Skip to content

sid-1998/Sign-Language-Recognition

Repository files navigation

Sign-Language-Recognition

The CNN is trained on a custom dataset containing alphabets A-Y (excluding J) of American Sign Language.

Usage

To run the pretrained model

Run:

python Gesture_recognize_sign.py

This will start the webcam.Press C then place your hand inside the green box while performing a gesture and you will get the letter to which the respective gesture corresponds. Press Q to quit.

To train your own model

Set up the path in the Image_capturing.py file

Run:

python Image_capturing.py

Place your hand in the green box and press C to start capturing the data.

Now set up the paths in Image_preprocessing.py file to preprocess the dataset.

Then Run:

python Image_preprocessing.py

After preprocessing set up the path in model.py file to get the preprocessed data for training.

Then Run:

python model.py

This will create model-<val_acc>.h5 files. Choose the appropriate file for Gesture_recognize_sign.py

Then Run:

python Gesture_recognize_sign.py

Demo

Demo

About

Trained a CNN to understand American Sign Language and convert gestures into text

Topics

Resources

License

Stars

Watchers

Forks

Releases

No releases published

Packages

No packages published