Global Average pooling operation for 3D data.
layer_global_average_pooling_3d( object, data_format = NULL, keepdims = FALSE, ... )
What to compose the new
A string, one of
A boolean, whether to keep the spatial dimensions or not. If
standard layer arguments.
data_format='channels_last': 5D tensor with shape:
(batch_size, spatial_dim1, spatial_dim2, spatial_dim3, channels)
data_format='channels_first': 5D tensor with shape:
(batch_size, channels, spatial_dim1, spatial_dim2, spatial_dim3)
2D tensor with shape:
Other pooling layers: