Skip to content

Instantly share code, notes, and snippets.

@jhoffman
Last active March 9, 2019 13:52
  • Star 3 You must be signed in to star a gist
  • Fork 0 You must be signed in to fork a gist
Star You must be signed in to star a gist
Save jhoffman/9a28bcaf354f21ad3169f0679d73f647 to your computer and use it in GitHub Desktop.
Caffe Prototxt example files to run supervised domain confusion experiment. Takes 3 data sources as input (1) source training labeled data (2) target training labeled data (3) target test unlabeled data
average_loss: 25
base_lr: 0.000100
display: 25
gamma: 0.1
iter_size: 1
lr_policy: "fixed"
max_iter: 100000
momentum: 0.9
net: "trainval_domain-confusion_super.prototxt"
snapshot: 1000
snapshot_prefix: "snapshots/amazon_webcam_dom-conf-super_split-0"
stepsize: 1000
test_initialization: false
test_interval: 50
test_iter: 256
weight_decay: 0.0005
layer {
name: "data_s"
type: "ImageData"
top: "data_s"
top: "label_s"
include {
phase: TRAIN
}
transform_param {
mirror: true
crop_size: 227
mean_value: 104.0
mean_value: 116.7
mean_value: 122.7
}
image_data_param {
source: "data/splits/office/amazon/source/train/split_00.txt"
batch_size: 15
shuffle: false
new_width: 256
new_height: 256
}
}
layer {
name: "data_t"
type: "ImageData"
top: "data_t"
top: "label_t"
include {
phase: TRAIN
}
transform_param {
mirror: true
crop_size: 227
mean_value: 104.0
mean_value: 116.7
mean_value: 122.7
}
image_data_param {
source: "data/splits/office/webcam/target/train/same_category-split_00.txt"
batch_size: 15
shuffle: false
new_width: 256
new_height: 256
}
}
layer {
name: "data"
type: "Concat"
bottom: "data_s"
bottom: "data_t"
top: "data"
include {
phase: TRAIN
}
concat_param {
concat_dim: 0
}
}
layer {
name: "data"
type: "ImageData"
top: "data"
top: "label"
include {
phase: TEST
}
transform_param {
mirror: false
crop_size: 227
mean_value: 104.0
mean_value: 116.7
mean_value: 122.7
}
image_data_param {
source: "data/splits/office/webcam/target/test/same_category-split_00.txt"
batch_size: 1
shuffle: false
new_width: 256
new_height: 256
}
}
layer {
name: "data"
type: "Concat"
bottom: "label_s"
bottom: "label_t"
top: "label"
include {
phase: TRAIN
}
concat_param {
concat_dim: 0
}
}
layer {
name: "source_domain_label"
type: "DummyData"
top: "source_domain_label"
include {
phase: TRAIN
}
dummy_data_param {
data_filler {
type: "constant"
value: 0
}
shape {
dim: 15
dim: 1
dim: 1
}
}
}
layer {
name: "target_domain_label"
type: "DummyData"
top: "target_domain_label"
include {
phase: TRAIN
}
dummy_data_param {
data_filler {
type: "constant"
value: 1
}
shape {
dim: 15
dim: 1
dim: 1
}
}
}
layer {
name: "domain_label"
type: "Concat"
bottom: "source_domain_label"
bottom: "target_domain_label"
top: "domain_label"
include {
phase: TRAIN
}
concat_param {
concat_dim: 0
}
}
layer {
name: "source_domain_label_inv"
type: "DummyData"
top: "source_domain_label_inv"
include {
phase: TRAIN
}
dummy_data_param {
data_filler {
type: "constant"
value: 1
}
shape {
dim: 15
dim: 1
dim: 1
}
}
}
layer {
name: "target_domain_label_inv"
type: "DummyData"
top: "target_domain_label_inv"
include {
phase: TRAIN
}
dummy_data_param {
data_filler {
type: "constant"
value: 0
}
shape {
dim: 15
dim: 1
dim: 1
}
}
}
layer {
name: "domain_label_inv"
type: "Concat"
bottom: "source_domain_label_inv"
bottom: "target_domain_label_inv"
top: "domain_label_inv"
include {
phase: TRAIN
}
concat_param {
concat_dim: 0
}
}
layer {
name: "conv1"
type: "Convolution"
bottom: "data"
top: "conv1"
param {
lr_mult: 1.0
decay_mult: 1.0
}
param {
lr_mult: 2.0
decay_mult: 0.0
}
convolution_param {
num_output: 96
pad: 0
kernel_size: 11
group: 1
stride: 4
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.0
}
}
}
layer {
name: "relu1"
type: "ReLU"
bottom: "conv1"
top: "conv1"
}
layer {
name: "pool1"
type: "Pooling"
bottom: "conv1"
top: "pool1"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "norm1"
