This python package provides an implementation of the Virtual Retina developed by Adrien Wohrer. It uses theano to simulate spike trains of retinal ganglion cells by directing the input through a number of computation nodes. Each node might do linear or nonlinear computations, eg. convolve the inpute with a spatio-temporal kernel or apply gain control.

• retina.py provides a VirtualRetina class and corresponding computation nodes
  • the VirtualRetina class computes the entire cascade of nodes
  • each node can retain its state, such that input can be chunked
• retina_base.py provides general functions for creating kernels
• retina_virtualretina.py provides default configuration options (as a dictionary) and methods for writing xml configuration files for the original program
• vison.py aims at reimplementing the nodes in a more general framework, such that they can be used for any vision computation.