WaveletScattering.jl

This package intends to provide fast scattering transforms of images, sounds and videos. Scattering coefficients are powerful features for signal-based machine learning tasks, as they offer a great discriminability while discarding non-informative variability, such as rigid motions, frequency transpositions, and elastic deformations.

Its design is inspired from the scattering.m toolbox, while taking advantage of Julia's multiple dispatch and in-place operations.

The development of this package has been stopped and superseded by Kymatio, a PyTorch/Keras/NumPy implementation: https://github.com/kymatio/kymatio