mindearth.cell.DEMNet

class mindearth.cell.DEMNet(in_channels=1, out_channels=256, kernel_size=3, scale=5, num_blocks=42)

Digital Elevation Model is based on deep residual network and transfer learning. The details can be found in Super-resolution reconstruction of a 3 arc-second global DEM dataset.

Parameters

• in_channels (int) – The channels of input image.

• out_channels (int) – The number of output channels.

• kernel_size (int) – Kernel size.

• scale (int) – The scale factor of new size of the tensor.

• num_blocks (int) – The number of blocks in the DEMNet.

Inputs:

• x (Tensor) - Tensor of shape \((batch\_size, out\_channels, height\_size, width\_size)\).

Outputs:

Tensor, the output of the DEMNet.

• output (Tensor) - Tensor of shape \((batch\_size, out\_channels, new\_height\_size, new_width\_size)\).

Supported Platforms:

Ascend GPU

Examples

>>> import numpy as np
>>> import mindspore as ms
>>> from mindspore import ops, Tensor
>>> from mindspore.nn import Cell
>>> from mindearth.cell import DEMNet
>>> input_images = np.random.rand(64, 1, 32, 32).astype(np.float32)
>>> net = DEMNet(in_channels=1, out_channels=256, kernel_size=3, scale=5, num_blocks=42)
>>> out = net(Tensor(input_images, ms.float32))
>>> print(out.shape)
(64, 1, 160, 160)