Input_shape in all the layers

Hi,

When the CNN was first taught, Prateek Bhaiya said that we don’t have to mention the ‘input_shape’ in every Convolution layer since it is only important in the first where the program doesn’t know about it. Eventually, in the following layers, the model realizes the ‘input_shape’ on its own. But, in these videos, we have defined the input_shape for all the layers and is it of the shape of the input image and not what it will receive from the previous layer. Why are we mentioning that in the following layers?

Hey @A18ML0031, it is not mandatory to mention shapes in each layer. But bhaiya has done so just to make you understand the shapes that will pass from one layer to another. You just have to mandatorily provide the input shape of the first layer.

I hope this helps ! :slightly_smiling_face:
Happy Learning ! :slight_smile: