letip / tracking-randomforest Goto Github PK

Please create a class TransitionClassifier that:

• holds a list of positive and negative examples and their features
• has a method addSample(featuresA, featuresB, label), where A and B denote the source and target object
• has a method train()
• has a method predictSample(featuresA, featuresB)
• bonus task: add methods addSamples and predictSamples for batches of features

Then, create a trainFromGT script (pretty much the same as RF-function, but using the new TransitionClassifier), that loads a dataset, sets up the TransitionClassifier, trains it, and saves the random forest to disk.

We want to know how much better the transition classifier does with respect to the pure-distance based computation of a "transition probability".

For this, predict the probabilities (add a predictProbability method to the TransitionClassifier) for all validation samples using the trained random forest. Now threshold the probabilities p at some threshold t in [0,1]: e.g. when t=0.3, then every sample with a probability p>t for being a good transition will be classified as positive, otherwise negative. (The Random Forest's predictLabel method does this with t=0.5.)

Do the same with transition probabilities derived from distances as follows:

import numpy as np

def distanceToProbability(distance, transitionParameter=5.0):
    np.exp(-1.0 * dist / transitionParameter)

Then compute precision, recall, and f-measure for each threshold and plot a graph that looks roughly like the following (curves are made up!):

... and save it using its writeHDF5 method.