ASLAssist aims to aid the deaf-mute towards a completely wireless platform by converting American Sign Language to a text form. It can be used as a web chat for the deaf-mute.
First, train the model using Train_CNN. After this step you will obtain a file with the trained model, called "my_model.h5". You may use the Predict_one_image file to test the accuracy of the model.
Next, create a folder called 'pictures' in the place where you have downloaded all the code. We first run webcam.py to open the webcam and take the pictures of your ASL signs. Click c to keep taking more pictures, and q twice to stop the webcam and to close all windows.
Next, put the file "image.html" into a new folder called templates. Then, run "model.py". model.py integrates the model with some front end features. model.py causes a localhost server to be set up. Open the localhost server in order to see the interface that is encoded by image.html.
On windows open command prompt centre and run the codes beginning with pip.requires python 3.7 Tensorflow pip install tensorflow keras pip install keras PIL pip install Pillow==2.2.1 flask pip install flask Socketio pip install flask-socketio pandas pip install pandas numpy pip install numpy