.com
Yes, that is the correct link.
Dan Shulman
Jul 26, 2016, 12:41:28 PM7/26/16
to Caffe Users, rud...@gmail.com
But this is the link to the deeplab project, it has no fork of Caffe, but uses the original one.
Dan Shulman
Jul 27, 2016, 5:36:29 AM7/27/16
to Caffe Users, rud...@gmail.com
ok, I compiled it, but still receive:
Unknown enumeration value of "IMAGE_SEG_DATA" for field "type"
What should I do?
Dan
Ruud
Jul 27, 2016, 1:21:25 PM7/27/16
to Caffe Users, rud...@gmail.com
Are you sure CAFFE_BIN=${CAFFE_DIR}/.build_release/tools/caffe.bin refers to the compiled Deeplab directory?
Dan Shulman
Jul 27, 2016, 1:32:34 PM7/27/16
to Caffe Users, rud...@gmail.com
yes, it is:
CAFFE_DIR=.
CAFFE_BIN=${CAFFE_DIR}/.build_release/tools/caffe.bin
where CAFFE_DIR (.) is the deeplab directory.
Ruud
Jul 28, 2016, 1:37:32 PM7/28/16
to Caffe Users, rud...@gmail.com
So you are still getting the error?
Which prototext are you using for training? With V2, you might be using an outdated Caffe prototext definition that was used in V1 but that is no longer supported without some adjustments.
Dan Shulman
Jul 29, 2016, 5:26:24 AM7/29/16
to Caffe Users, rud...@gmail.com
Yes, that was the problem, thanks!
Do you know how I can run and get results for my own images?
Tao Liu
Aug 6, 2016, 1:24:34 PM8/6/16
to Caffe Users, rud...@gmail.com
Did encounter the following error when your make runtest the caffe from the deeplab author
[ RUN ] SoftmaxWithLossLayerTest/2.TestGradientWeights
F0806 13:21:43.072187 3521 softmax_loss_layer.cpp:29] Check failed: infile.is_open()
*** Check failure stack trace: ***
@ 0x2b0d1998adaa (unknown)
@ 0x2b0d1998ace4 (unknown)
@ 0x2b0d1998a6e6 (unknown)
@ 0x2b0d1998d687 (unknown)
@ 0x82db08 caffe::SoftmaxWithLossLayer<>::LayerSetUp()
@ 0x4518c3 caffe::GradientChecker<>::CheckGradientExhaustive()
@ 0x6be271 caffe::SoftmaxWithLossLayerTest_TestGradientWeights_Test<>::TestBody()
@ 0x726a43 testing::internal::HandleExceptionsInMethodIfSupported<>()
@ 0x71d587 testing::Test::Run()
@ 0x71d62e testing::TestInfo::Run()
@ 0x71d735 testing::TestCase::Run()
@ 0x720a78 testing::internal::UnitTestImpl::RunAllTests()
@ 0x720d07 testing::UnitTest::Run()
@ 0x4248aa main
@ 0x2b0d1d036f45 (unknown)
@ 0x42eb67 (unknown)
@ (nil) (unknown)
Aborted (core dumped)
make: *** [runtest] Error 134
Message has been deleted
Alessandro Musumeci
Nov 14, 2016, 4:08:49 PM11/14/16
to Caffe Users, rud...@gmail.com
Hi Dan, I have right now the same problem that you had before: "Unknown
enumeration value of "IMAGE_SEG_DATA" for field "type" ". I'm using
deeplabV2, I changed the prototext file but now I have:
"Message type "caffe.ImageDataParameter" has no field named "label_type". " that sounds the same. Can you tell me please how did you solved the problem?
Thanks
Alessandro
"Message type "caffe.ImageDataParameter" has no field named "label_type". " that sounds the same. Can you tell me please how did you solved the problem?
Thanks
Alessandro
Alessandro Musumeci
Nov 14, 2016, 6:04:54 PM11/14/16
to Caffe Users
Hi, please can you share what have you done to solve this problem?
