There are a few sites that are definetely worth checking if you are looking for (research) code related to computer vision, music analysis or machine learning:
Gabriel Peyré: CNRS researcher working on image processing, computer graphics and applied math, in general. You can find many tutorials and links to important research papers on sparsity and wavelets, among other things.
SPArse Modeling Software: package developed in France and US. It implements algorithms on Sparse coding and Dictionary learning.
Torch: Newly developed language optimized to deal with large amount of data. There are libraries with already implemented neural networks, autoencoders, and other methods related to deep learning.
data group: Group lead by Stephane Mallat, which is lately working on the applications of the Scattering transform to music analysis, computer vision, time series or language processing.
ipol: journal with peer-reviewed code and online demos.
This will be an on-going post, as we discover new interesting links.