Israeli Sign Language Letters Recognition using a Convolutional Neural Network (CNN) implemented in Keras + TensorFlow + OpenCV. This project is part of the requirements for the Final Project in B.Sc. Software Engineering.
Delete & Space signs were added for simplicity.
Python 3.6.8
tensorflow 1.13.1 (or tensorflow-gpu 1.13.1)
Keras==2.2.4
opencv-python==4.1.0.25
seaborn
sklearn
imutils
Pillow
numpy
matplotlib
After downloading the repo, you will need to setup database and the model:
-
Create exe file by running from cmd:
pyinstaller --noconsole --onefile gui.py
This will create gui.exe inside dist folder. -
Run SetupScript.iss script file which will pack resources and models under single exe setup in Releases folder. (InnoSetup must be installed).
Installation setup could be downloaded from the following OneDrive Link
The Image preprocessing, OpenCV and CNN integration are based on Dr. Jared Vasquez's CNN-HowManyFingers project. The CNN architecture used in this project is based on Hemalatha Vakade's CNN-for-Image-Classification project.