Transfer Learning Using CNN(VGG 16)| Keras Tutorial|

[@simonegasparella1228]

Hi, How can we see the accuracy chart of the validation test set? How can we see the incorrect pictures in the test set? Thanks a lot

[@mohammedzia1015]

I have one question, the VGG16 model would have been trained for this type of Dataset ? I mean for the classes Benign and Malignant ?

[@dishachakraborty9769]

I need help. I don't have separate train, test and validation data for my dataset on Kaggle. The dataset only has 3 folders - Normal, Benign and Malginant which contain images but no subfolders for train and test.

[@komaldaoud3445]

how can we use grid search cv with model.fit??

[@sayansaha5348]

Thanks bhaiya can you please make a separate video for the prediction part only

[@umanggupta1003]

Can you show end to end code with model.Predict() to verify the results?

[@anchalsharma6163]

Hloo sir can you help ke for plot graph

[@rohit_mondal__]

Sir if you could explain the vgg.input thing we are putting in the inputs section of the model !

[@sherin7444]

1 question.Lets say in my train file couple of images are not of 224*224 .How can we convert them to standard size

[@pavantripathi1890]

Can you please show how to predict on a new image.

[@rakeshsahoo3711]

Can u apply k-fold cross-validation for this

[@TitoWPurnomo]

I've run code with jupyter-lab but during training process always appears statement "the kernel appears to have died. it will restart automatically" . do you know the solution?

[@vanshajgoel8783]

InvalidArgumentError: Graph execution error: -->Any idea why this error is coming ??

[@NachiketaHebbar]

Normally you get this error when there is a mismatch in the number of data classes that you defined in model and the one which you are actually giving it. Double check the input shape, target size , number of dataset folders and the number of output layers

[@alexomelak4200]

Sir ,help me please? I was doing gray scale image classification using VGG16. their is an error when I update the code from vgg = VGG16(input_shape=IMAGE_SIZE + [3], weights='imagenet', include_top=False) to vgg = VGG16(input_shape=IMAGE_SIZE + [1], weights='imagenet', include_top=False). help me please

[@minakshisonawane5929]

sir can u help me when i create folder but i have error .

[@darshitavijayakumar2021]

I'm getting that the object is not callable . Please help me

[@ashikroy4618]

Hi Nachiketa, Thank you for your tutorials. I am a beginner, and I am wandering why didn't you use Validation and Test, both of the datasets to evaluate your transfer learned model? Thanks

[@swatimishra1555]

I am also not getting validation accuracy and loss because they n have not to use a train test split. I really don't know how he is getting validation accuracy and loss for the model. Plz, include this in the code.

[@swatimishra1555]

Hi Nachiketa, plz clear this doubt

[@leo7662]

​@@swatimishra1555when u r writing u just need to write validation split =0.2 that will automatically take 20% of our data as validation set randomly so u will get both validation loss and accuracy

[@bhavukthakur9501]

sir, when you do ImageDataGenerator step how many images we get? Like, if we have 10 image in train dataset and we apply rotation_range, width_shift_range, height_shift_range=0.2 so we get 30 images or just 10 images?

[@girimadhavchowdhury1281]

I could not able to resize my input into 224*224*3... Can any one help me on this

[@aninditasaha0307]

Hi... Can anyone send me the confusion matrix code for this prob?

[@byDes]

Hi! How did you split the train, val, and test dataset?

[@sololife9403]

if you download the dataset from kaggle, you will find out that the train and test sets were already split for us. but i did not find val , though

[@Megh93]

getting error in model.fit_generator

[@sagnikroy6405]

Thanks @Nachiketa . Really helped. One question , what does the .h5 extension is for?? Storing trained NNs??

[@NachiketaHebbar]

Yeah, it stores the model architecture and weights

[@jaideeppanwar2730]

hey.....can we make a gradio app on this ......i mean it will the same process ....that u showed in the earlier video ?

[@NachiketaHebbar]

Yes, you can. I had shown an inception model in that video, you can just replace that by this vgg model. It will work

[@hamzanaeem4838]

What if we don't use steps per epoch and validation steps ? Can we use model.fit only if we don't apply augmentation.

[@sagnikroy6405]

I guess YES, as the fit_generator is deprecated in the newest version of tensorflow or keras.

[@nikolaostziokas6847]

Great video. I have only one question. What are the dimensions of your training and test images? Do they have to be 224*224 before you import them into the algorithm?

[@shreypatel3782]

NO.

[@sherin7444]

VGG16 expects img size to be 224*224

[@madhukanadeeshani6982]

Hi it's great .can you please share the code and link of the dataser

[@NachiketaHebbar]

Already there in the description

[@reemalmijmaj3215]

This video save my life :)) Thanks a lot !!

[@vanillasky1842]

So if I'm using grayscale images,where should I make the changes?

[@NachiketaHebbar]

in the input shape. Instead of 3 as the depth of image, use 1.

[@syedzoofa8711]

Ur videos have helped me a lot.. Thank u

[@byDes]

Hello, this helped me a lot! I hope you could also make a tutorial for medical images using mobilenetv2

[@OpeLeke]

Great Video, it's skin lesion dataset not skin cancer.

[@shivambaldha3914]

Very very useful for me , thank you :)

[@rasanjalirathnayake8873]

Instead of classification do something related to deep learning with regression analysis please.

[@kaouthergacem8135]

your video helped me a lot thank you

[@adarshrkr9355]

Bro when im doing prediction im only getting class 0 for all my test data Any idea?

[@NachiketaHebbar]

If your test accuracy is good, then most probably the mistake you are making is in pre processing the input. For example, if while training you are rescaling the input or something else, then you have to do the same on an image before making predictions also.

[@adarshrkr9355]

@@NachiketaHebbar ok bro thankyou will try that out

[@simonegasparella1228]

Hi Sir, after training how can we make a prediction in one specific picture? Can you share the code ? Thanks

[@simonegasparella1228]

@@NachiketaHebbar Can you provide the code for making a prediction on a single image? THANKS

[@pranshusharma2030]

Sir Using transfer learning in various dataset we ar getting same value for all the datsets, how to process our inputs for predicting can you please do it sir, It will be of great help

[@plabmadeeasy]

You've got a subscriber!

[@shivamkashyap6207]

can you make a video on covid 19 detetection using chest xray using transfer learning

[@bihanbanerjee]

thank you

[@Ticitacapepguordiola]

Where can i find the fucking dataset

[@casterwill01]

Check the fucking decsription

[@saisingireddy2359]

💎🛐

[@gunnu61]

x = Flatten()(vgg.output) prediction = Dense(6, activation='softmax')(x) model = Model(inputs=vgg.input, outputs=prediction) model.summary() getting this error: Attempt to convert a value (None) with an unsupported type () to a Tensor

[@leenaselvamani9065]

Can you help me if I try to fit my model it says kernel is dead what shld I do for this prblm