The code in these folders is research quality. No guarantees or support is provided. 

The algorithms folder is a placeholder for the different algorithms you may want to use. For the paper we used the following code all available online:
Black & Anandan
FlowLib
Horn & Schunck
TV_L1

For the training we used OpenCV 2.0 Random Forests. Additionally you will need code from Middlebury for reading and writing the .flo files. 

1. calcflows.m computes the flow for each image
2. generateFeatureVector.m computes the feature vector
3. writeFVToTextBinary.m writes the feature vector to a text file so you can use it with the random forest classifier. 
4. readForestPredictionsBinary.m readds the results of the classifier. 

The data is available from our website:
http://visual.cs.ucl.ac.uk/pubs/algorithmSuitability/

If you use this data please cite:
@INPROCEEDINGS{MacAodhaEtAlCVPR2010,
  author       = {Mac Aodha, Oisin and Brostow, Gabriel J. and Pollefeys, Marc},
  title        = {Segmenting Video Into Classes of Algorithm-Suitability},
  booktitle    = {CVPR},
  year         = {2010},
}

For feedback please contact:
Oisin Mac Aodha 
o.macaodha@cs.ucl.ac.uk