Trjtdtk

Was Dennis Ritchie being too modest in this quote about C and Pascal?

Are there moral objections to a life motivated purely by money? How to sway a person from this lifestyle?

Multiple fireplaces in an apartment building?

Contradiction proof for inequality of P and NP?

Why must Chinese maps be obfuscated?

What *exactly* is electrical current, voltage, and resistance?

Should the Product Owner dictate what info the UI needs to display?

finding a tangent line to a parabola

How to find the stem of any word?

Is there any pythonic way to find average of specific tuple elements in array?

Double-nominative constructions and “von”

How do I prove this combinatorial identity

Check if a string is entirely made of the same substring

Is it acceptable to use working hours to read general interest books?

How to keep bees out of canned beverages?

Is Electric Central Heating worth it if using Solar Panels?

My admission is revoked after accepting the admission offer

My bank got bought out, am I now going to have to start filing tax returns in a different state?

How to translate "red flag" into Spanish?

Why doesn't the standard consider a template constructor as a copy constructor?

What does a straight horizontal line above a few notes, after a changed tempo mean?

Can I criticise the more senior developers around me for not writing clean code?

Can you stand up from being prone using Skirmisher outside of your turn?

Did the Roman Empire have penal colonies?

Extract feature vector of a CNN

5

$begingroup$

How can I get the feature vector of my dataset. I have a fine-tuned CNN model with my data. Now I want to feed the features of all my dataset extracted from the last layer of the CNN into a LSTM.

So far I have

cnn_model = load_model('weights/pre-trained_CNN.hdf5')
cnn_output = cnn_model.get_layer('fc7').output

I know I will have to feed all my dataset into this model but I have no idea on how to "save" the features.

keras feature-extraction cnn

share|improve this question

asked Jan 23 '18 at 16:10

Daniel ZapataDaniel Zapata

264

$endgroup$

add a comment | 

5

$begingroup$

How can I get the feature vector of my dataset. I have a fine-tuned CNN model with my data. Now I want to feed the features of all my dataset extracted from the last layer of the CNN into a LSTM.

So far I have

cnn_model = load_model('weights/pre-trained_CNN.hdf5')
cnn_output = cnn_model.get_layer('fc7').output

I know I will have to feed all my dataset into this model but I have no idea on how to "save" the features.

keras feature-extraction cnn

share|improve this question

asked Jan 23 '18 at 16:10

Daniel ZapataDaniel Zapata

264

$endgroup$

add a comment | 

$begingroup$

How can I get the feature vector of my dataset. I have a fine-tuned CNN model with my data. Now I want to feed the features of all my dataset extracted from the last layer of the CNN into a LSTM.

So far I have

cnn_model = load_model('weights/pre-trained_CNN.hdf5')
cnn_output = cnn_model.get_layer('fc7').output

I know I will have to feed all my dataset into this model but I have no idea on how to "save" the features.

keras feature-extraction cnn

share|improve this question

asked Jan 23 '18 at 16:10

Daniel ZapataDaniel Zapata

264

$endgroup$

cnn_model = load_model('weights/pre-trained_CNN.hdf5')
cnn_output = cnn_model.get_layer('fc7').output

share|improve this question

asked Jan 23 '18 at 16:10

Daniel ZapataDaniel Zapata

264

share|improve this question

asked Jan 23 '18 at 16:10

Daniel ZapataDaniel Zapata

264

asked Jan 23 '18 at 16:10

Daniel ZapataDaniel Zapata

264

asked Jan 23 '18 at 16:10

Daniel ZapataDaniel Zapata

264

1 Answer
1

active

oldest

votes

1

$begingroup$

In order to "tap" intermediate layers of an existing model you could do the following:

# get your feature layer
cnn_model = load_model('weights/pre-trained_CNN.hdf5')
feature_layer = cnn_model.get_layer('fc7')

# stack your LSTM and other layers below, e.g.:
lstm_layer = TimeDistributed(LSTM(...), input_shape = ...)(feature_layer)
output = Dense(...)(lstm_layer)

# create a combined model 
model = Model(inputs = cnn_model.input, outputs = [output])

share|improve this answer

answered Apr 6 at 17:57

m0nzderrm0nzderr

863

$endgroup$

add a comment | 

Your Answer

StackExchange.ready(function()
var channelOptions =
tags: "".split(" "),
id: "557"
;
initTagRenderer("".split(" "), "".split(" "), channelOptions);

