SegNet Standard Model Config Error. # SET SAMPLE SIZE HERE
bzdfzfer opened this issue · 4 comments
bzdfzfer commented
Hi @navganti ,
Thanks a lot for sharing your work! When I run your program, the basic model perfoms well for me but the standard bayesian segnet does not work. It truned to be an error of the model layer size mismatch. Do you meet this error in your experiment? Following is running information:
ORB-SLAM2 Copyright (C) 2014-2016 Raul Mur-Artal, University of Zaragoza.
This program comes with ABSOLUTELY NO WARRANTY;
This is free software, and you are welcome to redistribute it
under certain conditions. See LICENSE.txt.
Loading ORB Vocabulary. This could take a while...
Vocabulary loaded!
WARNING: Logging before InitGoogleLogging() is written to STDERR
I1029 20:53:35.092469 6127 upgrade_proto.cpp:67] Attempting to upgrade input file specified using deprecated input fields: config/bayesian_segnet/standard/kitti/bayesian_segnet_kitti.prototxt
I1029 20:53:35.092579 6127 upgrade_proto.cpp:70] Successfully upgraded file specified using deprecated input fields.
W1029 20:53:35.092587 6127 upgrade_proto.cpp:72] Note that future Caffe releases will only support input layers and not input fields.
I1029 20:53:35.093302 6127 net.cpp:58] Initializing net from parameters:
name: "bayesian_segnet"
state {
phase: TEST
level: 0
}
layer {
name: "input"
type: "Input"
top: "data"
input_param {
shape {
dim: 2
dim: 3
dim: 352
dim: 1024
}
}
}
layer {
name: "conv1_1"
type: "Convolution"
bottom: "data"
top: "conv1_1"
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"
value: 0
}
}
}
layer {
name: "conv1_1_bn"
type: "BN"
bottom: "conv1_1"
top: "conv1_1"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu1_1"
type: "ReLU"
bottom: "conv1_1"
top: "conv1_1"
}
layer {
name: "conv1_2"
type: "Convolution"
bottom: "conv1_1"
top: "conv1_2"
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"
value: 0
}
}
}
layer {
name: "conv1_2_bn"
type: "BN"
bottom: "conv1_2"
top: "conv1_2"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu1_2"
type: "ReLU"
bottom: "conv1_2"
top: "conv1_2"
}
layer {
name: "pool1"
type: "Pooling"
bottom: "conv1_2"
top: "pool1"
top: "pool1_mask"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layer {
name: "conv2_1"
type: "Convolution"
bottom: "pool1"
top: "conv2_1"
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"
value: 0
}
}
}
layer {
name: "conv2_1_bn"
type: "BN"
bottom: "conv2_1"
top: "conv2_1"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu2_1"
type: "ReLU"
bottom: "conv2_1"
top: "conv2_1"
}
layer {
name: "conv2_2"
type: "Convolution"
bottom: "conv2_1"
top: "conv2_2"
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"
value: 0
}
}
}
layer {
name: "conv2_2_bn"
type: "BN"
bottom: "conv2_2"
top: "conv2_2"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu2_2"
type: "ReLU"
bottom: "conv2_2"
top: "conv2_2"
}
layer {
name: "pool2"
type: "Pooling"
bottom: "conv2_2"
top: "pool2"
top: "pool2_mask"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layer {
name: "conv3_1"
type: "Convolution"
bottom: "pool2"
top: "conv3_1"
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"
value: 0
}
}
}
layer {
name: "conv3_1_bn"
type: "BN"
bottom: "conv3_1"
top: "conv3_1"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu3_1"
type: "ReLU"
bottom: "conv3_1"
top: "conv3_1"
}
layer {
name: "conv3_2"
type: "Convolution"
bottom: "conv3_1"
top: "conv3_2"
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"
value: 0
}
}
}
layer {
name: "conv3_2_bn"
type: "BN"
bottom: "conv3_2"
top: "conv3_2"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu3_2"
type: "ReLU"
bottom: "conv3_2"
top: "conv3_2"
}
layer {
name: "conv3_3"
type: "Convolution"
bottom: "conv3_2"
top: "conv3_3"
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"
value: 0
}
}
}
layer {
name: "conv3_3_bn"
type: "BN"
bottom: "conv3_3"
top: "conv3_3"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu3_3"
type: "ReLU"
bottom: "conv3_3"
top: "conv3_3"
}
layer {
name: "pool3"
type: "Pooling"
bottom: "conv3_3"
top: "pool3"
top: "pool3_mask"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layer {
name: "pool3_drop"
type: "Dropout"
bottom: "pool3"
top: "pool3"
dropout_param {
dropout_ratio: 0.5
sample_weights_test: true
}
}
layer {
name: "conv4_1"
type: "Convolution"
bottom: "pool3"
top: "conv4_1"
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"
value: 0
}
}
}
layer {
name: "conv4_1_bn"
type: "BN"
bottom: "conv4_1"
top: "conv4_1"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu4_1"
type: "ReLU"
bottom: "conv4_1"
top: "conv4_1"
}
layer {
name: "conv4_2"
type: "Convolution"
bottom: "conv4_1"
top: "conv4_2"
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"
value: 0
}
}
}
layer {
name: "conv4_2_bn"
type: "BN"
bottom: "conv4_2"
top: "conv4_2"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu4_2"
type: "ReLU"
bottom: "conv4_2"
top: "conv4_2"
}
layer {
name: "conv4_3"
type: "Convolution"
bottom: "conv4_2"
top: "conv4_3"
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"
value: 0
}
}
}
layer {
name: "conv4_3_bn"
type: "BN"
bottom: "conv4_3"
top: "conv4_3"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu4_3"
type: "ReLU"
bottom: "conv4_3"
top: "conv4_3"
}
layer {
name: "pool4"
type: "Pooling"
bottom: "conv4_3"
top: "pool4"
