Select the correct statement.

Select the correct statement.

Convolution layers learn global patterns and dense layers learn local patterns.

The kernel can extract various features, resulting in a wide range of resolution effects.

The pooling layer in CNN is used to increase the spatial size of the convolved feature.

$3$D feature extraction in CNN cannot combine multiple filters.

Semantic segmentation can only associate one pixel of an image with a class label.