/patchnorm

Implementation of the PatchNormConv2D and EfficientPatchNormConv2D layers.

Primary LanguagePythonMIT LicenseMIT

Patch-normalized Convolution

patchnorm

Implementation of patch-normalized convolutional layers in PyTorch and Tensorflow Keras.

Installation

Over ssh:

python3 -m pip install -U git+ssh://git@github.com/benjamindkilleen/patchnorm.git --user

Or over https:

python3 -m pip install -U git+https://github.com/benjamindkilleen/patchnorm.git --user

Usage

This package contains both Tensorflow Keras and PyTorch implementations of the PatchNorm layers. They can be imported independently of one another, for example:

from patchnorm.pytorch import PatchNormConv2D
from patchnorm.pytorch import PatchNormConv3D

or

from patchnorm.tfkeras import PatchNormConv2D
from patchnorm.tfkeras import PatchNormConv3D