Customizing MobileNetV2 through transfer learning using Keras and Python [Part 4]

[@felixkunert4176]

Great video for getting started with transfer learning

[@MuskanRajivBathlaBCSE]

helpful videos!

[@adnanimran4414]

ur the best

[@anemone5264]

Very nice and descriptive! I saw that MobileNet also available at PyTorch. Does it has the same implementation to build new model by using PyTorch instead Keras?

Yes you can do everything with pytorch that you can do with keras/ tensorflow!

[@Vineas11]

can this mobilenet model automatically be used for tensorflow lite?

[@lLlLawliet1]

then how to save the model into file?

[@zahragholami7534]

Does this model work for images with other sizes like 112*112?

[@KashifShaheed]

Sigmoid we not used for binary classification?

I'm not sure this answers your question (?), but the softmax function actually is used as the activation function in the output layer to squeeze values into a 0 - 1 range and to make sure that they add up to 1.

[@KashifShaheed]

@ often called softmax in this context ? In Binary classification?

[@amanjain6680]

@@KashifShaheed Softmax can be used for binary classification as well. Although using sigmoid is most people prefer. But there is not difference in result

Thanks! I actually was a bit confused in my answer as well. Softmax is not the same as a sigmoid, but it does accomplish the similar goal of squeezing values into a fixed range.

[@ramakanthyadav828]

bro tell me which IDE you are using?

[@shokhrukhbibalaev8101]

Can we add more layers instead of one output layer to the model?

[@ethangunter9330]

You can add as many as you want following the same pattern. It's actually quite common to add a flatten, and couple of dense layers with classification problems like this

[@arinjain25]

Pls reply sir i am making project on skin disease classification, using HAM 10000 dataset, i am getting problem in modle.fit() fucntion while using Tensorflow2.0 it is showing attribute error -(' Directory Iterator ' object has no attribute) plz reply sir, its my university project i have to show it to my professors.