Open navanchauhan opened 4 years ago
Thanks @navanchauhan for the feature request. What specifically in this space, do you think is the most important problem that a new toolkit could solve?
These are some sample use cases
I personally would use it for generating the next movement, given the data extracted from OpenPose/Posenet models
We know that CoreML supports RNNs so is there any way we can add support for sequence generation / prediction models?
Some examples being :-
Compact Prediction Tree Andrew NV’s implementation of his LSTM RNN
This could be used for Generating text, predicting sequences
If you can tell me how I should go about implementing this ( now that Turicreate uses TF it might be easier ) I might be able to work up a PR myself