top: "pool4_mask"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layer {
name: "pool4_drop"
type: "Dropout"
bottom: "pool4"
top: "pool4"
dropout_param {
dropout_ratio: 0.5
sample_weights_test: true
}
}
layer {
name: "conv5_1"
type: "Convolution"
bottom: "pool4"
top: "conv5_1"
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"
value: 0
}
}
}
layer {
name: "conv5_1_bn"
type: "BN"
bottom: "conv5_1"
top: "conv5_1"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu5_1"
type: "ReLU"
bottom: "conv5_1"
top: "conv5_1"
}
layer {
name: "conv5_2"
type: "Convolution"
bottom: "conv5_1"
top: "conv5_2"
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"
value: 0
}
}
}
layer {
name: "conv5_2_bn"
type: "BN"
bottom: "conv5_2"
top: "conv5_2"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu5_2"
type: "ReLU"
bottom: "conv5_2"
top: "conv5_2"
}
layer {
name: "conv5_3"
type: "Convolution"
bottom: "conv5_2"
top: "conv5_3"
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"
value: 0
}
}
}
layer {
name: "conv5_3_bn"
type: "BN"
bottom: "conv5_3"
top: "conv5_3"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu5_3"
type: "ReLU"
bottom: "conv5_3"
top: "conv5_3"
}
layer {
name: "pool5"
type: "Pooling"
bottom: "conv5_3"
top: "pool5"
top: "pool5_mask"
pooling_param {
pool: MAX
kernel_size: 2
stride: 2
}
}
layer {
name: "pool5_drop"
type: "Dropout"
bottom: "pool5"
top: "pool5"
dropout_param {
dropout_ratio: 0.5
sample_weights_test: true
}
}
layer {
name: "upsample5"
type: "Upsample"
bottom: "pool5"
bottom: "pool5_mask"
top: "pool5_D"
upsample_param {
scale: 2
upsample_h: 23
upsample_w: 30
}
}
layer {
name: "conv5_3_D"
type: "Convolution"
bottom: "pool5_D"
top: "conv5_3_D"
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"
value: 0
}
}
}
layer {
name: "conv5_3_D_bn"
type: "BN"
bottom: "conv5_3_D"
top: "conv5_3_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu5_3_D"
type: "ReLU"
bottom: "conv5_3_D"
top: "conv5_3_D"
}
layer {
name: "conv5_2_D"
type: "Convolution"
bottom: "conv5_3_D"
top: "conv5_2_D"
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"
value: 0
}
}
}
layer {
name: "conv5_2_D_bn"
type: "BN"
bottom: "conv5_2_D"
top: "conv5_2_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu5_2_D"
type: "ReLU"
bottom: "conv5_2_D"
top: "conv5_2_D"
}
layer {
name: "conv5_1_D"
type: "Convolution"
bottom: "conv5_2_D"
top: "conv5_1_D"
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"
value: 0
}
}
}
layer {
name: "conv5_1_D_bn"
type: "BN"
bottom: "conv5_1_D"
top: "conv5_1_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu5_1_D"
type: "ReLU"
bottom: "conv5_1_D"
top: "conv5_1_D"
}
layer {
name: "upsample4_drop"
type: "Dropout"
bottom: "conv5_1_D"
top: "conv5_1_D"
dropout_param {
dropout_ratio: 0.5
sample_weights_test: true
}
}
layer {
name: "upsample4"
type: "Upsample"
bottom: "conv5_1_D"
bottom: "pool4_mask"
top: "pool4_D"
upsample_param {
scale: 2
upsample_h: 45
upsample_w: 60
}
}
layer {
name: "conv4_3_D"
type: "Convolution"
bottom: "pool4_D"
top: "conv4_3_D"
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"
value: 0
}
}
}
layer {
name: "conv4_3_D_bn"
type: "BN"
bottom: "conv4_3_D"
top: "conv4_3_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu4_3_D"
type: "ReLU"
bottom: "conv4_3_D"
top: "conv4_3_D"
}
layer {
name: "conv4_2_D"
type: "Convolution"
bottom: "conv4_3_D"
top: "conv4_2_D"
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"
value: 0
}
}
}
layer {
name: "conv4_2_D_bn"
type: "BN"
bottom: "conv4_2_D"
top: "conv4_2_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu4_2_D"
type: "ReLU"
bottom: "conv4_2_D"
top: "conv4_2_D"
}
layer {
name: "conv4_1_D"
type: "Convolution"
bottom: "conv4_2_D"
top: "conv4_1_D"
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"
value: 0
}
}
}
layer {
name: "conv4_1_D_bn"
type: "BN"
bottom: "conv4_1_D"
top: "conv4_1_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu4_1_D"
type: "ReLU"
bottom: "conv4_1_D"
top: "conv4_1_D"
}
layer {
name: "upsample3_drop"
type: "Dropout"
bottom: "conv4_1_D"
top: "conv4_1_D"
dropout_param {
dropout_ratio: 0.5
sample_weights_test: true
}
}
layer {
name: "upsample3"
type: "Upsample"
bottom: "conv4_1_D"
bottom: "pool3_mask"
top: "pool3_D"
upsample_param {
scale: 2
}
}
layer {
name: "conv3_3_D"
type: "Convolution"
bottom: "pool3_D"
top: "conv3_3_D"
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"
value: 0
}
}
}
layer {
name: "conv3_3_D_bn"
type: "BN"
bottom: "conv3_3_D"
top: "conv3_3_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu3_3_D"
type: "ReLU"
bottom: "conv3_3_D"
top: "conv3_3_D"
}
layer {
name: "conv3_2_D"
type: "Convolution"
bottom: "conv3_3_D"
top: "conv3_2_D"
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"
value: 0
}
}
}
layer {
name: "conv3_2_D_bn"
type: "BN"
bottom: "conv3_2_D"
top: "conv3_2_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu3_2_D"
type: "ReLU"
bottom: "conv3_2_D"
top: "conv3_2_D"
}
layer {
name: "conv3_1_D"
type: "Convolution"
bottom: "conv3_2_D"
top: "conv3_1_D"
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"
value: 0
}
}
}
layer {
name: "conv3_1_D_bn"
type: "BN"
bottom: "conv3_1_D"
top: "conv3_1_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu3_1_D"
type: "ReLU"
bottom: "conv3_1_D"
top: "conv3_1_D"
}
layer {
name: "upsample2_drop"
type: "Dropout"
bottom: "conv3_1_D"
top: "conv3_1_D"
