ofxMSATensorFlow (2016)

A C++ openFrameworks wrapper for Google’s graph based machine intelligence / deep learning library TensorFlow. Includes many examples for training (in python) and then loading pretrained models in C++ for running inference (or generation) on images (e.g. pix2pix, style transfer, classification) and sequences with RNNs/LSTMs (e.g. text, handwriting).

https://github.com/memo/ofxMSATensorFlow

Images

Examples

Realtime pix2pix

Realtime pix2pix on webcam

Handwriting generation with LSTM／RNNs

Text generation with LSTM／RNN

Image classification

Credits & Acknowledgements

Created during my PhD at Goldsmiths, funded by the EPSRC UK.

Words(3 / 3)

The artist’s own writing about this work.

2021
Deep Visual Instruments: Realtime Continuous, Meaningful Human Control over Deep Neural Networks for Creative Expression. Memo Akten. Goldsmiths, University of London, 2021. · PhD thesis
2016
Collaborative creativity with Monte Carlo Tree Search and Convolutional Neural Networks · Independent
2016
Collaborative Creativity with Monte Carlo Tree Search and Convolutional Neural Networks. Memo Akten, Mick Grierson. Constructive Machine Learning Workshop at Neural Information Processing Systems (NeurIPS 2016), 2016. · Academic