Thanks
Ruud
Nov 15, 2016, 10:43:14 AM11/15/16
to Caffe Users
Hi Allesandro,
I used the following prototext file for Deeplab V2, using the vanilla model:
# VGG 16-layer network convolutional finetuning
# Network modified to have smaller receptive field (128 pixels)
# and smaller stride (8 pixels) when run in convolutional mode.
#
# For alignment to work, we set:
# (1) input dimension equal to
# $n = 8 * k + 2$, e.g., 306 (for k = 38)
# (2) dimension after 3rd max-pooling (centered at -3.5)
# $m = k + 2$ (40 if k = 38)
# (3) dimension after 4th max-pooling (centered at -1.5)
# $m = k + 1$ (39 if k = 38)
# (4) Crop 1 pixels at the begin of label map and shrink by 8
# to produce the expected $m$
#
name: "${NET_ID}"
layer {
name: "data"
type: "ImageSegData"
top: "data"
top: "label"
image_data_param {
root_folder: "${DATA_ROOT}"
source: "${EXP}/list/${TRAIN_SET}.txt"
label_type: PIXEL
batch_size: 10
shuffle: true
}
transform_param {
# Use BGR as order!
# Use matlab script : calc_bgr_image_set_mean.m
mean_value: 34.7887
mean_value: 27.7252
mean_value: 38.9483
crop_size: 306
mirror: true
}
include: { phase: TRAIN }
}
### NETWORK ###
layer {
bottom: "data"
top: "conv1_1"
name: "conv1_1"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv1_1"
top: "conv1_1"
name: "relu1_1"
type: "ReLU"
}
layer {
bottom: "conv1_1"
top: "conv1_2"
name: "conv1_2"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 64
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv1_2"
top: "conv1_2"
name: "relu1_2"
type: "ReLU"
}
layer {
bottom: "conv1_2"
top: "pool1"
name: "pool1"
type: "Pooling"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
pad: 1
}
}
layer {
bottom: "pool1"
top: "conv2_1"
name: "conv2_1"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 128
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv2_1"
top: "conv2_1"
name: "relu2_1"
type: "ReLU"
}
layer {
bottom: "conv2_1"
top: "conv2_2"
name: "conv2_2"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 128
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv2_2"
top: "conv2_2"
name: "relu2_2"
type: "ReLU"
}
layer {
bottom: "conv2_2"
top: "pool2"
name: "pool2"
type: "Pooling"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
pad: 1
}
}
layer {
bottom: "pool2"
top: "conv3_1"
name: "conv3_1"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv3_1"
top: "conv3_1"
name: "relu3_1"
type: "ReLU"
}
layer {
bottom: "conv3_1"
top: "conv3_2"
name: "conv3_2"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv3_2"
top: "conv3_2"
name: "relu3_2"
type: "ReLU"
}
layer {
bottom: "conv3_2"
top: "conv3_3"
name: "conv3_3"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 256
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv3_3"
top: "conv3_3"
name: "relu3_3"
type: "ReLU"
}
layer {
bottom: "conv3_3"
top: "pool3"
name: "pool3"
type: "Pooling"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
pad: 1
}
}
layer {
bottom: "pool3"
top: "conv4_1"
name: "conv4_1"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv4_1"
top: "conv4_1"
name: "relu4_1"
type: "ReLU"
}
layer {
bottom: "conv4_1"
top: "conv4_2"
name: "conv4_2"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv4_2"
top: "conv4_2"
name: "relu4_2"
type: "ReLU"
}
layer {
bottom: "conv4_2"
top: "conv4_3"
name: "conv4_3"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv4_3"
top: "conv4_3"
name: "relu4_3"
type: "ReLU"
}
layer {
bottom: "conv4_3"
top: "pool4"
name: "pool4"
type: "Pooling"
pooling_param {
pool: MAX
kernel_size: 2
#pad: 1
#stride: 2
stride: 1
}
}
layer {
bottom: "pool4"
top: "conv5_1"
name: "conv5_1"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
#pad: 1
pad: 2
#hole is for V1, use 'dilation' instead
#hole: 2
dilation: 2
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv5_1"
top: "conv5_1"
name: "relu5_1"
type: "ReLU"
}
layer {
bottom: "conv5_1"
top: "conv5_2"
name: "conv5_2"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
#pad: 1
pad: 2
#hole is for V1, use 'dilation' instead for V2
#hole: 2
dilation: 2
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv5_2"
top: "conv5_2"
name: "relu5_2"
type: "ReLU"
}
layer {
bottom: "conv5_2"
top: "conv5_3"
name: "conv5_3"
type: "Convolution"
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 512
#pad: 1
pad: 2
#hole is for V1, use 'dilation' instead for V2
#hole: 2
dilation: 2
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "conv5_3"
top: "conv5_3"
name: "relu5_3"
type: "ReLU"
}
layer {
bottom: "conv5_3"
top: "pool5"
name: "pool5"
type: "Pooling"
pooling_param {
pool: MAX
#kernel_size: 2
#stride: 2
kernel_size: 3
stride: 1
pad: 1
}
}
layer {
bottom: "pool5"
top: "fc6"
name: "fc6"
type: "Convolution"