dropout_param {
dropout_ratio: 0.5
sample_weights_test: true
}
}
layer {
name: "upsample2"
type: "Upsample"
bottom: "conv3_1_D"
bottom: "pool2_mask"
top: "pool2_D"
upsample_param {
scale: 2
}
}
layer {
name: "conv2_2_D"
type: "Convolution"
bottom: "pool2_D"
top: "conv2_2_D"
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"
value: 0
}
}
}
layer {
name: "conv2_2_D_bn"
type: "BN"
bottom: "conv2_2_D"
top: "conv2_2_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu2_2_D"
type: "ReLU"
bottom: "conv2_2_D"
top: "conv2_2_D"
}
layer {
name: "conv2_1_D"
type: "Convolution"
bottom: "conv2_2_D"
top: "conv2_1_D"
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"
value: 0
}
}
}
layer {
name: "conv2_1_D_bn"
type: "BN"
bottom: "conv2_1_D"
top: "conv2_1_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu2_1_D"
type: "ReLU"
bottom: "conv2_1_D"
top: "conv2_1_D"
}
layer {
name: "upsample1"
type: "Upsample"
bottom: "conv2_1_D"
bottom: "pool1_mask"
top: "pool1_D"
upsample_param {
scale: 2
}
}
layer {
name: "conv1_2_D"
type: "Convolution"
bottom: "pool1_D"
top: "conv1_2_D"
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"
value: 0
}
}
}
layer {
name: "conv1_2_D_bn"
type: "BN"
bottom: "conv1_2_D"
top: "conv1_2_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 1
decay_mult: 0
}
bn_param {
scale_filler {
type: "constant"
value: 1
}
shift_filler {
type: "constant"
value: 0
}
bn_mode: INFERENCE
}
}
layer {
name: "relu1_2_D"
type: "ReLU"
bottom: "conv1_2_D"
top: "conv1_2_D"
}
layer {
name: "conv1_1_D"
type: "Convolution"
bottom: "conv1_2_D"
top: "conv1_1_D"
param {
lr_mult: 1
decay_mult: 1
}
param {
lr_mult: 2
decay_mult: 0
}
convolution_param {
num_output: 15
pad: 1
kernel_size: 3
weight_filler {
type: "xavier"
}
bias_filler {
type: "constant"
value: 0
}
}
}
layer {
name: "prob"
type: "Softmax"
bottom: "conv1_1_D"
top: "prob"
softmax_param {
engine: CAFFE
}
}
I1029 20:53:35.094120 6127 layer_factory.hpp:77] Creating layer input
I1029 20:53:35.094143 6127 net.cpp:100] Creating Layer input
I1029 20:53:35.094153 6127 net.cpp:408] input -> data
I1029 20:53:35.101982 6127 net.cpp:150] Setting up input
I1029 20:53:35.102032 6127 net.cpp:157] Top shape: 2 3 352 1024 (2162688)
I1029 20:53:35.102036 6127 net.cpp:165] Memory required for data: 8650752
I1029 20:53:35.102053 6127 layer_factory.hpp:77] Creating layer conv1_1
I1029 20:53:35.102109 6127 net.cpp:100] Creating Layer conv1_1
I1029 20:53:35.102130 6127 net.cpp:434] conv1_1 <- data
I1029 20:53:35.102143 6127 net.cpp:408] conv1_1 -> conv1_1
I1029 20:53:35.290997 6127 net.cpp:150] Setting up conv1_1
I1029 20:53:35.291038 6127 net.cpp:157] Top shape: 2 64 352 1024 (46137344)
I1029 20:53:35.291045 6127 net.cpp:165] Memory required for data: 193200128
I1029 20:53:35.291100 6127 layer_factory.hpp:77] Creating layer conv1_1_bn
I1029 20:53:35.291123 6127 net.cpp:100] Creating Layer conv1_1_bn
I1029 20:53:35.291128 6127 net.cpp:434] conv1_1_bn <- conv1_1
I1029 20:53:35.291136 6127 net.cpp:395] conv1_1_bn -> conv1_1 (in-place)
I1029 20:53:35.292418 6127 net.cpp:150] Setting up conv1_1_bn
I1029 20:53:35.292439 6127 net.cpp:157] Top shape: 2 64 352 1024 (46137344)
I1029 20:53:35.292443 6127 net.cpp:165] Memory required for data: 377749504
I1029 20:53:35.292455 6127 layer_factory.hpp:77] Creating layer relu1_1
I1029 20:53:35.292464 6127 net.cpp:100] Creating Layer relu1_1
I1029 20:53:35.292469 6127 net.cpp:434] relu1_1 <- conv1_1
I1029 20:53:35.292474 6127 net.cpp:395] relu1_1 -> conv1_1 (in-place)
I1029 20:53:35.292676 6127 net.cpp:150] Setting up relu1_1
I1029 20:53:35.292686 6127 net.cpp:157] Top shape: 2 64 352 1024 (46137344)
I1029 20:53:35.292690 6127 net.cpp:165] Memory required for data: 562298880
I1029 20:53:35.292693 6127 layer_factory.hpp:77] Creating layer conv1_2
I1029 20:53:35.292706 6127 net.cpp:100] Creating Layer conv1_2
I1029 20:53:35.292709 6127 net.cpp:434] conv1_2 <- conv1_1
I1029 20:53:35.292716 6127 net.cpp:408] conv1_2 -> conv1_2
I1029 20:53:35.294865 6127 net.cpp:150] Setting up conv1_2
I1029 20:53:35.294893 6127 net.cpp:157] Top shape: 2 64 352 1024 (46137344)
I1029 20:53:35.294898 6127 net.cpp:165] Memory required for data: 746848256
I1029 20:53:35.294909 6127 layer_factory.hpp:77] Creating layer conv1_2_bn
I1029 20:53:35.294920 6127 net.cpp:100] Creating Layer conv1_2_bn
I1029 20:53:35.294924 6127 net.cpp:434] conv1_2_bn <- conv1_2
I1029 20:53:35.294930 6127 net.cpp:395] conv1_2_bn -> conv1_2 (in-place)
I1029 20:53:35.296283 6127 net.cpp:150] Setting up conv1_2_bn
I1029 20:53:35.296311 6127 net.cpp:157] Top shape: 2 64 352 1024 (46137344)
I1029 20:53:35.296315 6127 net.cpp:165] Memory required for data: 931397632
I1029 20:53:35.296324 6127 layer_factory.hpp:77] Creating layer relu1_2
I1029 20:53:35.296334 6127 net.cpp:100] Creating Layer relu1_2
I1029 20:53:35.296339 6127 net.cpp:434] relu1_2 <- conv1_2
I1029 20:53:35.296345 6127 net.cpp:395] relu1_2 -> conv1_2 (in-place)
I1029 20:53:35.296674 6127 net.cpp:150] Setting up relu1_2
I1029 20:53:35.296684 6127 net.cpp:157] Top shape: 2 64 352 1024 (46137344)
I1029 20:53:35.296699 6127 net.cpp:165] Memory required for data: 1115947008
I1029 20:53:35.296702 6127 layer_factory.hpp:77] Creating layer pool1
I1029 20:53:35.296706 6127 layer_factory.cpp:91] cuDNN does not support multiple tops. Using Caffe's own pooling layer.
