carpedm20/pixel-rnn-tensorflow
A TensorFlow implementation of PixelCNN and PixelRNN deep generative models for image synthesis.
Not currently ranked — collecting fresh signals.
star history
This repository implements Pixel Recurrent Neural Networks, an autoregressive generative model that synthesizes images by predicting each pixel conditionally on previously generated pixels. It provides TensorFlow implementations of both PixelCNN using masked convolutions and PixelRNN using row LSTM and diagonal BiLSTM architectures. The project includes training scripts for MNIST and CIFAR-10 datasets, with support for generating new images from trained models.
Frequently asked
- What is carpedm20/pixel-rnn-tensorflow?
- A TensorFlow implementation of PixelCNN and PixelRNN deep generative models for image synthesis.
- Is pixel-rnn-tensorflow open source?
- Yes — carpedm20/pixel-rnn-tensorflow is open source, released under the MIT license.
- What language is pixel-rnn-tensorflow written in?
- carpedm20/pixel-rnn-tensorflow is primarily written in Python.
- How popular is pixel-rnn-tensorflow?
- carpedm20/pixel-rnn-tensorflow has 488 stars on GitHub.
- Where can I find pixel-rnn-tensorflow?
- carpedm20/pixel-rnn-tensorflow is on GitHub at https://github.com/carpedm20/pixel-rnn-tensorflow.