Christine Perez edited subsubsection_Data_Modeling_This_phase__.tex  about 8 years ago

Commit id: 8de2dee0a6ca77534d3bb96ba14acc97bbf8fd25

deletions | additions      

       

\subsubsection{Data Modeling}  This phase involves the modeling of the data with the application of machine learning algorithms. A model will be produced, where it will determine which among the activity types in the study were a correctly defined data provided in the data set. \\ \begin{enumerate}  \item Catching trashes using a net  \item Unloading the trashes from the net  \item Proper waste segregation of trashes  \end{enumerate}  The model to be produced should be able to know which activity type does the given feature sets and sequences would lead to. Additionally, the specific kind of hand gesture acquired from the data collection would be used as labels to help the model classify the given activity types. \\    The modeling approach would need a specialized machine learning tool for the kind of technology given in the study. The machine learning tool is used as a classifier for the skeletal tracking and hand events like grip and grip release, hand stabilization, and gesture recognition in Kinect. The skeletal tracking finds the users hand. A window of NxN pixels around each hand is then extracted. Machine learning executes and generates all the hand events. A machine learning task is presented wherein the machine will be trained and tested. The ID3: Iterative Dichotomizer 3 is used by the researchers as the This  machine learning tool to perform and correctly implement is already present in  the gesture recognition. Kinect sensor thus, gestures will be easily recognized.  Taking into consideration the researches presented by the study conducted by \cite{zhang2012microsoft} \cite{gonzalez2014kinect} \cite{yeung2014evaluation} , this process arrived with a standard basis for every hand data as a result of the training and is used in the implementation stage of the study.