I1029 20:53:35.296712 6127 net.cpp:100] Creating Layer pool1
I1029 20:53:35.296717 6127 net.cpp:434] pool1 <- conv1_2
I1029 20:53:35.296723 6127 net.cpp:408] pool1 -> pool1
I1029 20:53:35.296741 6127 net.cpp:408] pool1 -> pool1_mask
I1029 20:53:35.296797 6127 net.cpp:150] Setting up pool1
I1029 20:53:35.296814 6127 net.cpp:157] Top shape: 2 64 176 512 (11534336)
I1029 20:53:35.296819 6127 net.cpp:157] Top shape: 2 64 176 512 (11534336)
I1029 20:53:35.296823 6127 net.cpp:165] Memory required for data: 1208221696
I1029 20:53:35.296826 6127 layer_factory.hpp:77] Creating layer conv2_1
I1029 20:53:35.296839 6127 net.cpp:100] Creating Layer conv2_1
I1029 20:53:35.296844 6127 net.cpp:434] conv2_1 <- pool1
I1029 20:53:35.296855 6127 net.cpp:408] conv2_1 -> conv2_1
I1029 20:53:35.299191 6127 net.cpp:150] Setting up conv2_1
I1029 20:53:35.299223 6127 net.cpp:157] Top shape: 2 128 176 512 (23068672)
I1029 20:53:35.299226 6127 net.cpp:165] Memory required for data: 1300496384
I1029 20:53:35.299237 6127 layer_factory.hpp:77] Creating layer conv2_1_bn
I1029 20:53:35.299245 6127 net.cpp:100] Creating Layer conv2_1_bn
I1029 20:53:35.299249 6127 net.cpp:434] conv2_1_bn <- conv2_1
I1029 20:53:35.299255 6127 net.cpp:395] conv2_1_bn -> conv2_1 (in-place)
I1029 20:53:35.299832 6127 net.cpp:150] Setting up conv2_1_bn
I1029 20:53:35.299844 6127 net.cpp:157] Top shape: 2 128 176 512 (23068672)
I1029 20:53:35.299847 6127 net.cpp:165] Memory required for data: 1392771072
I1029 20:53:35.299854 6127 layer_factory.hpp:77] Creating layer relu2_1
I1029 20:53:35.299861 6127 net.cpp:100] Creating Layer relu2_1
I1029 20:53:35.299865 6127 net.cpp:434] relu2_1 <- conv2_1
I1029 20:53:35.299871 6127 net.cpp:395] relu2_1 -> conv2_1 (in-place)
I1029 20:53:35.300143 6127 net.cpp:150] Setting up relu2_1
I1029 20:53:35.300154 6127 net.cpp:157] Top shape: 2 128 176 512 (23068672)
I1029 20:53:35.300158 6127 net.cpp:165] Memory required for data: 1485045760
I1029 20:53:35.300161 6127 layer_factory.hpp:77] Creating layer conv2_2
I1029 20:53:35.300173 6127 net.cpp:100] Creating Layer conv2_2
I1029 20:53:35.300176 6127 net.cpp:434] conv2_2 <- conv2_1
I1029 20:53:35.300184 6127 net.cpp:408] conv2_2 -> conv2_2
I1029 20:53:35.302278 6127 net.cpp:150] Setting up conv2_2
I1029 20:53:35.302292 6127 net.cpp:157] Top shape: 2 128 176 512 (23068672)
I1029 20:53:35.302296 6127 net.cpp:165] Memory required for data: 1577320448
I1029 20:53:35.302304 6127 layer_factory.hpp:77] Creating layer conv2_2_bn
I1029 20:53:35.302314 6127 net.cpp:100] Creating Layer conv2_2_bn
I1029 20:53:35.302318 6127 net.cpp:434] conv2_2_bn <- conv2_2
I1029 20:53:35.302326 6127 net.cpp:395] conv2_2_bn -> conv2_2 (in-place)
I1029 20:53:35.302934 6127 net.cpp:150] Setting up conv2_2_bn
I1029 20:53:35.302947 6127 net.cpp:157] Top shape: 2 128 176 512 (23068672)
I1029 20:53:35.302959 6127 net.cpp:165] Memory required for data: 1669595136
I1029 20:53:35.302965 6127 layer_factory.hpp:77] Creating layer relu2_2
I1029 20:53:35.302983 6127 net.cpp:100] Creating Layer relu2_2
I1029 20:53:35.302986 6127 net.cpp:434] relu2_2 <- conv2_2
I1029 20:53:35.302991 6127 net.cpp:395] relu2_2 -> conv2_2 (in-place)
I1029 20:53:35.303185 6127 net.cpp:150] Setting up relu2_2
I1029 20:53:35.303194 6127 net.cpp:157] Top shape: 2 128 176 512 (23068672)
I1029 20:53:35.303206 6127 net.cpp:165] Memory required for data: 1761869824
I1029 20:53:35.303210 6127 layer_factory.hpp:77] Creating layer pool2
I1029 20:53:35.303215 6127 layer_factory.cpp:91] cuDNN does not support multiple tops. Using Caffe's own pooling layer.
I1029 20:53:35.303233 6127 net.cpp:100] Creating Layer pool2
I1029 20:53:35.303236 6127 net.cpp:434] pool2 <- conv2_2
I1029 20:53:35.303241 6127 net.cpp:408] pool2 -> pool2
I1029 20:53:35.303247 6127 net.cpp:408] pool2 -> pool2_mask
I1029 20:53:35.303294 6127 net.cpp:150] Setting up pool2
I1029 20:53:35.303303 6127 net.cpp:157] Top shape: 2 128 88 256 (5767168)
I1029 20:53:35.303318 6127 net.cpp:157] Top shape: 2 128 88 256 (5767168)
I1029 20:53:35.303320 6127 net.cpp:165] Memory required for data: 1808007168
I1029 20:53:35.303324 6127 layer_factory.hpp:77] Creating layer conv3_1
I1029 20:53:35.303333 6127 net.cpp:100] Creating Layer conv3_1
I1029 20:53:35.303337 6127 net.cpp:434] conv3_1 <- pool2
I1029 20:53:35.303344 6127 net.cpp:408] conv3_1 -> conv3_1
I1029 20:53:35.306005 6127 net.cpp:150] Setting up conv3_1
I1029 20:53:35.306030 6127 net.cpp:157] Top shape: 2 256 88 256 (11534336)
I1029 20:53:35.306032 6127 net.cpp:165] Memory required for data: 1854144512
I1029 20:53:35.306044 6127 layer_factory.hpp:77] Creating layer conv3_1_bn
I1029 20:53:35.306056 6127 net.cpp:100] Creating Layer conv3_1_bn
I1029 20:53:35.306059 6127 net.cpp:434] conv3_1_bn <- conv3_1
I1029 20:53:35.306066 6127 net.cpp:395] conv3_1_bn -> conv3_1 (in-place)
I1029 20:53:35.306267 6127 net.cpp:150] Setting up conv3_1_bn
I1029 20:53:35.306275 6127 net.cpp:157] Top shape: 2 256 88 256 (11534336)
I1029 20:53:35.306288 6127 net.cpp:165] Memory required for data: 1900281856
I1029 20:53:35.306293 6127 layer_factory.hpp:77] Creating layer relu3_1
I1029 20:53:35.306300 6127 net.cpp:100] Creating Layer relu3_1
I1029 20:53:35.306303 6127 net.cpp:434] relu3_1 <- conv3_1
I1029 20:53:35.306308 6127 net.cpp:395] relu3_1 -> conv3_1 (in-place)
I1029 20:53:35.306589 6127 net.cpp:150] Setting up relu3_1
I1029 20:53:35.306599 6127 net.cpp:157] Top shape: 2 256 88 256 (11534336)
