Ongoing Deep Learning Projects

We have a few deeplearning projects encapsulated in jupyter ipython notebooks in the the notebooks directory of this github repository The code written is mainly by us, But we have shown credit where ever we have used code from other repositories. We have used tensorflow and jupyter ipython notebooks with ipywidgets installed

github reference

github repository

requirements

We require jupyter ipywidgets to be installed before running these notebooks. ipywidget is a small add-on to jupyter technology. You can easily pip install this component once you have installed jupyter.

Before starting the server with ‘jupyter notebook --ip=0.0.0.0’ we do the following command to ensure that ipywidgets are enabled in the notebook. ‘jupyter nbextension enable --py widgetsnbextension.’

notebooks

• notebooks/tf_mnist.ipynb is a simple mnist project

• notebooks/tf_autoencoder.ipynb is a simple autoencoder written for mnist data

• notebooks/tf_cifar.ipynb is a simple training/validation framework for cifar-10 data

• notebooks/tf_cifar_optimized.ipynb attempts to get a higher validation accuracy than notebooks/tf_cifar.ipynb by means of a slew of optimizations.

• notebooks/tf_vgg16.ipynb loads a pre-trained vgg-16 weights, build a model and test random images taken by us

utility code

• notebooks/common/utils.ipynb is a utility notebook which can be imported by other notebooks. This utility notebook, provides the following utilities

  • ProgressImageWidget, is a custom ipywidget written for interactively displaying training data
  • Plotter, is a class for adding channels and sample data to channels , which can be plotted as a png image
  • ImgGrid, is a class for plotting a grid of images
  • ImgGridController, is a higher level class for managing ImgGrids and corresponding ProgressImageWidget-s

credits

• notebooks/common/load_notebooks.py is reproduced from its implementation here
• notebooks/common/imagenet_classes.py is taken from Davi Frossard’s site
• notebooks/tf_vgg16.ipynb, much of the code for building the tensorflow model from vgg16.npy is adapted from MarvinTeichmann’s tensorflow implementation of fc net
• notebooks/tf_cifar_optimized.ipynb, we have used the code from Jean Dut for GCN pre-processing of data
• notebooks/tf_cifar.ipynb, owes its origins to the official tensorflow cifar tutorial

---

---

blog comments powered by Disqus

Published