Computer Vision Sign Language Detection and Classification Model with TensorFlow, Keras, OpenCV and Mediapipe
This Computer Vision Model, detects the letters of the Sign Language. It can recognizes the letters "A", "B", "L", "H" and "O".
The Model was trained with more than 1500 Images, 300 Images from each Sign Language Letter, and using Transfer Learning, from the CNN VGG-16 TensorFlow Model.
Mediapipe detects any Hand from the Image, then is cropped, and last the Model classifies which letter is of the Sign Language.
The code to collect any Sign Language Letter Image of the Hand is the data_collection.ipynb file and the code of the Model is in the sign_language_model.ipynb file. To Test the Model in Real-Time, run the code in the sign_language_preview.ipynb file.
Remember first to train and save the model, by running the sign_language_model.ipynb file.
jupyter notebook sign_language_preview.ipynb