This is a model to classify Vietnamese sign language using Motion history image (MHI) algorithm and CNN. We have 6 classes for most basic vocabolary disabled people need in daily life :
- Uống (Drink)
- Đói (Hungry)
- Vui (Happy)
- Giận (Angry)
- Tôi (Me)
- and class None for indicating no actions.
First proposed by A.Bobick et.al as a method to capture both spatial and temporal information of an action. Here some demonstration.
This algorithm conveniently enables CNN to recognize actions, not just still images. Since sign languages are consisted of dynamic gesutres, it's perfect for this project.
Use MHI.py to capture images for dataset
Python3
Numpy
Tensorflow
Keras
opencv
Matplotlib
Seaborn