I1029 20:53:35.306612 6127 net.cpp:165] Memory required for data: 1946419200
I1029 20:53:35.306615 6127 layer_factory.hpp:77] Creating layer conv3_2
I1029 20:53:35.306627 6127 net.cpp:100] Creating Layer conv3_2
I1029 20:53:35.306630 6127 net.cpp:434] conv3_2 <- conv3_1
I1029 20:53:35.306638 6127 net.cpp:408] conv3_2 -> conv3_2
I1029 20:53:35.310780 6127 net.cpp:150] Setting up conv3_2
I1029 20:53:35.310804 6127 net.cpp:157] Top shape: 2 256 88 256 (11534336)
I1029 20:53:35.310809 6127 net.cpp:165] Memory required for data: 1992556544
I1029 20:53:35.310832 6127 layer_factory.hpp:77] Creating layer conv3_2_bn
I1029 20:53:35.310842 6127 net.cpp:100] Creating Layer conv3_2_bn
I1029 20:53:35.310847 6127 net.cpp:434] conv3_2_bn <- conv3_2
I1029 20:53:35.310854 6127 net.cpp:395] conv3_2_bn -> conv3_2 (in-place)
I1029 20:53:35.311046 6127 net.cpp:150] Setting up conv3_2_bn
I1029 20:53:35.311053 6127 net.cpp:157] Top shape: 2 256 88 256 (11534336)
I1029 20:53:35.311056 6127 net.cpp:165] Memory required for data: 2038693888
I1029 20:53:35.311072 6127 layer_factory.hpp:77] Creating layer relu3_2
I1029 20:53:35.311079 6127 net.cpp:100] Creating Layer relu3_2
I1029 20:53:35.311082 6127 net.cpp:434] relu3_2 <- conv3_2
I1029 20:53:35.311089 6127 net.cpp:395] relu3_2 -> conv3_2 (in-place)
I1029 20:53:35.311360 6127 net.cpp:150] Setting up relu3_2
I1029 20:53:35.311372 6127 net.cpp:157] Top shape: 2 256 88 256 (11534336)
I1029 20:53:35.311385 6127 net.cpp:165] Memory required for data: 2084831232
I1029 20:53:35.311388 6127 layer_factory.hpp:77] Creating layer conv3_3
I1029 20:53:35.311399 6127 net.cpp:100] Creating Layer conv3_3
I1029 20:53:35.311403 6127 net.cpp:434] conv3_3 <- conv3_2
I1029 20:53:35.311410 6127 net.cpp:408] conv3_3 -> conv3_3
I1029 20:53:35.316313 6127 net.cpp:150] Setting up conv3_3
I1029 20:53:35.316332 6127 net.cpp:157] Top shape: 2 256 88 256 (11534336)
I1029 20:53:35.316352 6127 net.cpp:165] Memory required for data: 2130968576
I1029 20:53:35.316360 6127 layer_factory.hpp:77] Creating layer conv3_3_bn
I1029 20:53:35.316371 6127 net.cpp:100] Creating Layer conv3_3_bn
I1029 20:53:35.316375 6127 net.cpp:434] conv3_3_bn <- conv3_3
I1029 20:53:35.316382 6127 net.cpp:395] conv3_3_bn -> conv3_3 (in-place)
I1029 20:53:35.316571 6127 net.cpp:150] Setting up conv3_3_bn
I1029 20:53:35.316578 6127 net.cpp:157] Top shape: 2 256 88 256 (11534336)
I1029 20:53:35.316582 6127 net.cpp:165] Memory required for data: 2177105920
I1029 20:53:35.316586 6127 layer_factory.hpp:77] Creating layer relu3_3
I1029 20:53:35.316593 6127 net.cpp:100] Creating Layer relu3_3
I1029 20:53:35.316596 6127 net.cpp:434] relu3_3 <- conv3_3
I1029 20:53:35.316603 6127 net.cpp:395] relu3_3 -> conv3_3 (in-place)
I1029 20:53:35.316762 6127 net.cpp:150] Setting up relu3_3
I1029 20:53:35.316771 6127 net.cpp:157] Top shape: 2 256 88 256 (11534336)
I1029 20:53:35.316774 6127 net.cpp:165] Memory required for data: 2223243264
I1029 20:53:35.316778 6127 layer_factory.hpp:77] Creating layer pool3
I1029 20:53:35.316782 6127 layer_factory.cpp:91] cuDNN does not support multiple tops. Using Caffe's own pooling layer.
I1029 20:53:35.316788 6127 net.cpp:100] Creating Layer pool3
I1029 20:53:35.316792 6127 net.cpp:434] pool3 <- conv3_3
I1029 20:53:35.316797 6127 net.cpp:408] pool3 -> pool3
I1029 20:53:35.316804 6127 net.cpp:408] pool3 -> pool3_mask
I1029 20:53:35.316841 6127 net.cpp:150] Setting up pool3
I1029 20:53:35.316848 6127 net.cpp:157] Top shape: 2 256 44 128 (2883584)
I1029 20:53:35.316851 6127 net.cpp:157] Top shape: 2 256 44 128 (2883584)
I1029 20:53:35.316855 6127 net.cpp:165] Memory required for data: 2246311936
I1029 20:53:35.316859 6127 layer_factory.hpp:77] Creating layer pool3_drop
I1029 20:53:35.316865 6127 net.cpp:100] Creating Layer pool3_drop
I1029 20:53:35.316869 6127 net.cpp:434] pool3_drop <- pool3
I1029 20:53:35.316874 6127 net.cpp:395] pool3_drop -> pool3 (in-place)
I1029 20:53:35.316908 6127 net.cpp:150] Setting up pool3_drop
I1029 20:53:35.316915 6127 net.cpp:157] Top shape: 2 256 44 128 (2883584)
I1029 20:53:35.316918 6127 net.cpp:165] Memory required for data: 2257846272
I1029 20:53:35.316921 6127 layer_factory.hpp:77] Creating layer conv4_1
I1029 20:53:35.316932 6127 net.cpp:100] Creating Layer conv4_1
I1029 20:53:35.316936 6127 net.cpp:434] conv4_1 <- pool3
I1029 20:53:35.316941 6127 net.cpp:408] conv4_1 -> conv4_1
I1029 20:53:35.325304 6127 net.cpp:150] Setting up conv4_1
I1029 20:53:35.325330 6127 net.cpp:157] Top shape: 2 512 44 128 (5767168)
I1029 20:53:35.325333 6127 net.cpp:165] Memory required for data: 2280914944
I1029 20:53:35.325343 6127 layer_factory.hpp:77] Creating layer conv4_1_bn
I1029 20:53:35.325367 6127 net.cpp:100] Creating Layer conv4_1_bn
I1029 20:53:35.325374 6127 net.cpp:434] conv4_1_bn <- conv4_1
I1029 20:53:35.325382 6127 net.cpp:395] conv4_1_bn -> conv4_1 (in-place)
I1029 20:53:35.325567 6127 net.cpp:150] Setting up conv4_1_bn
I1029 20:53:35.325574 6127 net.cpp:157] Top shape: 2 512 44 128 (5767168)
I1029 20:53:35.325587 6127 net.cpp:165] Memory required for data: 2303983616
I1029 20:53:35.325592 6127 layer_factory.hpp:77] Creating layer relu4_1
I1029 20:53:35.325600 6127 net.cpp:100] Creating Layer relu4_1
I1029 20:53:35.325604 6127 net.cpp:434] relu4_1 <- conv4_1
I1029 20:53:35.325608 6127 net.cpp:395] relu4_1 -> conv4_1 (in-place)
I1029 20:53:35.326001 6127 net.cpp:150] Setting up relu4_1
I1029 20:53:35.326012 6127 net.cpp:157] Top shape: 2 512 44 128 (5767168)
