1

I am beginner of medical deep learning. I am practicing medical deep learning. I don't know how to control stride in ConVNN. I want to know if I control stride in ConVNN,that I think image is different. so I want to see that.

already I saw many information about stride in ConVNN in Keras. but I didn't understand.

def build_SRCNN():
    input_img = Input(shape=(512, 512, 1), name='image_input')


    x = Conv2D(64, (9, 9), strides=(1,1), activation='relu',kernel_initializer='he_normal', padding='same')(input_img)
    x = Conv2D(32, (3, 3), strides=(1,1), activation='relu', kernel_initializer='he_normal', padding='same')(x)
    x = Conv2D(1, (5, 5), strides=(1,1), activation='relu',kernel_initializer='he_normal', padding='same')(x)

    # model
    SRCNN = Model(inputs=input_img, outputs=x)

    return SRCNN

I wrote stride = (1,1) to (3,3) or (5,5) or (7,7) ... but error is same.

hpwww
  • 535
  • 3
  • 8
조정효
  • 11
  • 1
  • 1
    Welcome to StackOverflow. What is the error you're seeing? – OverLordGoldDragon Oct 01 '19 at 16:45
  • 1
    Please post the error that you are getting. As I checked, there is no error in `model` compilation. Also, please choose `strides` carefully as it can severely affect the performance of the model in terms of overall model accuracy. – Rishab P Oct 01 '19 at 18:45

0 Answers0