# Works in V1, does not in V2?
# strict_dim: false
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 4096
pad: 6
#hole is for V1, use 'dilation' instead for V2
#hole: 4
dilation: 4
kernel_size: 4
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "fc6"
top: "fc6"
name: "relu6"
type: "ReLU"
}
layer {
bottom: "fc6"
top: "fc6"
name: "drop6"
type: "Dropout"
dropout_param {
dropout_ratio: 0.5
}
}
layer {
bottom: "fc6"
top: "fc7"
name: "fc7"
type: "Convolution"
# This parameter seems deprecated in V2
# strict_dim: false
# These parameter do not seem to be parsed in V2
#blobs_lr: 1
#blobs_lr: 2
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 4096
kernel_size: 1
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
}
}
}
layer {
bottom: "fc7"
top: "fc7"
name: "relu7"
type: "ReLU"
}
layer {
bottom: "fc7"
top: "fc7"
name: "drop7"
type: "Dropout"
dropout_param {
dropout_ratio: 0.5
}
}
layer {
bottom: "fc7"
top: "fc8_synth_to_real"
name: "fc8_synth_to_real"
type: "Convolution"
# This parameter seems deprecated in V2
#strict_dim: false
# These parameter do not seem to be parsed in V2
#blobs_lr: 10
#blobs_lr: 20
#weight_decay: 1
#weight_decay: 0
# For V2 use these instead
param {
lr_mult: 10
decay_mult: 1
}
param {
lr_mult: 20
decay_mult: 0
}
convolution_param {
num_output: ${NUM_LABELS}
kernel_size: 1
weight_filler {
type: "gaussian"
std: 0.01
}
bias_filler {
type: "constant"
value: 0
}
}
}
layer {
bottom: "label"
top: "label_shrink"
name: "label_shrink"
type: "Interp"
interp_param {
shrink_factor: 8
pad_beg: -1
pad_end: 0
}
}
layer {
name: "loss"
type: "SoftmaxWithLoss"
bottom: "fc8_synth_to_real"
bottom: "label_shrink"
top: "loss"
# Use this instead for V2
loss_param {
ignore_label: 255
normalize : true
}
#include: { phase: TRAIN }
}
layer {
name: "accuracy"
type: "SegAccuracy"
bottom: "fc8_synth_to_real"
bottom: "label_shrink"
top: "accuracy"
seg_accuracy_param {
ignore_label: 255
}
}
# layer {
# name: "im_data"
# type: IMSHOW
# bottom: "data"
# }
# layer {
# name: "im_scores"
# type: IMSHOW
# bottom: "fc8_pascal"
# }
#layer {
# name: "fc8_mat"
# type: "MatWrite"
# bottom: "fc8_synth_to_real"
# mat_write_param {
# #prefix: "voc12/features/${NET_ID}/${TEST_SET}/fc8/"
# #source: "voc12/list/${TEST_SET}_id.txt"
# prefix: "${EXP}/features/${NET_ID}/${TEST_SET}/fc8/"
# source: "${EXP}/list/${TEST_SET}_id.txt"
# strip: 0
# period: 1
# }
# include: { phase: TEST }
#}
Tsaku Nelson
Jun 11, 2018, 10:55:52 AM6/11/18
to Caffe Users
Tsaku Nelson
Jun 15, 2018, 11:11:07 AM6/15/18
to Caffe Users
Hello Alessandro, how did you finally solve the problem? because I had the same error as shown on the screenshot below, applied the fix by modifying the train_train_aug.prototxt file, but it seem to re-appear, and I have the feeling that the file "train_train_aug.prototxt" is automatically created with the parameters like "IMAGE_SEG_DATA" inside already.
Is there a way to prevent the automatic creation, in order to use the correct version of the prototext you used? Feedback most appreciated
Tsaku Nelson
Jun 22, 2018, 11:35:35 AM6/22/18
to Caffe Users
Hi Alessandro, I have the same error as you, after solving the "IMAGE_SEG_DATA" issue:
"Message type "caffe.ImageDataParameter" has no field named "label_type". "
Can you tell me please how did you solved the problem? Thanks
Reply all
Reply to author
Forward
0 new messages