I1029 20:53:35.326025 6127 net.cpp:165] Memory required for data: 2327052288
I1029 20:53:35.326028 6127 layer_factory.hpp:77] Creating layer conv4_2
I1029 20:53:35.326040 6127 net.cpp:100] Creating Layer conv4_2
I1029 20:53:35.326045 6127 net.cpp:434] conv4_2 <- conv4_1
I1029 20:53:35.326051 6127 net.cpp:408] conv4_2 -> conv4_2
I1029 20:53:35.340030 6127 net.cpp:150] Setting up conv4_2
I1029 20:53:35.340067 6127 net.cpp:157] Top shape: 2 512 44 128 (5767168)
I1029 20:53:35.340071 6127 net.cpp:165] Memory required for data: 2350120960
I1029 20:53:35.340090 6127 layer_factory.hpp:77] Creating layer conv4_2_bn
I1029 20:53:35.340102 6127 net.cpp:100] Creating Layer conv4_2_bn
I1029 20:53:35.340107 6127 net.cpp:434] conv4_2_bn <- conv4_2
I1029 20:53:35.340114 6127 net.cpp:395] conv4_2_bn -> conv4_2 (in-place)
I1029 20:53:35.340299 6127 net.cpp:150] Setting up conv4_2_bn
I1029 20:53:35.340306 6127 net.cpp:157] Top shape: 2 512 44 128 (5767168)
I1029 20:53:35.340309 6127 net.cpp:165] Memory required for data: 2373189632
I1029 20:53:35.340313 6127 layer_factory.hpp:77] Creating layer relu4_2
I1029 20:53:35.340319 6127 net.cpp:100] Creating Layer relu4_2
I1029 20:53:35.340323 6127 net.cpp:434] relu4_2 <- conv4_2
I1029 20:53:35.340339 6127 net.cpp:395] relu4_2 -> conv4_2 (in-place)
I1029 20:53:35.340616 6127 net.cpp:150] Setting up relu4_2
I1029 20:53:35.340626 6127 net.cpp:157] Top shape: 2 512 44 128 (5767168)
I1029 20:53:35.340629 6127 net.cpp:165] Memory required for data: 2396258304
I1029 20:53:35.340632 6127 layer_factory.hpp:77] Creating layer conv4_3
I1029 20:53:35.340663 6127 net.cpp:100] Creating Layer conv4_3
I1029 20:53:35.340667 6127 net.cpp:434] conv4_3 <- conv4_2
I1029 20:53:35.340683 6127 net.cpp:408] conv4_3 -> conv4_3
I1029 20:53:35.357230 6127 net.cpp:150] Setting up conv4_3
I1029 20:53:35.357260 6127 net.cpp:157] Top shape: 2 512 44 128 (5767168)
I1029 20:53:35.357265 6127 net.cpp:165] Memory required for data: 2419326976
I1029 20:53:35.357290 6127 layer_factory.hpp:77] Creating layer conv4_3_bn
I1029 20:53:35.357306 6127 net.cpp:100] Creating Layer conv4_3_bn
I1029 20:53:35.357312 6127 net.cpp:434] conv4_3_bn <- conv4_3
I1029 20:53:35.357322 6127 net.cpp:395] conv4_3_bn -> conv4_3 (in-place)
I1029 20:53:35.357561 6127 net.cpp:150] Setting up conv4_3_bn
I1029 20:53:35.357569 6127 net.cpp:157] Top shape: 2 512 44 128 (5767168)
I1029 20:53:35.357587 6127 net.cpp:165] Memory required for data: 2442395648
I1029 20:53:35.357594 6127 layer_factory.hpp:77] Creating layer relu4_3
I1029 20:53:35.357610 6127 net.cpp:100] Creating Layer relu4_3
I1029 20:53:35.357625 6127 net.cpp:434] relu4_3 <- conv4_3
I1029 20:53:35.357630 6127 net.cpp:395] relu4_3 -> conv4_3 (in-place)
I1029 20:53:35.357844 6127 net.cpp:150] Setting up relu4_3
I1029 20:53:35.357854 6127 net.cpp:157] Top shape: 2 512 44 128 (5767168)
I1029 20:53:35.357858 6127 net.cpp:165] Memory required for data: 2465464320
I1029 20:53:35.357862 6127 layer_factory.hpp:77] Creating layer pool4
I1029 20:53:35.357867 6127 layer_factory.cpp:91] cuDNN does not support multiple tops. Using Caffe's own pooling layer.
I1029 20:53:35.357887 6127 net.cpp:100] Creating Layer pool4
I1029 20:53:35.357892 6127 net.cpp:434] pool4 <- conv4_3
I1029 20:53:35.357898 6127 net.cpp:408] pool4 -> pool4
I1029 20:53:35.357908 6127 net.cpp:408] pool4 -> pool4_mask
I1029 20:53:35.357961 6127 net.cpp:150] Setting up pool4
I1029 20:53:35.357969 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.357985 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.357990 6127 net.cpp:165] Memory required for data: 2476998656
I1029 20:53:35.357995 6127 layer_factory.hpp:77] Creating layer pool4_drop
I1029 20:53:35.358003 6127 net.cpp:100] Creating Layer pool4_drop
I1029 20:53:35.358009 6127 net.cpp:434] pool4_drop <- pool4
I1029 20:53:35.358016 6127 net.cpp:395] pool4_drop -> pool4 (in-place)
I1029 20:53:35.358044 6127 net.cpp:150] Setting up pool4_drop
I1029 20:53:35.358060 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.358074 6127 net.cpp:165] Memory required for data: 2482765824
I1029 20:53:35.358079 6127 layer_factory.hpp:77] Creating layer conv5_1
I1029 20:53:35.358098 6127 net.cpp:100] Creating Layer conv5_1
I1029 20:53:35.358104 6127 net.cpp:434] conv5_1 <- pool4
I1029 20:53:35.358115 6127 net.cpp:408] conv5_1 -> conv5_1
I1029 20:53:35.380146 6127 net.cpp:150] Setting up conv5_1
I1029 20:53:35.380197 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.380204 6127 net.cpp:165] Memory required for data: 2488532992
I1029 20:53:35.380216 6127 layer_factory.hpp:77] Creating layer conv5_1_bn
I1029 20:53:35.380231 6127 net.cpp:100] Creating Layer conv5_1_bn
I1029 20:53:35.380249 6127 net.cpp:434] conv5_1_bn <- conv5_1
I1029 20:53:35.380268 6127 net.cpp:395] conv5_1_bn -> conv5_1 (in-place)
I1029 20:53:35.380529 6127 net.cpp:150] Setting up conv5_1_bn
I1029 20:53:35.380537 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.380551 6127 net.cpp:165] Memory required for data: 2494300160
I1029 20:53:35.380569 6127 layer_factory.hpp:77] Creating layer relu5_1
I1029 20:53:35.380578 6127 net.cpp:100] Creating Layer relu5_1
I1029 20:53:35.380584 6127 net.cpp:434] relu5_1 <- conv5_1
I1029 20:53:35.380590 6127 net.cpp:395] relu5_1 -> conv5_1 (in-place)
I1029 20:53:35.380939 6127 net.cpp:150] Setting up relu5_1
I1029 20:53:35.380959 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.380964 6127 net.cpp:165] Memory required for data: 2500067328
