Various utilities for automatically counting baobabs from satellite imagery ('bobibabbing') - VERY experimental
Based around a multilayer perceptron written by Issam H. Laradji, both mlp-example and classify-cascade do pretty much the same thing - take in training images, train the neural net and then identify any points in a test image that match the classifier.
This code is not finished, but if you really want to run it, download the latest scikit-learn, place all the .py files in the folder and run. But its probably better if you just write to me or wait for me to put it all together properly...