Transfer Learning | How to Extract Features from Images?

Transfer Learning enables you to use the power of the best machine learning models on your projects. In this video, I have explained how it works and how to implement it with Keras.
#transferlearning #keras #python

For more videos please subscribe -

Support me if you can ❤️

Notebook -

Keras Doc -

Playlist ML Algos from Scratch -

ML concepts -

Facebook -
Instagram -
Twitter -

[rdurian]

You *already* learned a language (English, Hindi, Spanish, whatever). Now you read a new post that has maybe 2% new vocab in it. But 98% is just recombinations of the language you *already* learned. That's the analogy.
The real trick is coming up with huge amounts of clean, properly labelled, organized and useful data. Then extracting *features* from it.
That was (apparently) done by the Model1 training on <Big Data>....
and even then....the truly *hard* part is having proper data at of images of it....

[switches_slips_turnouts]

how to create image classification model when classes are highly imbalances

[ahmedmohammed]

i would like to appericiate your act of presenting but i need to include actual class with predicted class

[hayaquraan]

please i got this error
ValueError: could not broadcast input array from shape (28, 7, 7, 512) into shape (32, 7, 7, 512)
how to solve it

[kayvengoh]

Sorry, I am new to machine learning (ML) and deep learning (DL) world. I have a question. Is this video of transfer learning considered to be DL application? Since I learned that Keras application is a DL models that are made available alongside pre-trained weights. I really need your help in these confusion. I am very sorry for the inconvenience.

[heshamkhaled]

I am trying your code to extract features from a small image dataset that i have but when i use the function(extract_features) and print the features i found it have a lot of 0.0 . so is this normal ?

[heshamkhaled]

i am really sorry for asking you again but i need your help, i tried to print the labels from the extract_feature function and it was like this : 0., 0., 0., 0., 0., 0.],
[1., 1., 1., 1., 1., 1.],
[0., 0., 0., 0., 0., 0.],
[0., 0., 0., 0., 0., 0.],
[1., 1., 1., 1., 1., 1.],
[1., 1., 1., 1., 1., 1.],
[0., 0., 0., 0., 0., 0.],
[1., 1., 1., 1., 1., 1.]
i have 2 classes cat and dog, like u see the labels of 0 are not consecutive there is (zero label) then the (1 label) then the (zero label) in random way, the labels of the cats are not consecutive and after the cats the dogs come .
so is this mean that cats features are not consecutive in the features array and the array of features has the images in random way ?
and if this is true what should i do to put all the cats features and labels then go for the dogs ?

[masbro]

hi sir, when i try to extract features, i got this error: MemoryError: Unable to allocate 9.68 GiB for an array with shape (51809, 7, 7, 512) and data type float64

how do i solve this.
many thanks

[faheemanjum]

sir i need help to evaluate classification report and confusion matrix

[himanshukharwar]

What exactly are these feature map ?
like from first layer what are the outputs we get?

[abhijeetanand]

can you name a few features of image which is being extracted by convolutional base? Anyone questioned me on which features my model is working?

[muhammadzubairbaloch]

Sir I want you make more lectures on NLP. thanks.
I like your work.

[pechaaloler]

Hey, i really like this video! I would like to build a model to identify textures in images, what kind of approach would you recommend?

[arshkatyal]

Which architecture do you consider best for feature extraction of sparse data (i.e images of line graphs)

[furqanafridi]

I am facing this error kindly help
"Found 0 images belonging to 0 classes.


ValueError Traceback (most recent call last)
Input In [102], in <cell line: 27>()
24 break
25 return features, labels
---> 27 train_features, train_labels = extract_features(train_dir, 30) # Agree with our small dataset size
28 validation_features, validation_labels = extract_features(validation_dir, 2)

Input In [102], in extract_features(directory, sample_count)
17 i = 0
18 for inputs_batch, labels_batch in generator:
---> 19 features_batch =
20 features[i * batch_size: (i + 1) * batch_size] = features_batch
21 labels[i * batch_size: (i + 1) * batch_size] = labels_batch"