ASL interpretation

Based on features get from LeapMotion sendor, using SDK toolit, this project uses Sckit-learn library to interpret American Signal Languages symbols.

Three models were been analyzed:

• ANN (Artificial neural netwod)
• KNN
• SVM
• Nearest Centroid

Features and training

The set of features used to train the set of models were (get from LeapMotion sensor SDK):

• RTP
• RTT
• RTJ

To get trained model:

• Run script:

python train.py --classifierType ANN

to get Confusion Matrix from SVM, for example.

The list of available models can be visualized with:

python train.py --help

Running trained model

All the trained models are in the directory data/. To run the ANN trained model, type:

python app.py

• Read the document: Letter_Leap.pdf to get more information

In case of you want the real time application, please send me a message!