Python based alternative to Weka?

So I'm experimenting with image segmentation, and I've been using Weka for machine learning.

Right now I have an ungodly mess where I'm calculating features in MATLAB, training in Weka Explorer, classifying with command line Weka, and using ipython notebooks reading and writing .tif, .mat and .arff files to glue this all together.

I'm starting to get some decent results, so I'd like to clean this up a bit. I think this will require moving everything either to Python or Java. I much prefer Python, so I'd like to use it if possible. I've also read in multiple places that Weka sucks and there are much better alternatives.

However, I've not found anything yet as simple to use or full-featured. Is there a python library with the same kind of plug-and-play GUI for quick prototyping?

Scikit-learn seems to be one of the better libraries, but it seems to have only a fraction of the algorithms available.

In general it seems like there are a lot of different ML packages for python, but no analogous unifying framework. Or is there something I'm missing?