I’m working on a gesture recognizer and I only want it to recognize 1 gesture. Should I train the model on a set of random gestures so it can learn the one I want it to recognize? Or, should I just create random noise? I’m using Accelerometer and Gyro.
Having a class with a set of random gestures will make the gesture detection more robust. You can also create a 3rd class with random noise.
Once you have collected enough data, you can tune the DSP and Neural Networks parameters to optimize the accuracy on your targeted gesture.