Deep learning
Receptive field
A receptive field is the input region that can influence one feature value.
How it works
Deeper convolutional layers combine information across progressively larger regions. Kernel size, stride, dilation, and depth determine its extent.
A concrete example
Illustrative: two stride-1 3×3 convolutions give a 5×5 receptive field.
Do not confuse it with Kernel
a kernel is the learned filter; the receptive field is the covered input region.