API Reference¶
This page lists the available functions and classes of pysegcnn
.
Dataset¶
Custom dataset classes compliant to the PyTorch standard.
Generic classes¶
Generic class to implement a custom dataset.
Base class for multispectral image data. |
Generic class to implement a custom dataset following a standard directory structure.
Base class for standard Earth Observation style datasets. |
Specific classes¶
Specific classes for some open-source image datasets. Currently, the following spaceborne multispectral image datasets are supported out-of-the-box:
Class for the Sparcs dataset by Hughes & Hayes (2014). |
|
Class for the Cloud-95 dataset by Mohajerani & Saeedi (2020). |
Models¶
Layers¶
Convolutional neural network layers.
Basic convolutional block. |
|
A convolution preserving the shape of its input. |
|
Block of convolution, batchnorm, relu and 2x2 max pool. |
|
Block of convolution, batchnorm, relu and 2x2 max unpool. |
|
Block of convolution, batchnorm, relu and nearest neighbor upsample. |
Encoder-Decoder architechture¶
Generic Encoder
and Decoder
classes to build an encoder-decoder
architecture.
Block of a convolutional encoder. |
|
Block of a convolutional decoder. |
|
Generic convolutional encoder. |
|
Generic convolutional decoder. |