We had about twenty people at the last Melbourne Python User Group meeting for 2012, celebrating with six litres of mulled wine, five dozen fruit mince pies, four speakers speaking and a partridge in a pear tree.
Nathan Faggian and Ed Schofield gave a great overview of image processing in Python, starting with some of the packages available, including scikit-image, which was recommended as very Pythonic. The scikit-image gallery is worth a look.
From there, they covered image segmentation, cellular automata and implementing versions of Conway’s Game of Life with a straight Python version and a version that used numpy for a significant performance improvement. This led to growcut.automata, Nathan’s implementation of the Growcut algorithm for user-interactive image segmentation using cellular automata. The user nominates a few regions within an image as being associated with either the object of interest or the background; Growcut can then segment out the object of interest. The live demonstrations were impressive: segmenting a flower out of a reasonably complex background and then a scanned piece of paper from a photo.
I’d point to Nathan’s code but the repository seems to have 404’ed 🙁
A recurring theme from the December meeting was the use of IPython notebooks for prototyping and demonstrations. If you aren’t familiar with notebooks, some good starting points are:
I’ll write about the other talks from the December meeting in part 2…