DenseBlock

tiatoolbox . models . architecture . hovernet . DenseBlock

class DenseBlock(in_ch, unit_ksizes, unit_chs, unit_count, split=1)

Dense Convolutional Block.

This convolutional block supports only valid padding.

References

Huang, Gao, et al. “Densely connected convolutional networks.” Proceedings of the IEEE conference on computer vision and pattern recognition. 2017.

Initialize DenseBlock.

Methods

forward

Logic for using layers defined in init.

Attributes

training

Parameters:

• in_ch (int)

• unit_ksizes (list[int])

• unit_chs (list[int])

• unit_count (int)

• split (int)

forward(prev_feat)

Logic for using layers defined in init.

Parameters:

• self (DenseBlock)

• prev_feat (Tensor)

Return type:

Sequential