Welcome to our repository for IT Project, this project was given by our profs from the Erasmus University College Brussels.
Our team consits of a Product Owner Gust Berchmans and 4 developers: Iben Vanthournout, Jente Tavernier, Senne Clauwaert, Bilal Belkasem.
People who use SMOG or wish to learn it face significant barriers in communication and accessibility due to the lack of engaging tools and real-time solutions. This limits their ability to connect with others in personal, educational, and professional contexts.
Our solution is a sign language translator app designed to support SMOG. It bridges the communication gap between signers and non-signers while offering an engaging way to learn SMOG.
Our app empowers users to:
- Translate gestures into text in real-time.
- Learn SMOG interactively through tailored lessons.
- Track their progress through gamified elements, including a statistics page showcasing milestones and achievements.
To use the project follow these steps:
- Clone the repository
- Open project in Visual Studio Code
- Put the serviceAccountKey.json int the root folder: the serviceAccountKey you can find in our files map in Teams (group-2) or via this lnk: https://ehb.sharepoint.com/sites/IT-Project24-25/_layouts/15/download.aspx?UniqueId=d09ce3a276c2481fbd425f452d4568fc&e=4UdnQL
- Create and activate virtual environment:
python -m venv .venv
.\.venv\Scripts\activate
- Install dependencies:
pip install -r requirements.txt
- Navigate to src directory and run app:
cd src
flet run
Optional: Mobile Phone Setup To use the app on your phone:
- Install the Flet app on your mobile device
- Install IP Webcame on an android phone and set the settings to:
- Main camera: front camera
- Resolution: 640 x 480
- Quality: 30
- Video orientation: Portrait
- FPS Limit: 15
- Start server and copy ip
- Past ip in the python code in the files translate.py and d1l3.py
- Run the app with Android flag:
flet run --android
- Scan the QR code displayed in terminal
- Enter the IP address in the Flet app
Gust:
Github Copilot
https://chatgpt.com/share/672b510d-d684-800b-afa2-a7bb171f265f
https://chatgpt.com/share/6747208d-b308-800b-a58e-20c0dca7ff4c
https://chatgpt.com/share/675976e9-f714-800b-976d-2b5cc7fedcd4
https://chatgpt.com/share/675ab584-57ac-800b-aadf-b1c46a6d6adf
https://chatgpt.com/share/675ab5fd-876c-800b-bfdb-77968f703f75
https://chatgpt.com/share/67629cb0-281c-800b-83e4-717d9e0b3252
https://chatgpt.com/share/6762ac2a-7f64-800b-9cbc-6d596ba54b74
https://chatgpt.com/share/6762ac2a-7f64-800b-9cbc-6d596ba54b74
https://chatgpt.com/share/6763ef37-bc7c-800b-9eb1-22fd6791d36d
https://chatgpt.com/share/6763ef47-6058-800b-aa53-35cb98e85b98
Iben: pip install protobuf==3.20.*
Jente:
Github Copilot
Help from Mr. Bervoets, who sent a video of an alumna who also did a project on Sign Language Translation with TensorFlow.
https://www.tensorflow.org/guide/keras
https://www.tensorflow.org/tutorials/quickstart/beginner
https://machinelearningmastery.com/tensorflow-tutorial-deep-learning-with-tf-keras/
https://keras.io/api/layers/recurrent_layers/lstm/
https://ai.google.dev/edge/mediapipe/solutions/vision/hand_landmarker/python
https://docs.opencv.org/3.4/dd/d43/tutorial_py_video_display.html
https://numpy.org/doc/stable/user/quickstart.html
https://keras.io/guides/sequential_model/
https://keras.io/guides/training_with_built_in_methods/
https://www.youtube.com/watch?v=7kHhz7nkpBw
https://www.quora.com/Why-does-an-LSTM-with-ReLU-activations-diverge
https://www.geeksforgeeks.org/sign-language-recognition-system-using-tensorflow-in-python/
Used ChatGPT to fix Tensorflow model, but because of an image with model statistics I cannot upload the link
Senne:
https://firebase.google.com/
https://www.cursor.com/
Bilal:
https://www.cursor.com
https://flet.dev
https://flet.dev/docs/