Documentation https://torchcvnn.github.io/torchcvnn/
Examples https://www.github.com/torchcvnn/examples
This is a library that uses pytorch as a back-end for complex valued neural networks. It provides :
- complex valued datasets from remote sensing and MRI,
- complex valued transforms,
- complex valued layers, some of them requiring specific implementations because their computation is specific with complex valued activations, others are implemented merely because the lower level implementation raise an exception if processing complex valued activations even though the computations are the same than for real valued activations
- complex valued neural networks
It was initially developed by Victor Dhédin and Jérémie Levi during their third year project at CentraleSupélec.
To install the library, it is simple as :
python -m pip install torchcvnn
or, using uv :
uv pip install torchcvnn
You might also be interested in some other projects:
Tensorflow based :
- cvnn developed by colleagues from CentraleSupélec
Pytorch based :