- Python 3, OpenCV, Tensorflow
- To install the required packages, run
pip install -r requirements.txt
.
The repository is currently compatible with tensorflow-2.0
and makes use of the Keras API using the tensorflow.keras
library.
- First, clone the repository and enter the folder
git clone https://github.com/Janmansh/Hackverse-2.0.git
cd Hackverse-2.0
python gesture.py --mode display
-
The folder structure is of the form:
src:gesture.py
(file)aGest.xml
(haar cascade file)model.h5
(file)
-
This implementation by default detects gestures on all hands in the webcam feed. With a simple 4-layer CNN, the test accuracy reached 58.2% in 50 epochs. Press 'q' to close the camera when finished
-
First, the haar cascade method is used to detect hands in each frame of the webcam feed.
-
The region of image containing the face is resized to 90x70 and is passed as input to the CNN.
-
The network outputs a list of softmax scores for the ten classes of emotions.
-
The emotion with maximum score is displayed on the screen.
- Made using NodeJs and WebRTC.
- npm
- Express
- Socket.io
- peerjs
- uuid
- Run
peerjs --port 3001
on terminal for establising peer to peer connection. - Run
npm run devStart
for starting server
- The video app needs to have the ML pipeline so that we can see the emoji's on live interaction.
- Gaurav Singh
- Aniket Agrawal
- Janmansh Agarwal