StackExchange.using("externalEditor", function()
// Have to fire editor after snippets, if snippets enabled
if (StackExchange.settings.snippets.snippetsEnabled)
StackExchange.using("snippets", function()
createEditor();
);

else
createEditor();

);

function createEditor()
StackExchange.prepareEditor(
heartbeatType: 'answer',
autoActivateHeartbeat: false,
convertImagesToLinks: false,
noModals: true,
showLowRepImageUploadWarning: true,
reputationToPostImages: null,
bindNavPrevention: true,
postfix: "",
imageUploader:
brandingHtml: "Powered by u003ca class="icon-imgur-white" href="https://imgur.com/"u003eu003c/au003e",
contentPolicyHtml: "User contributions licensed under u003ca href="https://creativecommons.org/licenses/by-sa/3.0/"u003ecc by-sa 3.0 with attribution requiredu003c/au003e u003ca href="https://stackoverflow.com/legal/content-policy"u003e(content policy)u003c/au003e",
allowUrls: true
,
onDemand: true,
discardSelector: ".discard-answer"
,immediatelyShowMarkdownHelp:true
);



);

draft saved

draft discarded

Sign up or log in

StackExchange.ready(function ()
StackExchange.helpers.onClickDraftSave('#login-link');
);

Sign up using Google

Sign up using Facebook

Sign up using Email and Password

Post as a guest

Name

Email

Required, but never shown

StackExchange.ready(
function ()
StackExchange.openid.initPostLogin('.new-post-login', 'https%3a%2f%2fdatascience.stackexchange.com%2fquestions%2f26963%2fextract-feature-vector-of-a-cnn%23new-answer', 'question_page');

);

Post as a guest

Name

Email

Required, but never shown

1

$begingroup$

In order to "tap" intermediate layers of an existing model you could do the following:

# get your feature layer
cnn_model = load_model('weights/pre-trained_CNN.hdf5')
feature_layer = cnn_model.get_layer('fc7')

# stack your LSTM and other layers below, e.g.:
lstm_layer = TimeDistributed(LSTM(...), input_shape = ...)(feature_layer)
output = Dense(...)(lstm_layer)

# create a combined model 
model = Model(inputs = cnn_model.input, outputs = [output])

share|improve this answer

answered Apr 6 at 17:57

m0nzderrm0nzderr

863

$endgroup$

add a comment | 

1

$begingroup$

In order to "tap" intermediate layers of an existing model you could do the following:

# get your feature layer
cnn_model = load_model('weights/pre-trained_CNN.hdf5')
feature_layer = cnn_model.get_layer('fc7')

# stack your LSTM and other layers below, e.g.:
lstm_layer = TimeDistributed(LSTM(...), input_shape = ...)(feature_layer)
output = Dense(...)(lstm_layer)

# create a combined model 
model = Model(inputs = cnn_model.input, outputs = [output])

share|improve this answer

answered Apr 6 at 17:57

m0nzderrm0nzderr

863

$endgroup$

add a comment | 

$begingroup$

In order to "tap" intermediate layers of an existing model you could do the following:

# get your feature layer
cnn_model = load_model('weights/pre-trained_CNN.hdf5')
feature_layer = cnn_model.get_layer('fc7')

# stack your LSTM and other layers below, e.g.:
lstm_layer = TimeDistributed(LSTM(...), input_shape = ...)(feature_layer)
output = Dense(...)(lstm_layer)

# create a combined model 
model = Model(inputs = cnn_model.input, outputs = [output])

share|improve this answer

answered Apr 6 at 17:57

m0nzderrm0nzderr

863

$endgroup$

# get your feature layer
cnn_model = load_model('weights/pre-trained_CNN.hdf5')
feature_layer = cnn_model.get_layer('fc7')

# stack your LSTM and other layers below, e.g.:
lstm_layer = TimeDistributed(LSTM(...), input_shape = ...)(feature_layer)
output = Dense(...)(lstm_layer)

# create a combined model 
model = Model(inputs = cnn_model.input, outputs = [output])

share|improve this answer

answered Apr 6 at 17:57

m0nzderrm0nzderr

863

answered Apr 6 at 17:57

m0nzderrm0nzderr

863

answered Apr 6 at 17:57

m0nzderrm0nzderr

863