I1029 20:53:35.380969 6127 layer_factory.hpp:77] Creating layer conv5_2
I1029 20:53:35.380993 6127 net.cpp:100] Creating Layer conv5_2
I1029 20:53:35.381000 6127 net.cpp:434] conv5_2 <- conv5_1
I1029 20:53:35.381009 6127 net.cpp:408] conv5_2 -> conv5_2
I1029 20:53:35.394979 6127 net.cpp:150] Setting up conv5_2
I1029 20:53:35.395015 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.395020 6127 net.cpp:165] Memory required for data: 2505834496
I1029 20:53:35.395038 6127 layer_factory.hpp:77] Creating layer conv5_2_bn
I1029 20:53:35.395051 6127 net.cpp:100] Creating Layer conv5_2_bn
I1029 20:53:35.395056 6127 net.cpp:434] conv5_2_bn <- conv5_2
I1029 20:53:35.395063 6127 net.cpp:395] conv5_2_bn -> conv5_2 (in-place)
I1029 20:53:35.395256 6127 net.cpp:150] Setting up conv5_2_bn
I1029 20:53:35.395263 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.395277 6127 net.cpp:165] Memory required for data: 2511601664
I1029 20:53:35.395282 6127 layer_factory.hpp:77] Creating layer relu5_2
I1029 20:53:35.395288 6127 net.cpp:100] Creating Layer relu5_2
I1029 20:53:35.395292 6127 net.cpp:434] relu5_2 <- conv5_2
I1029 20:53:35.395298 6127 net.cpp:395] relu5_2 -> conv5_2 (in-place)
I1029 20:53:35.395965 6127 net.cpp:150] Setting up relu5_2
I1029 20:53:35.395975 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.395988 6127 net.cpp:165] Memory required for data: 2517368832
I1029 20:53:35.395992 6127 layer_factory.hpp:77] Creating layer conv5_3
I1029 20:53:35.396013 6127 net.cpp:100] Creating Layer conv5_3
I1029 20:53:35.396018 6127 net.cpp:434] conv5_3 <- conv5_2
I1029 20:53:35.396025 6127 net.cpp:408] conv5_3 -> conv5_3
I1029 20:53:35.409796 6127 net.cpp:150] Setting up conv5_3
I1029 20:53:35.409823 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.409827 6127 net.cpp:165] Memory required for data: 2523136000
I1029 20:53:35.409837 6127 layer_factory.hpp:77] Creating layer conv5_3_bn
I1029 20:53:35.409859 6127 net.cpp:100] Creating Layer conv5_3_bn
I1029 20:53:35.409864 6127 net.cpp:434] conv5_3_bn <- conv5_3
I1029 20:53:35.409880 6127 net.cpp:395] conv5_3_bn -> conv5_3 (in-place)
I1029 20:53:35.410104 6127 net.cpp:150] Setting up conv5_3_bn
I1029 20:53:35.410112 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.410125 6127 net.cpp:165] Memory required for data: 2528903168
I1029 20:53:35.410131 6127 layer_factory.hpp:77] Creating layer relu5_3
I1029 20:53:35.410138 6127 net.cpp:100] Creating Layer relu5_3
I1029 20:53:35.410142 6127 net.cpp:434] relu5_3 <- conv5_3
I1029 20:53:35.410147 6127 net.cpp:395] relu5_3 -> conv5_3 (in-place)
I1029 20:53:35.410320 6127 net.cpp:150] Setting up relu5_3
I1029 20:53:35.410328 6127 net.cpp:157] Top shape: 2 512 22 64 (1441792)
I1029 20:53:35.410331 6127 net.cpp:165] Memory required for data: 2534670336
I1029 20:53:35.410334 6127 layer_factory.hpp:77] Creating layer pool5
I1029 20:53:35.410337 6127 layer_factory.cpp:91] cuDNN does not support multiple tops. Using Caffe's own pooling layer.
I1029 20:53:35.410343 6127 net.cpp:100] Creating Layer pool5
I1029 20:53:35.410359 6127 net.cpp:434] pool5 <- conv5_3
I1029 20:53:35.410365 6127 net.cpp:408] pool5 -> pool5
I1029 20:53:35.410372 6127 net.cpp:408] pool5 -> pool5_mask
I1029 20:53:35.410421 6127 net.cpp:150] Setting up pool5
I1029 20:53:35.410429 6127 net.cpp:157] Top shape: 2 512 11 32 (360448)
I1029 20:53:35.410432 6127 net.cpp:157] Top shape: 2 512 11 32 (360448)
I1029 20:53:35.410434 6127 net.cpp:165] Memory required for data: 2537553920
I1029 20:53:35.410437 6127 layer_factory.hpp:77] Creating layer pool5_drop
I1029 20:53:35.410444 6127 net.cpp:100] Creating Layer pool5_drop
I1029 20:53:35.410457 6127 net.cpp:434] pool5_drop <- pool5
I1029 20:53:35.410464 6127 net.cpp:395] pool5_drop -> pool5 (in-place)
I1029 20:53:35.410495 6127 net.cpp:150] Setting up pool5_drop
I1029 20:53:35.410501 6127 net.cpp:157] Top shape: 2 512 11 32 (360448)
I1029 20:53:35.410513 6127 net.cpp:165] Memory required for data: 2538995712
I1029 20:53:35.410516 6127 layer_factory.hpp:77] Creating layer upsample5
I1029 20:53:35.410537 6127 net.cpp:100] Creating Layer upsample5
I1029 20:53:35.410552 6127 net.cpp:434] upsample5 <- pool5
I1029 20:53:35.410557 6127 net.cpp:434] upsample5 <- pool5_mask
I1029 20:53:35.410562 6127 net.cpp:408] upsample5 -> pool5_D
I1029 20:53:35.410598 6127 net.cpp:150] Setting up upsample5
I1029 20:53:35.410604 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.410609 6127 net.cpp:165] Memory required for data: 2541821952
I1029 20:53:35.410611 6127 layer_factory.hpp:77] Creating layer conv5_3_D
I1029 20:53:35.410622 6127 net.cpp:100] Creating Layer conv5_3_D
I1029 20:53:35.410626 6127 net.cpp:434] conv5_3_D <- pool5_D
I1029 20:53:35.410632 6127 net.cpp:408] conv5_3_D -> conv5_3_D
I1029 20:53:35.424585 6127 net.cpp:150] Setting up conv5_3_D
I1029 20:53:35.424613 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.424629 6127 net.cpp:165] Memory required for data: 2544648192
I1029 20:53:35.424639 6127 layer_factory.hpp:77] Creating layer conv5_3_D_bn
I1029 20:53:35.424652 6127 net.cpp:100] Creating Layer conv5_3_D_bn
I1029 20:53:35.424659 6127 net.cpp:434] conv5_3_D_bn <- conv5_3_D
I1029 20:53:35.424665 6127 net.cpp:395] conv5_3_D_bn -> conv5_3_D (in-place)
I1029 20:53:35.424873 6127 net.cpp:150] Setting up conv5_3_D_bn
I1029 20:53:35.424880 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.424883 6127 net.cpp:165] Memory required for data: 2547474432
