No description, website, or topics provided.
Switch branches/tags
Nothing to show
Clone or download
Latest commit d137b2b Oct 19, 2017
Failed to load latest commit information.
CIFAR CIFAR Mar 29, 2017
STL Update Apr 3, 2017
imagenet Update Oct 19, 2017 point to arxiv abstract page Mar 29, 2017 initial Mar 28, 2017

Scaling The Scattering Transform : Deep Hybrid Networks

This repository contains the experiments found in the paper:


In order to run our experiments you will need at minimum the following python packages: pytorch,opencv,pyscatwave package. The simplest way to install pytorch and opencv is through anaconda. We recommend python 2.7 + anaconda. The pyscatwave package can be found here


We provide a pre-trained model similar to the one described in the paper.

To run the trained model of scattering+resnet on imagenet ILSVRC validation set:

  1. Make sure you have downloaded at least the validation set of ILSVRC2012 and have it organized by class categories Note: due to problems with pytorch dataset constructors make sure your imagenet directory has no hidden files, or extra directories besides the 1000 ILSVRC categories.. otherwise all the images will be mislabeled
  2. Download the model file from
  3. Add this to the imagenet/ directory
  4. Run the script to evaluate on the ILSVRC validation set specifying --imagenetpath to point to your imagenet directory

Training scripts for imagenet and SLE feature extractor will be added soon


Simply run python script in the STL directory


To run the small sample experiments Example:

python --model resnet12_8_scat --save "test"  --seed 1 --sampleSize 500 --mul 20