PatRecog - Pattern Recognition Framework

PatRecog (Pattern Recognition) is a framework for both static (i.e., "traditional" static features) and dynamic (i.e., time-sequence) classification. It was developed to bypass the challenge felt in adequately preparing both training and testing data for different classification methods.

The framework allows to load data from a given dataset, be it static (i.e., "traditional" features) or dynamic (i.e., time sequence data). Datasets need to be prepared in EXCEL, bearing in mind that:
1. Data from different classes need to be on different files
2. Data from different trials of a given class need to be on different tabs from the same file

An example of a dataset is provided with the framework, following the aforementioned structure. The dataset includes 5 golfers (S01 to S05) executing the putting at 1 meter (D1) and 4 meters (D4) away from the hole. Data was obtained using Ingeniarius' InPutter.

The following classifiers have been implemented (so far).
Static classification (i.e., for "traditional" static features):
* SVM - Based on the multiclass implementation of the Least Squares Support Vector Machine (LS-SVM) by J.A.K. Suykens, T. Van Gestel, J. De Brabanter, B. De Moor, J. Vandewalle, Least Squares Support Vector Machines, World Scientific, Singapore, 2002 (ISBN 981-238-151-1).
* ANN - Runs MatLab's artificial neural network.

Dynamic (time-sequence) classification (i.e., for dynamic features):
* LSTM - Runs MatLab's deep learning Long Short-Term Memory (LSTM) networks.

A more thorough explanation on how to use the framework will follow in a near future.

This work was supported by the Portuguese Foundation for Science and Technology (FCT) under the grant SFRH/BPD/99655/2014, Ingeniarius, Ltd., CIPER, Faculdade de Motricidade Humana, Universidade de Lisboa, Laboratory of Expertise in Sport (SpertLab), and Centre for Sports Engineering Research (CSER).