The Vision Toolkit is a genetic-algorithm-based circle-finding tool. I wrote it to learn more about how image data could be processed during the 2010-2011 FIRST robotics season. The fitness function can be changed to support other simple shapes and, if used continuously, the chromosomes could be held over between frames of video. Vision Toolkit uses QwtPlot3D to plot the fitness landscapes. Edge detection is done by looking for discontinuities in the second partial derivatives of the luminance at each point.