I1029 20:53:35.424888 6127 layer_factory.hpp:77] Creating layer relu5_3_D
I1029 20:53:35.424896 6127 net.cpp:100] Creating Layer relu5_3_D
I1029 20:53:35.424909 6127 net.cpp:434] relu5_3_D <- conv5_3_D
I1029 20:53:35.424914 6127 net.cpp:395] relu5_3_D -> conv5_3_D (in-place)
I1029 20:53:35.425227 6127 net.cpp:150] Setting up relu5_3_D
I1029 20:53:35.425238 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.425252 6127 net.cpp:165] Memory required for data: 2550300672
I1029 20:53:35.425257 6127 layer_factory.hpp:77] Creating layer conv5_2_D
I1029 20:53:35.425268 6127 net.cpp:100] Creating Layer conv5_2_D
I1029 20:53:35.425273 6127 net.cpp:434] conv5_2_D <- conv5_3_D
I1029 20:53:35.425279 6127 net.cpp:408] conv5_2_D -> conv5_2_D
I1029 20:53:35.439083 6127 net.cpp:150] Setting up conv5_2_D
I1029 20:53:35.439112 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.439117 6127 net.cpp:165] Memory required for data: 2553126912
I1029 20:53:35.439128 6127 layer_factory.hpp:77] Creating layer conv5_2_D_bn
I1029 20:53:35.439141 6127 net.cpp:100] Creating Layer conv5_2_D_bn
I1029 20:53:35.439146 6127 net.cpp:434] conv5_2_D_bn <- conv5_2_D
I1029 20:53:35.439157 6127 net.cpp:395] conv5_2_D_bn -> conv5_2_D (in-place)
I1029 20:53:35.439373 6127 net.cpp:150] Setting up conv5_2_D_bn
I1029 20:53:35.439380 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.439394 6127 net.cpp:165] Memory required for data: 2555953152
I1029 20:53:35.439400 6127 layer_factory.hpp:77] Creating layer relu5_2_D
I1029 20:53:35.439407 6127 net.cpp:100] Creating Layer relu5_2_D
I1029 20:53:35.439411 6127 net.cpp:434] relu5_2_D <- conv5_2_D
I1029 20:53:35.439417 6127 net.cpp:395] relu5_2_D -> conv5_2_D (in-place)
I1029 20:53:35.439730 6127 net.cpp:150] Setting up relu5_2_D
I1029 20:53:35.439743 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.439756 6127 net.cpp:165] Memory required for data: 2558779392
I1029 20:53:35.439760 6127 layer_factory.hpp:77] Creating layer conv5_1_D
I1029 20:53:35.439774 6127 net.cpp:100] Creating Layer conv5_1_D
I1029 20:53:35.439779 6127 net.cpp:434] conv5_1_D <- conv5_2_D
I1029 20:53:35.439786 6127 net.cpp:408] conv5_1_D -> conv5_1_D
I1029 20:53:35.453780 6127 net.cpp:150] Setting up conv5_1_D
I1029 20:53:35.453816 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.453820 6127 net.cpp:165] Memory required for data: 2561605632
I1029 20:53:35.453831 6127 layer_factory.hpp:77] Creating layer conv5_1_D_bn
I1029 20:53:35.453843 6127 net.cpp:100] Creating Layer conv5_1_D_bn
I1029 20:53:35.453850 6127 net.cpp:434] conv5_1_D_bn <- conv5_1_D
I1029 20:53:35.453858 6127 net.cpp:395] conv5_1_D_bn -> conv5_1_D (in-place)
I1029 20:53:35.454074 6127 net.cpp:150] Setting up conv5_1_D_bn
I1029 20:53:35.454082 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.454085 6127 net.cpp:165] Memory required for data: 2564431872
I1029 20:53:35.454116 6127 layer_factory.hpp:77] Creating layer relu5_1_D
I1029 20:53:35.454123 6127 net.cpp:100] Creating Layer relu5_1_D
I1029 20:53:35.454128 6127 net.cpp:434] relu5_1_D <- conv5_1_D
I1029 20:53:35.454133 6127 net.cpp:395] relu5_1_D -> conv5_1_D (in-place)
I1029 20:53:35.454320 6127 net.cpp:150] Setting up relu5_1_D
I1029 20:53:35.454329 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.454344 6127 net.cpp:165] Memory required for data: 2567258112
I1029 20:53:35.454346 6127 layer_factory.hpp:77] Creating layer upsample4_drop
I1029 20:53:35.454355 6127 net.cpp:100] Creating Layer upsample4_drop
I1029 20:53:35.454360 6127 net.cpp:434] upsample4_drop <- conv5_1_D
I1029 20:53:35.454363 6127 net.cpp:395] upsample4_drop -> conv5_1_D (in-place)
I1029 20:53:35.454392 6127 net.cpp:150] Setting up upsample4_drop
I1029 20:53:35.454399 6127 net.cpp:157] Top shape: 2 512 23 30 (706560)
I1029 20:53:35.454401 6127 net.cpp:165] Memory required for data: 2570084352
I1029 20:53:35.454404 6127 layer_factory.hpp:77] Creating layer upsample4
I1029 20:53:35.454411 6127 net.cpp:100] Creating Layer upsample4
I1029 20:53:35.454413 6127 net.cpp:434] upsample4 <- conv5_1_D
I1029 20:53:35.454418 6127 net.cpp:434] upsample4 <- pool4_mask
I1029 20:53:35.454424 6127 net.cpp:408] upsample4 -> pool4_D
F1029 20:53:35.454461 6127 upsample_layer.cpp:59] Check failed: bottom[0]->height() == bottom[1]->height() (23 vs. 22)
*** Check failure stack trace: ***
Aborted (core dumped)
I set the input data size like this:
input_shape {
dim: 2 # SET SAMPLE SIZE HERE
dim: 3
dim: 352
dim: 1024
}
navganti commented
Hi @bzdfzfer - my apologies for the late response, I completely missed this notification. Glad to hear that the basic model worked for you!
I did run into this issue, but I was pretty sure I fixed it before I pushed. It is related to issues with the original SegNet model as seen here and here. It has to do with dimensionality mismatches while upsampling.
I'll take a look into it and see if I can find a fix.
navganti commented
@bzdfzfer - I think I have found it (but won't be able to test it until later).
Within the kitti/bayesian_segnet_standard.prototxt file at line 848/849, remove the upsample_h and upsample_w parameters. Keep the scale value. And repeat that at line 1042/1043 also. It looks like that's what's causing the mismatch, and I had removed it in the bayesian_segnet_basic.prototxt file.
Let me know if that works for you 🙂 .
bzdfzfer commented
I tested the standard model and it worked! Thank you very much for this problem.