The goal is to train the machine to train and predict which type of emoji hand you are showing to the camera. You can try it here https://bufolab.github.io/Handemoji/
- Select an emoji from top buttons
- Place your hand in front of the camera and press Capture
- Select another emoji and repeat the process
- Once you have taken snapshots for every emoji. Press Start Learning
- Press Predict to start
This version has been made just to play with tensors and some functionalities of the model. I have imported the ModelNet already trained to ease the traning. Also this version of the code does not go into any improvement. So that is why it there might be some memory leak or the capture images has been done in a naive way. There is huge room for improvements!