Validation accuracy not increasing

#1

I was trying the video lecture on Tranfer learning using resnet50 model, Here sir is getting good validation as well as train accuracy.while practicing i used train and validation generator (sir have used normal split),here my train accuracy is good but validation accuracy is only 25%. I don’t know what is the problem. please help

collab notebok link:

https://colab.research.google.com/drive/1xj6IKk8xIPu2TWSHY_7390M0Las9T_RB

thanks

#2

Thanks,

I have solved this problem. But for all those who are facing similar issue, the SOLUTION is:

When using Image data generator in transfer learning problems we dont have to rescale the image but instead we have to use preprocessing _function as preprocess_input in the resnet50 library

train_gen=ImageDataGenerator(
preprocessing_function=preprocess_input,
width_shift_range=0.2,
height_shift_range=0.2,
horizontal_flip=True,
zoom_range=0.3,
rotation_range=40
)

val_gen=ImageDataGenerator(
preprocessing_function=preprocess_input
)