This project was to make an interactive way to practice your alphabet in sign language. Some images of the GUI and the results are shown below.
First thing you need to do is change absolute paths in the botton_gui_V3.py script to the correct ones for your computer. Next please see the folder "Models" in the "yolo-hand-detection" folder and download the "download_models.sh" file. After that and after you have installed all of the dependencies you can use the GUI by typing the following code into your terminal.
python3 botton_gui_V3.pyThe YOLO object detection was pre-trained and can be found at "https://github.com/cansik/yolo-hand-detection". Currently that is not working so a new hand detection model will need to be found and implemented into the code.
A CNN was trained on a dataset obtained from Kaggle to predict what letter a hand was signing. The YOLO hand detection was used to draw a box around the users hand so it could be cropped for the image analysis using the Tensorflow CNN model trained on the large Kaggle dataset.