Human Gesture Recognition

Human Gesture Prediction using PoseNet data.

This is a server application that serves with the name of a gesture when provided the json data generated by posenet using the videos of the gestures.

This was initially started as a class project for CSE 535, at Arizona State University under Dr. Ayan Banerjee.

Details

The server mainly provides with API that receives 'application/json' data and responds with json data as well. It uses some pretrained models to predict the human gestures based on the input. The input format is same as the json files generate by the 'scale_to_videos.js' file in this repository: https://github.com/prashanthnetizen/posenet_nodejs_setup

How to start the server

1. Install docker for your OS:
   https://www.docker.com/
2. For Linux users, add your user to the docker group:
   https://docs.docker.com/install/linux/linux-postinstall/
3. Clone this repository in your machine:
   

   git clone https://github.com/tnasim/cse535-2019F-A2.git

4. Go to the downoad directory:
   

   cd cse535-2019F-A2

5. Build the docker image (notice the dot, '.', at the end):
   

   docker build -t myserver .

6. Run the docker image just created:
   

   docker run -ti --rm -p 5000:5000 myserver python3 app.py

7. For Windows, use the below line to run the docker image just created:
   

   winpty docker run -ti --rm -p 5000:5000 myserver python3 app.py

8. Check if the server is running on browser:
   http://{your-server-url}:5000/
   example: http://127.0.0.1:5000/