I am going to be collecting data to train a CNN/deep-learning model for gesture recognition. I have never worked in the domain of gesture recognition before, so I want to know if there are any tips or notes to keep in mind when collecting and processing video data.
For instance, when training a CNN image classifier, it’s good to have the image as a square (but not always needed) and to also normalize the input pixels to a range of, for instance, [0,1].
So are there any best practices or things to keep in mind when collecting & processing data for gesture recognition. Specifically for 3D-CNN/CNN-LSTM model? I plan on building a 3D-CNN/CNN-LSTM model for my gesture recognition, but I don't know for sure yet. I might try other machine learning algorithms as well.