Keras ImageDataGenerator | Keras Data Augmentation | Create image dataset for CNN Deep Learning

preview_player
Показать описание
We need to have huge image dataset for convolutional neural network, this video will explain you, how you can generate huge image from few images.

Keras ImageDataGenerator concept is also very important for Keras image classifier code.

My Other YouTube Videos:
Рекомендации по теме
Комментарии
Автор

Aapka thet desi accent sun k bhai jo accha laga hai....
kano ko sukoon mil gya.

IItianrider
Автор

Thanks for your nice video.
How about the last instructionwhich does not appear inthe video:
for batch in datagen.flow(pic_array, batch_size=5, sae_to_dir="", save_prefix='dog',

BoularbahSouissi
Автор

hello i want to ask you now the imagedatagenerator is deprecated now is the same with the directoryiterator?

variiit
Автор

what is the use of the nearest augmentation parameter, sir?

mihretdesta
Автор

Can't we pass other features along with directory and classes in the generator to train our model?

rsnadata
Автор

I did not understand why we added one more dimension by reshape method ? Can someone explain it?

alperenbaskan
Автор

sir the Imagedatagenerator from keras, does it apply all the specified augmentation operations for every image passed to it?

shrinivas
Автор

Thanks for explaining this, I'm not a believer on blind implementation and phrases from some wiki. My question is, (1) suppose if I've got 4000 images of Dogs, then the image_datagen object will use flow_from_directory method, apply the attributes, and will create how many training images for a single image/batch_size say 32? like you showed here we can create as many image using the for loop..(2) I believe we're creating these images in order to train our model to identify for a test image that might be in different orientations right?

dipankarrahuldey
Автор

can you explain what it does ImageDataGenerator

АсхатБекенов-ъу
Автор

Awesome ! is ImageDataGenerator only used for deriving many image of different position from one image? or it is used for other works also?

kaushalkumar
Автор

when to use class_mode= 'multi_output' in generator from dataframe

darkemotions
Автор

Great explanation ! Many thanks, i was wondering does the order of the parameters used in ImageDatagenerator matter ? for example should Fillmode always be placed after settings that distort the aspect ratio and will it rotate first then do width shift etc.

LAKXx
Автор

Good explanation . Thank you. Can you explain how to perform data augmentation for a number of images at a time. I mean Images in a folder.

marydayanaa
Автор

sir, how i can use this code for whole image dateset instead of single image

Mrnorthpolestar
Автор

Great video thanks! But your code doesn't seem to normalise pixels i..e 'rescale' does not work and I haven't found a way to make this work... any ideas??? thanks

felixbeacher
Автор

Could not import PIL.Image. The use of `load_img` requires PIL.
i am getting this error
i have pillow installed

renuverma
Автор

How to split image data set into xtrain n y train

PadminiMansingh
Автор

Sir if I have a training data set and I have to augment that... How to create a loop so that it take images one by one and augment them and save in some folder??

Prachi_Mayank
Автор

YOU'RE FUN (and funny!) TO WATCH !!
OF COURSE I SUBSCRIBED - looking forward to LOTS MORE of your videos Binod !!!

-Mark Vogt, Solution Architect, Data & AI, AVANADE (USA)

VOGTLANDOUTDOORS
Автор

Very nice explanation0!!
can you tell something about flow_from_dataframe

avinash-xhqw