This is the Multi-Modal Digital Assistant of Group 9.
This repository contains all code necessary to run the application, as well as the full source code of the application.
This application uses Java Swing as graphical engine.
- Shireen Aynetchi
- Rafael Diederen
- Aurelien Giuglaris-Michael
- Thierry Guiot
- Lale Kosucu Ozgen
- Gunes Ozmen Bakan
- Thomas Vroom
-
Open this repository. Make sure you have Java, Python, OpenCV and Gradle installed.
-
Launch the application by running the main class located in Main.java.
-
Press the
authentication
button to start the authentication process:
- The assistant will access your webcam to see if there is a human in front of it,
- If the assistant finds a human you will be granted access to the chat.
- If you don't have a webcam press the
No access to webcam
button and enter the passwordXXXXX
.
- Talk to our digital assistant by typing in the chat box, then press the Enter key to send your message.
- Note that after selecting a different language model, you need to press the
clear chat
button to reload the chat window.
-
Gradle instructions:
- make sure you have Gradle installed
- open a terminal and navigate to this repository
- to run the application, type
gradle run
in the terminal (starting up might take a while)
-
Java Runtime Environment must be needed to launch the application.
-
Python and the OpenCV library must be needed to use the webcam functionality:
- https://www.python.org/
- https://pypi.org/project/opencv-python/
- Other packages include
numpy
,mediapipe
andface_detection
-
To use the neural network the following packages are needed:
tensorflow
fastapi
sklearn
numpy
bert_score
transformers
nltk
-
To use the neural network first do the following steps before launching the main class:
- Open a new terminal.
- CD to
src\main\java\org\Project22\LargeLanguageModel
. - Type
uvicorn main:app --reload
. - Wait for the message
Application startup complete.
to appear.
-
Citation of code is mentioned in comments throughout the source code.