type: "LRN"
bottom: "pool1"
top: "norm1"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "conv2"
type: "Convolution"
bottom: "norm1"
top: "conv2"
param {
lr_mult: 1.0
decay_mult: 1.0
}
param {
lr_mult: 2.0
decay_mult: 0.0
}
convolution_param {
num_output: 256
pad: 2
kernel_size: 5
group: 2
stride: 1
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.0
}
}
}
layer {
name: "relu2"
type: "ReLU"
bottom: "conv2"
top: "conv2"
}
layer {
name: "pool2"
type: "Pooling"
bottom: "conv2"
top: "pool2"
pooling_param {
pool: MAX
kernel_size: 3
stride: 2
}
}
layer {
name: "norm2"
type: "LRN"
bottom: "pool2"
top: "norm2"
lrn_param {
local_size: 5
alpha: 0.0001
beta: 0.75
}
}
layer {
name: "conv3"
type: "Convolution"
bottom: "norm2"
top: "conv3"
param {
lr_mult: 1.0
decay_mult: 1.0
}
param {
lr_mult: 2.0
decay_mult: 0.0
}
convolution_param {
num_output: 384
pad: 1
kernel_size: 3
group: 1
stride: 1
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.0
}
}
}
layer {
name: "relu3"
type: "ReLU"
bottom: "conv3"
top: "conv3"
}
layer {
name: "conv4"
type: "Convolution"
bottom: "conv3"
top: "conv4"
param {
lr_mult: 1.0
decay_mult: 1.0
}
param {
lr_mult: 2.0
decay_mult: 0.0
}
convolution_param {
num_output: 384
pad: 1
kernel_size: 3
group: 2
stride: 1
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.0
}
}
}
layer {
name: "relu4"
type: "ReLU"
bottom: "conv4"
top: "conv4"
}
layer {
name: "conv5"
type: "Convolution"
bottom: "conv4"
top: "conv5"
param {
lr_mult: 1.0
decay_mult: 1.0
}
param {
lr_mult: 2.0
decay_mult: 0.0
}
convolution_param {
num_output: 256
pad: 1
kernel_size: 3
group: 2
stride: 1
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.0
}
}
}
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: 1.0
decay_mult: 1.0
}
param {
lr_mult: 2.0
decay_mult: 0.0
}
inner_product_param {
num_output: 4096
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.0
}
}
}
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: 1.0
decay_mult: 1.0
}
param {
lr_mult: 2.0
decay_mult: 0.0
}
inner_product_param {
num_output: 4096
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.0
}
}
}
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_new"
type: "InnerProduct"
bottom: "fc7"
top: "fc8_new"
param {
lr_mult: 1.0
decay_mult: 1.0
}
param {
lr_mult: 2.0
decay_mult: 0.0
}
inner_product_param {
num_output: 31
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.0
}
}
}
layer {
name: "domain"
type: "DomainConfusionInnerProduct"
bottom: "fc7"
top: "domain"
top: "domain_inv"
param {
lr_mult: 1.0
decay_mult: 1.0
}
param {
lr_mult: 2.0
decay_mult: 0.0
}
include {
phase: TRAIN
}
inner_product_param {
num_output: 2
weight_filler {
type: "gaussian"
std: 0.005
}
bias_filler {
type: "constant"
value: 0.0
}
}
}
layer {
name: "loss_domain"
type: "SoftmaxWithLoss"
bottom: "domain"
bottom: "domain_label"
top: "loss_domain"
loss_weight: 0.1
include {
phase: TRAIN
}
}
layer {
name: "loss_domain_inv"
type: "SoftmaxWithLoss"
bottom: "domain_inv"
bottom: "domain_label_inv"
top: "loss_domain_inv"
loss_weight: 0.1
include {
phase: TRAIN
}
}
layer {
name: "domain_accuracy"
type: "Accuracy"
bottom: "domain"
bottom: "domain_label"
top: "domain_accuracy"
include {
phase: TRAIN
}
}
layer {
name: "loss_tr"
type: "SoftmaxWithLoss"
bottom: "fc8_new"
bottom: "label"
top: "loss_tr"
loss_weight: 1.0
include {
phase: TRAIN
}
}
layer {
name: "loss"
type: "SoftmaxWithLoss"
bottom: "fc8_new"
bottom: "label"
top: "loss"
loss_weight: 1.0
include {
phase: TEST
}
}
layer {
name: "prob"
type: "Softmax"
bottom: "fc8_new"
top: "prob"
include {
phase: TEST
}
}
layer {
name: "accuracy"
type: "Accuracy"
bottom: "prob"
bottom: "label"
top: "accuracy"
include {
phase: TEST
}
}
@hala3
Copy link

hala3 commented Mar 9, 2019

me too i want to know how i can prepare the data for the multi source classification

Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment